1 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4 a bool success value. Don't call output_operand_lossage here.
5 (aarch64_print_ldpstp_address): Return a bool success value.
6 (aarch64_print_operand_address): Call output_addr_const if
7 aarch64_print_address_internal fails.
8 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
9 'y'; call output_operand_lossage instead. Call output_operand_lossage
10 if aarch64_print_ldpstp_address fails.
12 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
14 * tree-vector-builder.h
15 (tree_vector_builder::binary_encoded_nelts): Declare.
16 * tree-vector-builder.c
17 (tree_vector_builder::binary_encoded_nelts): New function.
18 * fold-const.c (negate_expr_p): Likewise.
19 (operand_equal_p, fold_checksum_tree): Likewise.
20 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
21 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
22 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
23 (uniform_vector_p): Likewise.
24 * varasm.c (const_hash_1, compare_constant): Likewise.
25 * tree-ssa-ccp.c: Include tree-vector-builder.h.
26 (valid_lattice_transition): Operate directly on the VECTOR_CST
28 * ipa-icf.c: Include tree-vector-builder.h.
29 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
30 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
32 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
34 * tree.c (build_vector): Delete.
35 * tree.h (build_vector): Make static and move into the self-testing
38 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
40 * vector-builder.h (vector_builder::derived): New const overload.
41 (vector_builder::elt): New function.
42 * tree-vector-builder.h (tree_vector_builder::type): New function.
43 (tree_vector_builder::apply_step): Declare.
44 * tree-vector-builder.c (tree_vector_builder::apply_step): New
46 * gimple-fold.h (tree_vector_builder): Declare.
47 (gimple_build_vector): Take a tree_vector_builder instead of a
48 type and vector of elements.
49 * gimple-fold.c (gimple_build_vector): Likewise.
50 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
52 (get_initial_defs_for_reduction): Likewise.
53 (vectorizable_induction): Likewise.
55 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
57 * tree-vector-builder.h
58 (tree_vector_builder::new_binary_operation): Declare.
59 * tree-vector-builder.c
60 (tree_vector_builder::new_binary_operation): New function.
61 * fold-const.c (fold_relational_const): Use it.
62 (const_binop): Likewise. Check that both input vectors have
63 the same number of elements, thus excluding things like WIDEN_SUM.
64 Check whether it is possible to operate directly on the encodings
67 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
69 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
70 new_unary_operation, operating only on the encoded elements.
71 (const_unop): Likewise.
72 (exact_inverse): Likewise.
73 (distributes_over_addition_p): New function.
74 (const_binop): Use tree_vector_builder and new_unary_operation
75 for combinations of VECTOR_CST and INTEGER_CST. Operate only
76 on the encoded elements unless the encoding is strided and the
77 operation does not distribute over addition.
78 (fold_convert_const): Use tree_vector_builder and
79 new_unary_operation. Operate only on the encoded elements
80 for truncating integer conversions, or for non-stepped encodings.
82 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
84 * config/sparc/sparc.c: Include tree-vector-builder.h.
85 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
86 * expmed.c: Include tree-vector-builder.h.
87 (make_tree): Use tree_vector_builder instead of build_vector.
88 * fold-const.c: Include tree-vector-builder.h.
89 (const_binop): Use tree_vector_builder instead of build_vector.
90 (const_unop): Likewise.
91 (native_interpret_vector): Likewise.
92 (fold_vec_perm): Likewise.
93 (fold_ternary_loc): Likewise.
94 * gimple-fold.c: Include tree-vector-builder.h.
95 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
97 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
98 (simplify_vector_constructor): Use tree_vector_builder instead
100 * tree-vect-generic.c: Include tree-vector-builder.h.
101 (add_rshift): Use tree_vector_builder instead of build_vector.
102 (expand_vector_divmod): Likewise.
103 (optimize_vector_constructor): Likewise.
104 * tree-vect-loop.c: Include tree-vector-builder.h.
105 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
106 of build_vector. Explicitly use a stepped encoding for
108 * tree-vect-slp.c: Include tree-vector-builder.h.
109 (vect_get_constant_vectors): Use tree_vector_builder instead
111 (vect_transform_slp_perm_load): Likewise.
112 (vect_schedule_slp_instance): Likewise.
113 * tree-vect-stmts.c: Include tree-vector-builder.h.
114 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
115 (vect_gen_perm_mask_any): Likewise.
116 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
117 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
119 (build_vector_from_val): Likewise. Explicitly use a duplicate
122 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
124 * doc/generic.texi (VECTOR_CST): Describe new representation of
126 * vector-builder.h: New file.
127 * tree-vector-builder.h: Likewise.
128 * tree-vector-builder.c: Likewise.
129 * Makefile.in (OBJS): Add tree-vector-builder.o.
130 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
131 * tree-core.h (tree_base): Add a vector_cst field to the u union.
132 (tree_vector): Change the number of elements to
133 vector_cst_encoded_nelts.
134 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
135 (VECTOR_CST_ELTS): Delete.
136 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
137 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
138 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
139 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
140 (VECTOR_CST_ENCODED_ELT): Likewise.
141 (vector_cst_encoded_nelts): New function.
142 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
143 VECTOR_CST_NELTS_PER_PATTERN as arguments.
144 (vector_cst_int_elt, vector_cst_elt): Declare.
145 * tree.c: Include tree-vector-builder.h.
146 (tree_code_size): Abort if passed VECTOR_CST.
147 (tree_size): Update for new VECTOR_CST layout.
148 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
149 VECTOR_CST_NELTS_PER_PATTERN as arguments.
150 (build_vector): Use tree_vector_builder.
151 (vector_cst_int_elt, vector_cst_elt): New functions.
152 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
153 encoded elements and then create the vector in the canonical form.
154 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
155 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
156 (tree_c_tests): Call test_vector_cst_patterns.
157 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
159 (hash_tree): Likewise.
160 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
161 (streamer_write_tree_header): Likewise.
162 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
163 (streamer_alloc_tree): Likewise. Update call to make_vector.
164 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
166 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
168 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
169 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
170 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
172 2017-12-07 Bin Cheng <bin.cheng@arm.com>
173 Richard Biener <rguenther@suse.de>
175 PR tree-optimization/81303
176 * Makefile.in (gimple-loop-interchange.o): New object file.
177 * common.opt (floop-interchange): Reuse the option from graphite.
178 * doc/invoke.texi (-floop-interchange): Ditto. New document for
179 -floop-interchange and mention it for -O3.
180 * opts.c (default_options_table): Enable -floop-interchange at -O3.
181 * gimple-loop-interchange.cc: New file.
182 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
183 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
184 * passes.def (pass_linterchange): New pass.
185 * timevar.def (TV_LINTERCHANGE): New time var.
186 * tree-pass.h (make_pass_linterchange): New declaration.
187 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
188 interchange. Record IV before/after increment in new parameters.
189 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
190 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
192 (check_reduction_path): ...New function here.
193 * tree-vectorizer.h (check_reduction_path): New declaration.
195 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
198 PR rtl-optimization/80818
199 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
200 always early clobbered.
201 * lra-lives.c (process_bb_lives): Check input hard regs for early
202 clobbered non-operand hard reg.
204 2017-12-07 Jakub Jelinek <jakub@redhat.com>
207 * tree-cfg.c (verify_gimple_assign_binary): Don't require
208 types_compatible_p, just that TYPE_MODE is the same.
210 2017-12-07 Martin Sebor <msebor@redhat.com>
213 * attribs.c (empty_attribute_table): Initialize new member of
214 struct attribute_spec.
215 (decl_attributes): Add argument. Handle mutually exclusive
216 combinations of attributes.
217 (selftests::test_attribute_exclusions): New function.
218 (selftests::attribute_c_tests): Ditto.
219 * attribs.h (decl_attributes): Add default argument.
220 * selftest.h (attribute_c_tests): Declare.
221 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
222 * tree-core.h (attribute_spec::exclusions, exclude): New type and
224 * doc/extend.texi (Common Function Attributes): Update const and pure.
225 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
226 of struct attribute_spec.
227 * config/arc/arc.c (arc_attribute_table): Same.
228 * config/arm/arm.c (arm_attribute_table): Same.
229 * config/avr/avr.c ( avr_attribute_table): Same.
230 * config/bfin/bfin.c (bfin_attribute_table): Same.
231 * config/cr16/cr16.c (cr16_attribute_table): Same.
232 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
233 * config/h8300/h8300.c (h8300_attribute_table): Same.
234 * config/i386/i386.c (ix86_attribute_table): Same.
235 * config/ia64/ia64.c (ia64_attribute_table): Same.
236 * config/m32c/m32c.c (m32c_attribute_table): Same.
237 * config/m32r/m32r.c (m32r_attribute_table): Same.
238 * config/m68k/m68k.c (m68k_attribute_table): Same.
239 * config/mcore/mcore.c (mcore_attribute_table): Same.
240 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
241 * config/mips/mips.c (mips_attribute_table): Same.
242 * config/msp430/msp430.c (msp430_attribute_table): Same.
243 * config/nds32/nds32.c (nds32_attribute_table): Same.
244 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
245 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
246 * config/rl78/rl78.c (rl78__attribute_table): Same.
247 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
248 * onfig/rx/rx.c (rx_attribute_table): Same.
249 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
250 * config/sh/sh.c (sh_attribute_table): Same.
251 * config/sparc/sparc.c (sparc_attribute_table): Same.
252 * config/spu/spu.c (spu_attribute_table): Same.
253 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
254 * config/v850/v850.c (v850_attribute_table): Same.
255 * config/visium/visium.c (visium_attribute_table): Same.
257 2017-12-07 Tamar Christina <tamar.christina@arm.com>
260 * config/arm/arm.c (INCLUDE_STRING): Define.
261 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
262 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
265 2017-12-07 Michael Matz <matz@suse.de>
267 Add unroll and jam pass
269 * gimple-loop-jam.c: New file.
270 * Makefile.in (OBJS): Add gimple-loop-jam.o.
271 * common.opt (funroll-and-jam): New option.
272 * opts.c (default_options_table): Add unroll-and-jam at -O3.
273 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
274 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
275 * passes.def: Add pass_loop_jam.
276 * timevar.def (TV_LOOP_JAM): Add.
277 * tree-pass.h (make_pass_loop_jam): Declare.
278 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
279 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
280 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
281 to flow_loop_tree_node_add.
282 (duplicate_subloops, copy_loops_to): Append to sibling list.
283 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
284 * doc/invoke.texi (-funroll-and-jam): Document new option.
285 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
287 2017-12-07 Richard Biener <rguenther@suse.de>
289 PR tree-optimization/83296
290 PR tree-optimization/67769
291 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
292 flow sensitive info in an unrelated BB.
293 (value_replacement): Use reset_flow_sensitive_info.
294 (minmax_replacement): Reset flow sensitive info on the def
295 we move. Do not reset flow sensitive info in the whole BB
297 (abs_replacement): Likewise.
299 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
302 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
303 rs6000_cpu to the given -mcpu=, or to the default processor.
305 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
307 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
308 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
309 instead of rs6000_cpu_attr.
310 (rs6000_adjust_cost): Ditto.
311 (is_microcoded_insn): Ditto.
312 (rs6000_adjust_priority): Ditto.
313 (rs6000_issue_rate): Ditto.
314 (rs6000_use_sched_lookahead): Ditto.
315 (rs6000_use_sched_lookahead_guard): Ditto.
316 (rs6000_sched_reorder): Ditto.
317 (force_new_group): Ditto.
318 * config/rs6000/rs6000.md (cpu attribute): Ditto.
319 (group_ending_nop): Ditto.
321 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
323 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
324 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
325 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
326 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
328 (rs6000_reassociation_width): Ditto.
329 (rs6000_emit_epilogue): Ditto.
330 (rs6000_adjust_cost): Ditto.
331 (is_microcoded_insn): Ditto.
332 (is_cracked_insn): Ditto.
333 (rs6000_adjust_priority): Ditto.
334 (rs6000_sched_reorder): Ditto.
335 (rs6000_sched_reorder2): Ditto.
336 (insn_must_be_first_in_group): Ditto.
337 (insn_must_be_last_in_group): Ditto.
338 (rs6000_register_move_cost): Ditto.
339 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
342 2017-12-07 Julia Koval <julia.koval@intel.com>
344 * config.gcc: Add vaesintrin.h.
345 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI): New type.
346 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
347 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi): New builtins.
348 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
349 * config/i386/immintrin.h: Include vaesintrin.h.
350 * config/i386/sse.md (vaesdec_<mode>): New pattern.
351 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
352 _mm_aesdec_epi128): New intrinsics.
354 2017-12-06 David Malcolm <dmalcolm@redhat.com>
356 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
357 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
358 c-family/c-spellcheck.cc.
359 (best_macro_match::best_macro_match): Likewise.
361 (struct edit_distance_traits<cpp_hashnode *>): Move to
362 c-family/c-spellcheck.h.
363 (class best_macro_match): Likewise.
365 2017-12-06 Jakub Jelinek <jakub@redhat.com>
367 PR tree-optimization/83293
368 * gimple-ssa-strength-reduction.c (insert_initializers): Use
369 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
370 might insert into empty bb.
373 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
374 simplify for plus with :c added, and pointer_plus without that.
375 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
376 with undefined overflow and the conversion is not widening,
377 perform negation in utype and only convert to type afterwards.
378 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
379 simplify for plus with :c added, and pointer_plus without that.
380 If type is integral with undefined overflow and the conversion is
381 not widening, perform minus in utype and only convert to type
382 afterwards. Move the last pointer_diff_expr simplify into the
385 2017-12-06 Martin Sebor <msebor@redhat.com>
387 PR tree-optimization/82646
388 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
391 2017-12-06 Martin Sebor <msebor@redhat.com>
393 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
396 PR tree-optimization/83075
397 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
398 strncat/strncpy don't change length of source string.
400 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
403 2017-11-29 Martin Aberg <maberg@gaisler.com>
405 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
406 to prevent b2bst errata sequence.
407 (sqrtdf2_fix): Likewise.
409 2017-12-06 Jakub Jelinek <jakub@redhat.com>
411 PR tree-optimization/81945
412 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
413 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
414 to dest_cfun has orig_loop_num set, either remap it to the new
415 loop number if the loop got moved too, or clear it.
417 2017-12-05 Steve Ellcey <sellcey@cavium.com>
419 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
421 (thunderx2t99_nothing): New insn reservation.
422 (thunderx2t99_mrs): New insn reservation.
423 (thunderx2t99_multiple): New insn reservation.
424 (thunderx2t99_alu_basi): Add bfx to reservation.
425 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
427 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
430 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
432 2017-12-05 Bin Cheng <bin.cheng@arm.com>
434 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
435 tree-ssa-pre.c::remove_dead_inserted_code.
436 * tree-ssa-dce.h: New file.
437 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
438 (remove_dead_inserted_code): Move and rename to function
439 tree-ssa-dce.c::simple_dce_from_worklist.
440 (pass_pre::execute): Update use.
442 2017-12-05 Richard Biener <rguenther@suse.de>
444 PR tree-optimization/83277
445 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
446 to code-gen liveout vars.
448 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
450 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
451 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
452 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
455 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
457 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
459 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
460 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
463 2017-12-05 Richard Biener <rguenther@suse.de>
465 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL):
467 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
468 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
469 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
471 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
473 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
475 2017-12-05 Martin Liska <mliska@suse.cz>
476 Jakub Jelinek <jakub@redhat.com>
478 * doc/invoke.texi: Document the options.
479 * flag-types.h (enum sanitize_code): Add
480 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
481 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
482 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
483 * opts.c: Define new sanitizer options.
484 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
485 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
487 2017-12-05 Julia Koval <julia.koval@intel.com>
489 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
490 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
491 (ix86_handle_option): Handle -mavx512vnni.
492 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
493 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
494 * config/i386/i386-c (__AVX512VNNI__): New.
495 * config/i386/i386.c (ix86_target_string): Handle new option.
496 (ix86_valid_target_attribute_inner_p): Handle new option.
497 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
498 * config/i386/i386.opt (mavx512vnni): New option.
500 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
503 * athlon.md: Disable for generic.
504 * haswell.md: Enable for generic.
505 * i386.c (ix86_sched_init_global): Add core hooks for generic.
506 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
508 (ix86_adjust_cost): Move generic to haswell path.
510 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
512 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
513 instead of MEM_P in a couple more places. Fix formatting issues.
515 2017-12-04 Jim Wilson <jimw@sifive.com>
517 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
518 instead of GP_REG_LAST-1.
519 (riscv_adjust_libcall_cfi_prologue): Likewise.
520 (riscv_adjust_libcall_cri_epilogue): Likewise.
521 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
524 2017-12-04 Luis Machado <luis.machado@linaro.org>
526 * ipa-pure-const.c (check_decl): Add missing newline.
527 (state_from_flags): Likewise.
529 2017-12-04 Jeff Law <law@redhat.com>
531 PR tree-optimizatin/78496
532 * gimple-ssa-evrp-analyze.h
533 (evrp_range_analyzer::get_vr_values): Simplify.
534 * gimple-ssa-evrp-analyze.c: Corresponding changes.
535 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
536 and gimple-ssa-evrp-analyze.h.
537 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
538 (simplify_stmt_for_jump_threading): Copy a blob of code from
539 tree-vrp.c to use ranges to simplify statements.
540 (dom_opt_dom_walker::before_dom_children): Call
541 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
542 (dom_opt_dom_walker::after_dom_children): Similarly for
543 evrp_range_analyzer::leave.
544 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
547 * gimple-ssa-evrp-analyze.c
548 (evrp_range_analyzer::extract_range_from_stmt): Always use
549 vr_values::update_value_range so preexisting range info is
550 medged with new range info, even if the new range is VR_VARYING.
552 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
554 * combine.c: Adjust comment.
555 (use_crosses_set_p): Delete.
556 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
557 (try_combine): Ditto.
559 2017-12-04 Richard Biener <rguenther@suse.de>
561 PR tree-optimization/83255
562 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
563 Re-add zero-iteration check.
565 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
567 PR rtl-optimization/83245
568 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
569 hard registers as earlyclobber, also if not in an asm.
571 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
575 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
578 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
579 rs6000_cpu based on cpu_index, not tune_index.
581 2017-12-04 Richard Biener <rguenther@suse.de>
583 PR tree-optimization/83238
584 * graphite-scop-detection.c (scop_detection::merge_sese): Make
585 code match comment, rejecting invalid SESE regions.
587 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
589 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
590 require base operand is a REG_POINTER prior to reload on targets
591 with non-equivalent space registers.
593 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
595 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
596 (update_specialized_profile): Fix updating of counts.
597 (perhaps_add_new_callers): Likewise.
599 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
602 * x86-tune.def: Remove obsolette FIXMEs.
603 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
604 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
605 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
607 (X86_TUNE_PAD_RETURNS): Disable for generic.
609 2017-12-02 Jakub Jelinek <jakub@redhat.com>
611 PR tree-optimization/83170
612 PR tree-optimization/83241
613 * gimple-ssa-store-merging.c
614 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
615 gimple_vuse (ins_stmt) in case it has changed.
616 (imm_store_chain_info::output_merged_store): Likewise.
618 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
622 * tree-cfg.c (pass_warn_function_return::execute): Handle
623 __builtin_ubsan_handle_missing_return like __builtin_unreachable
624 with BUILTINS_LOCATION.
628 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
629 is BLKmode for vector field with vector raw mode, use TYPE_MODE
630 instead of DECL_MODE.
632 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
633 last argument to rtx pointer.
634 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
635 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
636 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
637 depending on the chosen ISAs.
638 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
639 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
640 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
642 * config/i386/sse.md (mov<mode>_internal): Likewise.
643 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
645 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
647 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
649 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
651 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
652 parameters from prototype.
653 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
654 parameters. Don't print a comment.
655 (emit_fusion_gpr_load): Adjust.
656 (emit_fusion_load_store): Adjust.
657 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
658 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
659 comment on the second line.
661 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
664 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
665 rs6000_cpu based on cpu_index, not tune_index.
667 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
669 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
670 which_alternative instead of which_alternative + 1.
671 (output_asm_insn): Print an extra tab if the template is short.
673 2017-12-01 Jim Wilson <jimw@sifive.com>
675 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
677 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
678 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
680 * doc/tm.texi: Regenerate.
682 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
685 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
686 whether we can allocate pseudos before trying to fix an address.
687 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
688 memory address is indexed or indirect.
689 (floatuns_<mode>si2_hw2): Likewise.
691 2017-12-01 Jason Merrill <jason@redhat.com>
693 * Makefile.in (TAGS): Add c-family/*.cc.
695 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
697 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
698 (call_value_insn): Likewise.
699 (sibcall_insn): Likewise.
700 (sibcall_value_insn): Likewise.
701 (movsi_aarch64): Likewise.
702 (movdi_aarch64): Likewise.
703 (add_losym_): Likewise.
704 (ldr_got_small_): Likewise.
705 (ldr_got_small_sidi): Likewise.
706 (ldr_got_small_28k_): Likewise.
707 (ldr_got_small_28k_sidi): Likewise.
708 * config/aarch64/aarch64.c (aarch64_print_address_internal):
709 Move output_addr_const to symbolic case. Add error check.
711 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
713 * config/s390/predicates.md (plus16_Q_operand): New predicate.
714 * config/s390/s390.md: Disable MVC merging peephole if it would
715 disable operand forwarding.
716 (new peephole2): Split MVCs if it would turn them into up to 2
719 2017-12-01 Richard Biener <rguenther@suse.de>
721 PR tree-optimization/83232
722 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
723 detection of same access. Instead of breaking the group here
724 do not consider the duplicate. Add comment explaining real fix.
726 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
728 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
730 2017-12-01 Sudakshina Das <sudi.das@arm.com>
732 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
734 2017-12-01 Jakub Jelinek <jakub@redhat.com>
736 * function.h (struct function): Remove cilk_frame_decl,
737 is_cilk_function and calls_cilk_spawn fields.
738 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
740 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
741 cilk_elemental field.
742 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
743 * target.def: Adjust comment.
744 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
745 Don't test cilk_elemental.
747 PR tree-optimization/83233
748 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
749 bswap_stat name for the struct.
752 * tree.h (SWITCH_BREAK_LABEL_P): Define.
753 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
754 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
755 SWITCH_BREAK_LABEL_P set on the label.
756 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
757 added for default case if it was missing and not all cases covered.
758 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
759 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
761 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
762 to avoid -Wimplicit-fallthrough warning.
763 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
764 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
766 PR tree-optimization/83221
767 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
769 (init_reassoc): Formatting fix.
772 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
773 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
777 * tree-cfg.c: Include asan.h.
778 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
780 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
782 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
784 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
786 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
787 vpcmpeqd instruction.
789 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
791 * config/i386/i386.c (standard_sse_constant_opcode): Fix
792 registers type for 128bit mode.
794 2017-11-30 Jakub Jelinek <jakub@redhat.com>
796 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
798 * gengtype-state.c (read_state_pair): Likewise.
799 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
800 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
801 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
802 * ifcvt.c (noce_try_store_flag_constants): Likewise.
803 * tree-ssa-ccp.c (ccp_finalize): Likewise.
804 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
805 * builtins.c (fold_builtin_3): Likewise.
806 * graphite-scop-detection.c
807 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
808 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
811 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
814 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
815 parameter. Return true if ignore_decl_rtl_set_p is true and other
816 conditions are satisfied.
817 * asan.h (asan_protect_global): Add new parameter.
818 * varasm.c (categorize_decl_for_section): Pass true as second parameter
819 to asan_protect_global calls.
821 2017-11-30 Jim Wilson <jimw@sifive.com>
823 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
824 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
825 -mstrict-align, add info on default value. Delete redundant lines for
826 -mabi. Add missing -mexplicit-relocs docs.
828 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
830 * config/arc/arc.md (trap): New pattern.
832 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
834 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
835 instruction to end into a delay slot.
836 * config/arc/arc.md (cond_delay_insn): Check if the instruction
837 can be placed into a delay slot against reg_note.
839 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
841 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
842 labels number of usages.
844 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
846 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
848 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
850 2017-11-30 Jakub Jelinek <jakub@redhat.com>
853 * internal-fn.c (expand_mul_overflow): Optimize unsigned
854 multiplication by power of 2 constant into two shifts + comparison.
856 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
859 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
861 2017-11-30 Richard Biener <rguenther@suse.de>
863 PR tree-optimization/83202
864 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
865 allow_peel argument and guard peeling.
866 (canonicalize_loop_induction_variables): Likewise.
867 (canonicalize_induction_variables): Pass false.
868 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
869 peeling from cunrolli.
871 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
873 * combine.c (try_combine): Print a message to dump file whenever
874 I0, I1, or I2 cannot be combined into I3.
876 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
878 PR rtl-optimization/83156
879 PR rtl-optimization/82621
880 * combine.c (try_combine): Don't split an I2 if one of the dests is
881 set again before I3. Allow unused dests.
883 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
885 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
887 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
889 PR rtl-optimization/80818
890 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
891 recursively. Use insn code for clobber.
892 (lra_set_insn_recog_data): Pass the new arg to
893 collect_non_operand_hard_regs.
894 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
896 (lra_update_insn_regno_info): Pass insn to
897 add_regs_to_insn_regno_info.
899 2017-11-29 Jim Wilson <jimw@sifive.com>
900 Andrew Waterman <andrew@sifive.com>
902 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
903 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
904 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
905 use SINGLE_SHIFT_COST.
906 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
907 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
909 2017-11-29 Julia Koval <julia.koval@intel.com>
911 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
912 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
913 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
914 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
915 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
916 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
917 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
918 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
919 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
920 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
921 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
922 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
923 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
924 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
925 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
926 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
927 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
928 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
929 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
930 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
931 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
932 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
933 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
934 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
935 __builtin_ia32_vpshldv_v2di_mask,
936 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
937 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
938 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
940 2017-11-29 Julia Koval <julia.koval@intel.com>
942 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
943 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
944 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
945 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
946 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
947 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
948 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
949 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
950 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
951 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
952 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
953 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
954 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
955 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
956 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
957 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
958 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
959 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
960 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
961 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
962 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
964 2017-11-29 Daniel Cederman <cederman@gaisler.com>
966 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
967 movsi_pic_gotdata_op instruction as a load for the UT699 errata
970 2017-11-29 Martin Aberg <maberg@gaisler.com>
972 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
973 to prevent b2bst errata sequence.
974 (sqrtdf2_fix): Likewise.
976 2017-11-29 Daniel Cederman <cederman@gaisler.com>
978 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
979 (div_sqrt_insn_p): New function.
980 (sparc_do_work_around_errata): Insert NOP instructions to
981 prevent sequences that could trigger the TN-0013 errata for
982 certain LEON3 processors.
983 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
984 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
985 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
986 (in_branch_delay): Prevent div and sqrt in delay slot if
988 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
990 2017-11-29 Daniel Cederman <cederman@gaisler.com>
992 * config/sparc/sparc.c (atomic_insn_p): New function.
993 (sparc_do_work_around_errata): Insert NOP instructions to
994 prevent sequences that could trigger the TN-0010 errata for
996 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
997 instruction referable in atomic_insns_p.
999 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1001 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
1002 (atomic_compare_and_swap_leon3_1): Likewise.
1005 2017-11-29 Daniel Cederman <cederman@gaisler.com>
1007 * config/sparc/sparc.c (fpop_insn_p): New function.
1008 (sparc_do_work_around_errata): Insert NOP instructions to
1009 prevent sequences that could trigger the TN-0012 errata for
1011 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
1012 * config/sparc/sparc.md (fix_gr712rc): New attribute.
1013 (in_branch_annul_delay): Prevent floating-point instructions
1014 in delay slot of annulled integer branch.
1016 2017-11-29 Richard Biener <rguenther@suse.de>
1018 PR tree-optimization/83202
1019 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
1021 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
1022 nodes vectorized to the same stmts multiple times.
1023 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
1024 (vect_analyze_slp_instance): Adjust.
1025 (scalar_stmts_to_slp_tree_map_t): New typedef.
1026 (vect_schedule_slp_instance): Add a map recording the SLP node
1027 representing the vectorized stmts for a set of scalar stmts.
1028 Avoid code-generating redundancies.
1029 (vect_schedule_slp): Allocate map and pass it down.
1031 2017-11-29 Nathan Sidwell <nathan@acm.org>
1034 * tree.c (build_complex_type): Fix canonicalization. Only fill in
1037 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
1039 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
1040 cases for printing LDP/STP memory addresses.
1041 (aarch64_print_address_internal): Renamed from
1042 aarch64_print_operand_address, added parameter, add Pmode check.
1043 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
1044 (aarch64_print_operand_address): Indirect to
1045 aarch64_print_address_internal.
1046 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
1047 'y' operand output specifier.
1049 2017-11-29 Jakub Jelinek <jakub@redhat.com>
1052 * tree.c (build_simple_mem_ref_loc): Handle
1053 get_addr_base_and_unit_offset returning a MEM_REF.
1056 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1059 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1062 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1064 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1066 (layout::move_to_column): Likewise.
1068 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1071 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1072 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1074 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1075 Martin Jambor <mjambor@suse.cz>
1078 * tree.h (expr_type_first_operand_type_p): Declare
1079 * tree.c (expr_type_first_operand_type_p): New function.
1080 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1081 (ipa_value_from_jfunc): Adjust declaration.
1082 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1083 Use it as result type for arithmetics, unless it is NULL in which case
1084 be more conservative.
1085 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1086 ipa_get_jf_pass_through_result.
1087 (propagate_vals_across_pass_through): Likewise.
1088 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1089 is to propagate_vals_across_pass_through.
1090 (propagate_constants_across_call): Pass PARM_TYPE to
1091 propagate_scalar_across_jump_function.
1092 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1093 ipa_value_from_jfunc.
1094 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1095 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1096 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1097 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1098 target_type, pass it to ipa_value_from_jfunc.
1099 (update_indirect_edges_after_inlining): Pass parameter type to
1100 try_make_edge_direct_simple_call.
1102 2017-11-28 Jeff Law <law@redhat.com>
1104 * gimple-ssa-evrp-analyze.c
1105 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1106 refine ranges if scev_initialized_p returns true.
1107 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1109 2017-11-28 Julia Koval <julia.koval@intel.com>
1111 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1112 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
1113 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
1114 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
1115 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
1116 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
1117 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
1118 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
1119 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
1120 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
1121 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
1122 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
1123 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
1124 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
1125 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
1126 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
1127 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
1128 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
1129 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
1130 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
1131 __builtin_ia32_vpshrd_v2di_mask): New builtins.
1132 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
1134 2017-11-28 Julia Koval <julia.koval@intel.com>
1136 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
1137 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
1138 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
1139 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
1140 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
1141 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
1142 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
1143 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
1144 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
1145 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
1146 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
1147 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
1148 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
1149 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
1150 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
1151 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
1152 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
1153 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
1154 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
1155 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
1156 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
1157 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
1158 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
1159 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
1160 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
1161 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
1162 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
1163 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
1164 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
1165 __builtin_ia32_vpshld_v2di_mask): New builtins.
1166 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1167 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
1169 2017-11-28 Richard Biener <rguenther@suse.de>
1171 PR tree-optimization/80776
1172 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
1174 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
1176 (evrp_range_analyzer::record_ranges_from_incoming_edges):
1177 If the incoming edge is an effective fallthru because the other
1178 edge only reaches a __builtin_unreachable () then record ranges
1179 derived from the controlling condition in SSA info.
1180 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
1181 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1183 2017-11-28 Olivier Hainque <hainque@adacore.com>
1185 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
1186 on mingw build hosts.
1188 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1191 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
1193 2017-11-28 Richard Biener <rguenther@suse.de>
1196 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
1197 copies generated from memcpy use a character array as reference
1200 2017-11-28 Julia Koval <julia.koval@intel.com>
1201 Sebastian Peryt <sebastian.peryt@intel.com>
1203 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
1204 c-family/c-cilkplus.o, c-family/array-notation-common.o,
1205 cilk-common.o, cilk.h, cilk-common.c): Remove.
1207 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
1208 * builtins.c (is_builtin_name): Remove cilkplus condition.
1209 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
1210 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
1211 cilk-builtins.def, cilkplus.def): Remove.
1212 * cif-code.def (CILK_SPAWN): Remove.
1213 * cilk-builtins.def: Delete.
1214 * cilk-common.c: Ditto.
1216 * cilkplus.def: Ditto.
1217 * config/darwin.h (fcilkplus): Delete.
1218 * cppbuiltin.c: Ditto.
1219 * doc/extend.texi: Remove cilkplus doc.
1220 * doc/generic.texi: Ditto.
1221 * doc/invoke.texi: Ditto.
1222 * doc/passes.texi: Ditto.
1223 * gcc.c (fcilkplus): Remove.
1224 * gengtype.c (cilk.h): Remove.
1225 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
1227 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
1229 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
1231 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
1232 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
1233 cilkplus conditions.
1234 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
1235 inline_read_section): Ditto.
1236 * ipa-inline-analysis.c (cilk.h): Remove.
1237 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
1238 * lto-wrapper.c (merge_and_complain, append_compiler_options,
1239 append_linker_options): Remove condition for fcilkplus.
1240 * lto/lto-lang.c (cilk.h): Remove.
1241 (lto_init): Remove condition for fcilkplus.
1242 * omp-expand.c (expand_cilk_for_call): Delete.
1243 (expand_omp_taskreg, expand_omp_for_static_chunk,
1244 expand_omp_for): Remove cilkplus
1246 (expand_cilk_for): Delete.
1247 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
1248 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
1249 execute_lower_omp, diagnose_sb_0): Ditto.
1250 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
1251 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1252 * tree-nested.c: Ditto.
1253 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
1254 (dump_generic_node): Ditto.
1255 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1256 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
1258 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
1260 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1262 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
1263 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
1264 for vector mode and !TARGET_SIMD.
1266 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1268 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
1270 * tree.h (SWITCH_LABELS): Remove.
1271 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
1272 assert SWITCH_BODY is non-NULL.
1273 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
1275 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
1277 PR tree-optimization/80788
1278 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
1279 has TREE_OVERFLOW set, call drop_tree_overflow.
1281 2017-11-28 Richard Biener <rguenther@suse.de>
1283 PR tree-optimization/83158
1284 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more
1287 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
1290 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
1291 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
1293 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
1295 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
1296 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
1298 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1301 * varasm.c (compare_constant): Take the mode of the constants into
1302 account when comparing floating point constants.
1304 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
1306 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
1307 from explicit instantiation of debug_helper.
1308 * vec.h (DEFINE_DEBUG_VEC): Ditto.
1310 2017-11-27 Richard Biener <rguenther@suse.de>
1312 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
1315 2017-11-27 Richard Biener <rguenther@suse.de>
1317 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
1318 (type_hash_canon): Also clear int_cst_hash_table entry for
1320 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
1322 2017-11-27 Tamar Christina <tamar.christina@arm.com>
1324 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1325 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1326 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1328 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1330 * hash-map.h (gt_cleare_cache): Avoid UB.
1332 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
1334 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
1335 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
1336 (decide_unroll_runtime_iterations): Likewise.
1337 (decide_unroll_stupid): Likewise.
1339 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1342 * config/i386/i386.md: Add a loop with incssp.
1344 2017-11-27 Martin Jambor <mjambor@suse.cz>
1346 PR tree-optimization/81248
1347 * tree-sra.c (splice_param_accesses): Remove size check.
1348 (decide_one_param_reduction): Fix size check.
1349 * gimple-pretty-print.c (dump_profile): Silence warning.
1350 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
1352 2017-11-27 Jakub Jelinek <jakub@redhat.com>
1355 * dbxout.c (lastlineno): New variable.
1356 (dbx_debug_hooks): Use dbxout_switch_text_section as
1357 switch_text_section debug hook.
1358 (dbxout_function_end): Switch to current_function_section
1359 rather than function_section. If crtl->has_bb_partition,
1360 output just one N_FUN, depending on in_cold_section_p.
1361 (dbxout_source_line): Remember last lineno in lastlineno.
1362 (dbxout_switch_text_section): New function.
1363 (dbxout_function_decl): Adjust dbxout_block caller.
1364 (dbx_block_with_cold_children): New function.
1365 (dbxout_block): Return true if any LBRAC/RBRAC have been
1366 emitted. Use dbx_block_with_cold_children at depth == 0
1367 in second partition. Add PARENT_BLOCKNUM argument, pass
1368 it optionally adjusted to children. Output LBRAC/RBRAC
1369 around recursive call only if the block is in the current
1370 partition, if not and anything was output, emit empty
1372 * final.c (final_scan_insn): Compute cold_function_name
1373 before calling switch_text_section debug hook. Call
1374 that hook even if dwarf2out_do_frame if not emitting
1378 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
1379 TREE_READONLY decls.
1381 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1383 PR rtl-optimization/82488
1384 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
1386 2017-11-26 Julia Koval <julia.koval@intel.com>
1388 * config/i386/i386.c (processor_target_table): Add skylake_cost for
1390 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
1393 2017-11-26 Julia Koval <julia.koval@intel.com>
1395 * config/i386/driver-i386.c (host_detect_local_cpu):
1396 Detect skylake-avx512.
1398 2017-11-26 Julia Koval <julia.koval@intel.com>
1400 * config.gcc: Add -march=cannonlake.
1401 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
1402 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
1403 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
1404 (PTA_CANNONLAKE): New.
1405 (processor_target_table): Add cannonlake.
1406 (ix86_option_override_internal): Ditto.
1407 (fold_builtin_cpu): Ditto.
1408 (get_builtin_code_for_version): Handle cannonlake.
1409 (M_INTEL_COREI7_CANNONLAKE): New.
1410 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
1411 * doc/invoke.texi: Add -march=cannonlake.
1413 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
1415 * plugin.c (add_new_plugin): Use platform-specific library extensions.
1416 (try_init_one_plugin): Alternative implementation for MinGW.
1417 * Makefile.in (plugin_implib): New.
1418 (gengtype-lex.c): Fix broken AIX workaround.
1419 * configure: Regenerate.
1420 * doc/plugins.texi: Document support for MinGW.
1422 2017-11-25 Jakub Jelinek <jakub@redhat.com>
1424 PR rtl-optimization/81553
1425 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
1426 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
1427 is a shift where C1 has different mode than the whole shift, use C1's
1428 mode for MULT rather than the shift's mode.
1431 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
1432 builtins not enabled in the currently selected ISA.
1434 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
1436 PR tree-optimization/71026
1437 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
1438 (insert_reciprocals): Change to insert reciprocals before a division
1439 by a square and to insert the square of a reciprocal.
1440 (execute_cse_reciprocals_1): Change to consider division by a square.
1441 (register_division_in): Add importance parameter.
1443 2017-11-24 Richard Biener <rguenther@suse.de>
1445 PR tree-optimization/82402
1446 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
1447 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
1449 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
1451 * match.pd (0-ptr): New transformation.
1453 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
1456 * ipa-inline.c (inline_small_functions): Set current badnes correctly
1457 when skipping checking.
1459 2017-11-24 Richard Biener <rguenther@suse.de>
1461 PR tree-optimization/83128
1462 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
1463 (vn_reference_lookup_3): Likewise.
1465 2017-11-24 Jakub Jelinek <jakub@redhat.com>
1468 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
1469 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
1472 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
1473 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
1474 check for the latter with gimple_call_builtin_p. Do not handle
1475 BUILT_IN_STPNCPY_CHK which is not a pass through call.
1477 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
1479 * config/arm/arm_neon.h: Fix pragma GCC push_options before
1482 2017-11-23 Julia Koval <julia.koval@intel.com>
1484 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
1485 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
1486 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
1487 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
1488 _mm512_maskz_expandloadu_epi16): New intrinsics.
1489 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
1490 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
1491 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
1492 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
1493 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
1494 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
1495 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
1496 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
1497 _mm256_maskz_expandloadu_epi8): New intrinsics.
1498 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
1499 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
1500 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
1501 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
1502 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
1503 * config/i386/sse.md (VI248_VLBW): New iterator.
1504 (expand<mode>_mask, expand<mode>_maskz): New patterns.
1506 2017-11-23 Julia Koval <julia.koval@intel.com>
1508 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
1509 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
1510 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
1511 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
1512 _mm512_mask_compressstoreu_epi16): New.
1513 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
1514 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
1515 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
1516 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
1517 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
1518 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
1519 _mm256_mask_compressstoreu_epi8): New.
1520 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
1521 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
1522 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
1523 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
1524 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
1525 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
1526 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
1527 __builtin_ia32_compresshi128_mask,
1528 __builtin_ia32_compressstoreuqi512_mask,
1529 __builtin_ia32_compressstoreuhi512_mask,
1530 __builtin_ia32_compressstoreuqi256_mask,
1531 __builtin_ia32_compressstoreuqi128_mask,
1532 __builtin_ia32_compressstoreuhi256_mask,
1533 __builtin_ia32_compressstoreuhi128_mask): New builtins.
1534 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
1536 (ix86_expand_special_args_builtin): Handle new types.
1537 (s4fma_expand): Handle new builtin array.
1538 * config/i386/immintrin.h: Include new headers.
1539 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
1540 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
1542 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1544 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
1545 fixes. Declare temp and g variables at the top in order to avoid
1546 {} in most of the cases.
1548 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
1550 * match.pd (ptr-0): New transformation.
1552 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1554 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
1555 (struct addr_mode_cost_table): New.
1556 (struct tune_params): Add field addr_mode_costs.
1557 * config/arm/arm.c (generic_addr_mode_costs): New.
1558 (arm_slowmul_tune): Initialise addr_mode_costs field.
1559 (arm_fastmul_tune): Likewise.
1560 (arm_strongarm_tune): Likewise.
1561 (arm_xscale_tune): Likewise.
1562 (arm_9e_tune): Likewise.
1563 (arm_marvell_pj4_tune): Likewise.
1564 (arm_v6t2_tune): Likewise.
1565 (arm_cortex_tune): Likewise.
1566 (arm_cortex_a8_tune): Likewise.
1567 (arm_cortex_a7_tune): Likewise.
1568 (arm_cortex_a15_tune): Likewise.
1569 (arm_cortex_a35_tune): Likewise.
1570 (arm_cortex_a53_tune): Likewise.
1571 (arm_cortex_a57_tune): Likewise.
1572 (arm_exynosm1_tune): Likewise.
1573 (arm_xgene1_tune): Likewise.
1574 (arm_cortex_a5_tune): Likewise.
1575 (arm_cortex_a9_tune): Likewise.
1576 (arm_cortex_a12_tune): Likewise.
1577 (arm_cortex_a73_tune): Likewise.
1578 (arm_v7m_tune): Likewise.
1579 (arm_cortex_m7_tune): Likewise.
1580 (arm_v6m_tune): Likewise.
1581 (arm_fa726te_tune): Likewise.
1582 (arm_mem_costs): Use table lookup to calculate cost of addressing
1585 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1587 * config/arm/arm.c (arm_mem_costs): New function.
1588 (arm_rtx_costs_internal): Use arm_mem_costs.
1590 2017-11-23 Mark Wielaard <mark@klomp.org>
1592 * dwarf2out.c (init_sections_and_labels): Use generation to create
1593 unique ranges_section_label and ranges_base_label. Return generation.
1594 (output_rnglists): Add generation argument. Use generation to create
1595 unique ranges labels.
1596 (dwarf2out_finish): Get generation from init_sections_and_labels
1597 and pass generation to output_rnglists.
1599 2017-11-23 Mike Stump <mikestump@comcast.net>
1600 Eric Botcazou <ebotcazou@adacore.com>
1602 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
1603 * cfgloop.h (struct loop): Add unroll field.
1604 * function.h (struct function): Add has_unroll bitfield.
1605 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
1606 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
1607 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
1608 (pass_rtl_unroll_loops::gate): Likewise.
1609 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
1610 for which loop->unroll==1.
1611 (decide_unroll_constant_iterations): Use note for consistency and
1612 take loop->unroll into account. Return early if loop->unroll is set.
1613 Fix thinko in existing test.
1614 (decide_unroll_runtime_iterations): Use note for consistency and
1615 take loop->unroll into account.
1616 (decide_unroll_stupid): Likewise.
1617 * lto-streamer-in.c (input_cfg): Read loop->unroll.
1618 * lto-streamer-out.c (output_cfg): Write loop->unroll.
1619 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
1621 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
1622 (print_loop): Print loop->unroll if set.
1623 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
1624 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
1625 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
1627 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
1628 loop->unroll is set and smaller than the trip count. Otherwise bypass
1629 entirely the heuristics if loop->unroll is set. Remove dead note.
1630 Fix off-by-one bug in other note.
1631 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
1632 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
1634 (tree_unroll_loops_completely): Make static.
1635 (pass_complete_unroll::execute): Use correct type for variable.
1636 (pass_complete_unrolli::execute): Fix formatting.
1637 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
1639 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
1641 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
1642 enable when TARGET_PREFER_AVX128 is set.
1644 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1646 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
1648 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
1649 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
1651 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1653 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
1654 frequencies. Fix estimation of aggregate parameters.
1656 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1658 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
1661 2017-11-23 Tom de Vries <tom@codesourcery.com>
1663 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
1665 2017-11-23 Tom de Vries <tom@codesourcery.com>
1667 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
1670 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
1673 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
1674 sibcall_value_pcrel_fdpic): Use local variable instead of
1676 (calli_tbr_rel): Add missing operand 2.
1677 (call_valuei_tbr_rel): Add missing operand 3.
1679 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1682 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
1683 bitpos/bitsize covering the whole destination, use store_expr only if
1684 the complex mode is the same. Otherwise, use expand_normal and if
1685 it returns CONCAT, subreg each part separately instead of trying to
1686 subreg the whole result.
1688 2017-11-23 Richard Biener <rguenther@suse.de>
1690 PR tree-optimization/23094
1691 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
1692 come here from walking over backedges in the first iteration.
1693 (vn_reference_lookup_3): Skip clobbers that store the same value.
1695 2017-11-23 Richard Biener <rguenther@suse.de>
1697 PR tree-optimization/81403
1698 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
1699 a block we need a leader relative to.
1700 (phi_translate_1): For nary processing require a leader from
1701 get_representative_for given we run expression simplification
1702 using match-and-simplify. Remove previous fix.
1704 2017-11-22 Jeff Law <law@redhat.com>
1706 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
1707 Use new method allocate_value_range rather than accessing the
1708 vrp_value_range_pool data member directly.
1709 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
1710 to use extract_range_from_stmt method to avoid need for
1711 extract_range_from_assignment method.
1712 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
1713 method rather than setting values_propgated data member directly.
1714 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
1715 and values propagated data members and extract_range_from_assignment
1716 method. Reorder private data members to conform to standards.
1717 Add new methods set_lattice_propagation_complete and
1718 allocate_value_range.
1720 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
1722 PR rtl-optimization/83030
1723 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
1724 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
1725 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
1726 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
1727 CROSSING_JUMP_P flag is already set before setting it.
1728 * cfgrtl.c (fixup_partition_crossing): Likewise.
1729 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
1732 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1734 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
1735 Handle the case where both arguments are using gen_const_vec_series.
1737 2017-11-22 David Malcolm <dmalcolm@redhat.com>
1740 * pretty-print.c (pp_format): Move quoting implementation to
1741 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
1742 to pass address of "quote" local.
1743 (pp_begin_quote): New function.
1744 (pp_end_quote): New function.
1745 * pretty-print.h (printer_fn): Convert penultimate param from bool
1747 (pp_begin_quote): New decl.
1748 (pp_end_quote): New decl.
1749 * tree-diagnostic.c (default_tree_printer): Convert penultimate
1750 param from bool to bool *.
1751 * tree-diagnostic.h (default_tree_printer): Likewise.
1753 2017-11-22 Jeff Law <law@redhat.com>
1755 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
1756 Initialize vr_values.
1757 (evrp_range_analyzer::try_find_new_range): Call methods attached to
1758 vr_values via vr_values class instance rather than delegators.
1759 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
1760 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1761 (evrp_range_analyzer::push_value_range): Likewise.
1762 (evrp_range_analyzer::pop_value_range): Likewise.
1763 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
1764 most delegators. Those remaining are exposed as public interfaces.
1765 Make vr_values a pointer and private.
1766 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
1768 (evrp_range_analyzer::get_vr_value): New method.
1769 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
1770 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
1771 (class evrp_dom_walker): Attach evrp_folder class, initialize
1772 it in the ctor. Remove temporary delegators.
1773 (evrp_dom_walker::before_dom_children): Call methods in attached
1774 evrp_range_analyzer class via class instance pointer. Use
1775 free value_range_constant_singleton to remove need for
1776 op_with_constant_singleton_value delegator method. Do not
1777 create a vrp_prop class instance for every call! Narrow
1778 scope of a couple variables.
1779 (evrp_dom_walker::cleanup): Call methods in attached
1780 evrp_range_analyzer class via class instance pointer.
1781 * vr-values.h (class vr_values): Privatize many methods and
1784 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1786 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
1787 * cfgexpand.c (expand_debug_expr): Remove handling for them.
1788 * expr.c (expand_expr_real_2): Likewise.
1789 * fold-const.c (const_unop): Likewise.
1790 * optabs-tree.c (optab_for_tree_code): Likewise.
1791 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1792 * tree-inline.c (estimate_operator_cost): Likewise.
1793 * tree-pretty-print.c (dump_generic_node): Likewise.
1794 (op_code_prio): Likewise.
1795 (op_symbol_code): Likewise.
1796 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
1797 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
1798 * internal-fn.c (direct_internal_fn_optab): New function.
1799 (direct_internal_fn_array, direct_internal_fn_supported_p
1800 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
1801 * fold-const-call.c (fold_const_reduction): New function.
1802 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
1804 * tree-vect-loop.c: Include internal-fn.h.
1805 (reduction_code_for_scalar_code): Rename to...
1806 (reduction_fn_for_scalar_code): ...this and return an internal
1808 (vect_model_reduction_cost): Take an internal_fn rather than
1810 (vect_create_epilog_for_reduction): Likewise. Build calls rather
1812 (vectorizable_reduction): Use internal functions rather than tree
1813 codes for the reduction operation. Update calls to the functions
1815 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
1816 Use calls to internal functions rather than REDUC tree codes.
1817 * config/aarch64/aarch64-simd.md: Update comment accordingly.
1819 2017-11-22 Olivier Hainque <hainque@adacore.com>
1821 * config/vxworks.c (vxworks_override_options): Pick default
1822 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
1823 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
1824 DWARF_GNAT_ENCODINGS_DEFAULT.
1825 * config/vxworksae.h: Likewise.
1827 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
1829 PR tree-optimization/83104
1830 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
1833 2017-11-22 Marek Polacek <polacek@redhat.com>
1834 H.J. Lu <hongjiu.lu@intel.com>
1835 Jason Merrill <jason@redhat.com>
1840 * calls.c (initialize_argument_information): Call
1841 warn_parameter_passing_abi target hook.
1842 (store_one_arg): Use 0 for empty record size. Don't push 0 size
1843 argument onto stack.
1844 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
1845 * common.opt: Update -fabi-version description.
1846 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
1847 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
1849 (ix86_is_empty_record): New function.
1850 (ix86_warn_parameter_passing_abi): New function.
1851 (TARGET_EMPTY_RECORD_P): Redefine.
1852 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
1853 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
1854 * doc/tm.texi: Regenerated.
1855 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
1856 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
1857 * dwarf2out.c (get_ultimate_context): Move to tree.c.
1858 * explow.c (hard_function_value): Call arg_int_size_in_bytes
1859 instead of int_size_in_bytes.
1860 * expr.c (copy_blkmode_to_reg): Likewise.
1861 * function.c (aggregate_value_p): Return 0 for empty types.
1862 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
1863 (locate_and_pad_parm): Call arg size_in_bytes instead
1865 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
1866 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
1867 * target.def (empty_record_p, warn_parameter_passing_abi): New target
1869 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
1870 (std_gimplify_va_arg_expr): Skip empty records. Call
1871 arg_size_in_bytes instead size_in_bytes.
1872 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
1873 * tree-core.h (tree_type_common): Add empty_flag.
1874 (tree_decl_common): Update comments.
1875 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
1877 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1878 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
1880 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1881 * tree.c (default_is_empty_type): New function.
1882 (default_is_empty_record): New function.
1883 (arg_int_size_in_bytes): New function.
1884 (arg_size_in_bytes): New function.
1885 (get_ultimate_context): New function.
1886 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
1887 TRANSLATION_UNIT_WARN_EMPTY_P.
1888 (default_is_empty_record, arg_int_size_in_bytes,
1889 arg_size_in_bytes, get_ultimate_context): Declare.
1891 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1893 * config/arm/arm.c (cmse_clear_registers): New function.
1894 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
1895 code by call to cmse_clear_registers.
1896 (cmse_nonsecure_entry_clear_before_return): Likewise.
1898 2017-11-22 Tamar Christina <tamar.christina@arm.com>
1900 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
1901 (vdot_s32, vdotq_s32): New.
1902 (vdot_lane_u32, vdotq_lane_u32): New.
1903 (vdot_lane_s32, vdotq_lane_s32): New.
1906 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1909 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
1910 for unsigned values with fewer HWIs than the precision.
1911 (test_overflow): New function.
1912 (wide_int_cc_tests): Call it.
1914 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1916 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
1917 has the same mode class as Pmode.
1919 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1921 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
1922 padding_bits_to_clear_ptr.
1923 (cmse_nonsecure_entry_clear_before_return): Likewise.
1925 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1927 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
1928 auto_sbitap instead of integer bitfield to control register needing
1931 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1933 PR tree-optimization/83044
1934 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
1935 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
1936 that need the upper bound. Subtract offset from
1937 get_addr_base_and_unit_offset only if positive and subtract it
1938 before division by eltsize rather than after it.
1941 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
1942 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
1943 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
1947 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
1948 expand_binop with *mul_widen_optab, make sure at least one of the
1949 operands doesn't have VOIDmode.
1952 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
1954 PR rtl-optimization/82044
1955 PR tree-optimization/82042
1956 * dse.c (record_store): Check for overflow.
1957 (check_mem_read_rtx): Properly check for overflow if width == -1, call
1958 add_wild_read instead of clear_rhs_from_active_local_stores on
1959 overflow and log it into dump_file.
1961 2017-11-22 Richard Biener <rguenther@suse.de>
1963 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
1964 fake edges to exit when looking for a place to insert.
1965 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
1967 (insert_into_preds_of_block): Commit edge insertion immediately,
1968 assert that doesn't require new BBs.
1969 (fini_pre): Release expressions.
1970 (pass_pre::execute): Shuffle things around a bit, if the fn
1971 is too large do not compute AVAIL either as this is really the
1974 2017-11-22 Richard Biener <rguenther@suse.de>
1976 PR tree-optimization/83089
1977 * tree-if-conv.c (pass_if_conversion::execute): If anything
1978 changed reset SCEV and free the number of iteration estimates.
1980 2017-11-21 Martin Sebor <msebor@redhat.com>
1982 PR tree-optimization/82945
1983 * calls.h (warn_nonstring_bound): Remove unused function.
1985 2017-11-21 Martin Sebor <msebor@redhat.com>
1987 PR tree-optimization/82945
1988 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
1989 * calls.h (maybe_warn_nonstring_arg): Declare new function.
1990 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
1992 (initialize_argument_information): Call maybe_warn_nonstring_arg.
1993 * calls.h (get_attr_nonstring_decl): Declare new function.
1994 * doc/extend.texi (attribute nonstring): Update.
1995 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
1996 get_attr_nonstring_decl and handle it.
1997 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
1998 detection of nul-termination.
1999 (strlen_to_stridx): Change to a pointer.
2000 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
2001 (pass_strlen::execute): Same.
2003 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
2005 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
2006 for the new option -mprefer-vector-width=[none|128|256|512].
2007 * config/i386/i386.c (ix86_target_string): remove old style options
2008 -mprefer-avx256 and make -mprefer-avx128 as alias.
2009 (ix86_option_override_internal): Apply defaults for the
2010 -mprefer-vector-width=[128|256] option.
2011 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
2012 Implement macros to work with -mprefer-vector-width=.
2013 * config/i386/i386.opt: Implemented option
2014 -mprefer-vector-width=[none|128|256|512].
2015 * doc/invoke.texi: Documentation for
2016 -mprefer-vector-width=[none|128|256|512].
2018 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
2020 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
2022 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
2024 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
2026 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
2028 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
2030 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
2031 * expr.c (expand_expr_real_2): Likewise.
2032 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
2033 fold_binary_loc): Likewise.
2034 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
2035 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
2036 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
2037 MINUS_EXPR transformations.
2038 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
2039 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
2040 * tree-inline.c (estimate_operator_cost): Likewise.
2041 * tree-pretty-print.c (dump_generic_node, op_code_prio,
2042 op_symbol_code): Likewise.
2043 * tree-vect-stmts.c (vectorizable_operation): Likewise.
2044 * vr-values.c (extract_range_from_binary_expr): Likewise.
2045 * varasm.c (initializer_constant_valid_p_1): Likewise.
2046 * tree.def: New tree code POINTER_DIFF_EXPR.
2048 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
2050 * config/i386/i386.md (*bswap<mode>2_movbe): Add
2051 integer suffix to movbe mnemonic.
2052 (*bswaphi2_movbe): Ditto.
2053 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2055 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2058 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2059 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2060 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2062 2017-11-21 Martin Liska <mliska@suse.cz>
2064 * tree-inline.c (expand_call_inline): Remove not needed
2067 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2068 Jakub Jelinek <jakub@redhat.com>
2071 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2072 Remove static keyword from f variable.
2074 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2076 PR tree-optimization/83086
2077 * gimple-ssa-store-merging.c
2078 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2079 rather than n.base_addr.
2081 2017-11-21 Martin Liska <mliska@suse.cz>
2083 PR rtl-optimization/82044
2084 PR tree-optimization/82042
2085 * dse.c (check_mem_read_rtx): Check for overflow.
2087 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2089 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2092 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2094 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2095 and its corresponding call arg location note.
2097 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2098 Andrew Burgess <andrew.burgess@embecosm.com>
2100 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2102 (arc_return_slot_offset): Likewise.
2103 (arc_eh_return_address_location): New declaration.
2104 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2105 (MUST_SAVE_REGISTER): Add exception handler case.
2106 (MUST_SAVE_RETURN_ADDR): Likewise.
2107 (arc_frame_pointer_required): Likewise.
2108 (arc_frame_pointer_needed): New function.
2109 (arc_compute_frame_size): Changed.
2110 (arc_expand_prologue): Likewise.
2111 (arc_expand_epilogue): Likewise.
2112 (arc_initial_elimination_offset): Likewise.
2113 (arc_return_slot_offset): Delete.
2114 (arc_eh_return_address_location): New function.
2115 (arc_builtin_setjmp_frame_value): Likewise.
2116 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
2117 (EH_RETURN_STACKADJ_RTX): Define.
2118 (EH_RETURN_HANDLER_RTX): Likewise.
2119 * config/arc/arc.md (eh_return): Delete.
2121 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2123 * print-tree.h (debug_vec_tree): Remove prototype.
2124 * gdbinit.in (pvt): Remove macro.
2126 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2128 PR tree-optimization/83047
2129 * gimple-ssa-store-merging.c
2130 (imm_store_chain_info::output_merged_store): If the loads with the
2131 same vuse are in different basic blocks, for load_gsi pick a load
2132 location that is dominated by the other loads.
2135 * config/i386/i386.c (ix86_memmodel_check): Start
2136 -Winvalid-memory-model diagnostics with lowercase letter.
2139 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
2140 set in_cold_section_p to the partition containing loc_list->first.
2141 When seeing loc_list->last_before_switch node, update secname and
2142 perform range_across_switch second partition handling only after that.
2145 * run-rtl-passes.c: Include debug.h.
2146 (run_rtl_passes): Call debug_hooks->assembly_start.
2147 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
2151 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
2152 OPTAB_DIRECT in calls to expand_simple_binop.
2154 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2157 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
2159 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2162 * spellcheck.h (best_match::blithely_get_best_candidate): New
2165 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2167 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
2168 with lower case letter.
2170 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
2172 * config/i386/i386.md (bswaphi2): New expander.
2173 (*bswaphi2_movbe): New insn pattern.
2174 (bswaphi -> rorhi pepehole2): New peephole pattern.
2176 2017-11-20 Jeff Law <law@redhat.com>
2178 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
2179 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
2180 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
2181 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
2182 gimple-ssa-evrp-analyze.h.
2184 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
2185 set BB_VISITED here.
2186 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
2188 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2190 * tree-predcom.c: Add general comment on Store-Store chains.
2191 (split_data_refs_to_components): Postpone clearing eliminate_store_p
2193 (get_chain_last_ref_at): Rename into...
2194 (get_chain_last_write_at): ...this.
2195 (get_chain_last_write_before_load): New function.
2196 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
2197 CT_STORE_STORE when write reference is added.
2198 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
2199 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
2200 (initialize_root_vars_store_elim_1): Ditto.
2201 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
2202 definition is created.
2203 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
2204 chain by replacing it with dominant stored value.
2206 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2208 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
2210 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
2212 * vr-values.c (extract_range_from_binary_expr): Use a full range
2215 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
2217 * config/arm/arm.md (R4_REGNUM): Define constant.
2218 (nonsecure_call_internal): Remove r4 clobber.
2219 (nonsecure_call_value_internal): Likewise.
2220 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
2221 clobber and resequence match_operands.
2222 (nonsecure_call_value_reg_thumb1_v5): Likewise.
2223 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
2224 (nonsecure_call_value_reg_thumb2): Likewise.
2226 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2228 PR tree-optimization/78821
2229 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
2230 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
2231 to the address of the base rather than the base itself.
2232 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
2233 (find_bswap_or_nop_finalize): New function.
2234 (find_bswap_or_nop): Use it.
2235 (bswap_replace): Return a tree rather than bool, change first
2236 argument from gimple * to gimple_stmt_iterator, allow inserting
2237 into an empty sequence, allow ins_stmt to be NULL - then emit
2238 all stmts into gsi. Fix up MEM_REF address gimplification.
2239 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
2240 (struct store_immediate_info): Add N and INS_STMT non-static
2242 (store_immediate_info::store_immediate_info): Initialize them
2243 from newly added ctor args.
2244 (merged_store_group::apply_stores): Formatting fixes. Sort by
2246 (stmts_may_clobber_ref_p): For stores call also
2247 refs_anti_dependent_p.
2248 (gather_bswap_load_refs): New function.
2249 (imm_store_chain_info::try_coalesce_bswap): New method.
2250 (imm_store_chain_info::coalesce_immediate_stores): Use it.
2251 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
2252 (imm_store_chain_info::output_merged_store): Fail if number of
2253 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
2254 and NOP_EXPR rhs_code.
2255 (pass_store_merging::process_store): Compute n and ins_stmt, if
2256 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
2257 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
2259 (pass_store_merging::execute): Calculate dominators.
2261 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
2262 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
2263 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
2264 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2265 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
2266 class pass_optimize_bswap, bswap_replace,
2267 pass_optimize_bswap::execute): Moved to ...
2268 * gimple-ssa-store-merging.c: ... this file.
2269 Include optabs-tree.h.
2270 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
2271 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2272 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
2273 anonymous namespace, remove static keywords.
2274 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
2275 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
2277 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
2280 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
2282 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
2284 * vec.h (debug_helper): New function.
2285 (DEFINE_DEBUG_VEC): New macro.
2286 * hash-set.h (debug_helper): New function.
2287 (DEFINE_DEBUG_HASH_SET): New macro.
2288 * cfg.c (debug_slim (edge)): New function.
2289 Call DEFINE_DEBUG_VEC for edges.
2290 Call DEFINE_DEBUG_HASH_SET for edges.
2291 * cfghooks.c (debug_slim (basic_block)): New function.
2292 Call DEFINE_DEBUG_VEC for basic blocks.
2293 Call DEFINE_DEBUG_HASH_SET for basic blocks.
2294 * print-tree.c (debug_slim): New function to handle trees.
2295 Call DEFINE_DEBUG_VEC for trees.
2296 Call DEFINE_DEBUG_HASH_SET for trees.
2297 (debug (vec<tree, va_gc>) &): Remove.
2298 (debug (<vec<tree, va_gc>) *): Remove.
2299 * print-rtl.c (debug_slim): New function to handle const_rtx.
2300 Call DEFINE_DEBUG_VEC for rtx_def.
2301 Call DEFINE_DEBUG_VEC for rtx_insn.
2302 Call DEFINE_DEBUG_HASH_SET for rtx_def.
2303 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
2304 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
2305 (debug (vec<rtx_insn *> *ptr): Remove.
2306 (debug_insn_vector): Remove.
2307 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
2309 2017-11-20 Tom de Vries <tom@codesourcery.com>
2311 PR rtl-optimization/82020
2312 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
2313 IF_THEN_ELSE condition.
2315 2017-11-19 Jeff Law <law@redhat.com>
2317 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2318 of degenerates resulting from ignoring an edge.
2320 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2323 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
2325 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2328 * profile-count.c (profile_count::to_sreal_scale): Fix return value
2329 for uninitialied counts.
2331 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2334 * tree-inline.c (estimate_num_insns): Set to 1 at least.
2336 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2339 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
2341 2017-11-19 Tom de Vries <tom@codesourcery.com>
2343 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
2346 2017-11-19 Tom de Vries <tom@codesourcery.com>
2348 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
2351 2017-11-19 Tom de Vries <tom@codesourcery.com>
2353 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2356 2017-11-19 Tom de Vries <tom@codesourcery.com>
2358 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
2360 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
2362 2017-11-19 Tom de Vries <tom@codesourcery.com>
2364 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
2366 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
2367 semicolon after MCORE_EXPORT_NAME call.
2369 2017-11-19 Tom de Vries <tom@codesourcery.com>
2372 * vmsdbgout.c (vmsdbgout_early_finish): New function.
2373 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
2375 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
2377 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
2378 (cgraph_edge::clone): Cleanup updating of profile.
2379 * ipa-cp.c (update_profiling_info): Likewise.
2380 * ipa-inline-transform.c (inline_transform): Likewise.
2381 * ipa-inline.c (inline_small_functions): Add missing space to dump.
2382 * ipa-split.c (execute_split_functions): Do not split when function
2384 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
2385 * profile-count.c (profile_count::dump): Add global0.
2386 (profile_count::to_cgraph_frequency): Do not ICE when entry is
2388 (profile_count::to_sreal_scale): Likewise.
2389 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
2390 (profile_count::combine_with_ipa_count): New function.
2391 * profile-count.h (profile_guessed_global0adjusted): New.
2392 (profile_count::adjusted_zero): New.
2393 (profile_count::global0adjusted): New.
2394 (profile_count::combine_with_ipa_count): New.
2395 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
2396 correct profile of return block of split functions.
2397 (copy_cfg_body): Remove unused profile_count.
2398 (copy_body): Likewise.
2399 (expand_call_inline): Update.
2400 (tree_function_versioning): Update.
2402 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
2404 * hash-set.h (hash_set::empty): New.
2405 * tree-ssa-threadbackward.h: Delete.
2406 * tree-ssa-threadbackward.c (class thread_jumps): New.
2407 Move max_threaded_paths into class.
2408 (fsm_find_thread_path): Remove arguments that are now in class.
2409 (profitable_jump_thread_path): Rename to...
2410 (thread_jumps::profitable_jump_thread_path): ...this.
2411 (convert_and_register_jump_thread_path): Rename to...
2412 (thread_jumps::convert_and_register_current_path): ...this.
2413 (check_subpath_and_update_thread_path): Rename to...
2414 (thread_jumps::check_subpath_and_update_thread_path): ...this.
2415 (register_jump_thread_path_if_profitable): Rename to...
2416 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
2417 (handle_phi): Rename to...
2418 (thread_jumps::handle_phi): ...this.
2419 (handle_assignment): Rename to...
2420 (thread_jumps::handle_assignment): ...this.
2421 (fsm_find_control_statement_thread_paths): Rename to...
2422 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
2423 (find_jump_threads_backwards): Rename to...
2424 (thread_jumps::find_jump_threads_backwards): ...this.
2425 Initialize path local data.
2426 (pass_thread_jumps::execute): Call find_jump_threads_backwards
2427 from within thread_jumps class.
2428 (pass_early_thread_jumps::execute): Same.
2430 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2432 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
2434 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2436 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
2437 to canonical location.
2439 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2442 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
2444 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
2446 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
2448 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
2449 rcrt1.o%s/grcrt1.o%s for -static-pie.
2451 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2453 * i386.c (ix86_multiplication_cost, ix86_division_cost,
2454 ix86_shift_rotate_cost): Break out from ...
2455 (ix86_rtx_costs): ... here.
2456 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
2459 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2461 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
2462 when entry block was promoted unlikely.
2463 (estimate_bb_frequencies): Increase frequency scale.
2464 * profile-count.h (profile_count): Export precision info.
2466 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2468 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
2469 disturbin profile of entry block.
2471 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2473 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
2476 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2478 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
2479 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
2481 2017-11-17 Jeff Law <law@redhat.com>
2483 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
2484 from evrp_dom_walker class. Various methods moved into new class.
2485 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
2486 (evrp_range_analyzer::enter): New method.
2487 (evrp_range_analyzer::leave): New method.
2488 (evrp_dom_walker): Remove delegators no longer needed by this class.
2489 Replace vr_values data member with evrp_range_analyzer
2491 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
2492 method extracted from evrp_dom_walker::before_dom_children.
2493 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
2494 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
2496 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
2497 Add private copy constructor and move assignment operators.
2498 Privatize methods and class data where trivially possible.
2499 (evrp_dom_walker::cleanup): New function, extracted from
2500 execute_early_vrp. Simplify access to class data.
2502 * vr-values.h (get_output_for_vrp): Prototype.
2503 * vr-values.c (get_output_for_vrp): New function extracted from
2504 vrp_visit_assignment_or_call and extract_range_from_stmt.
2505 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
2507 2017-11-17 Luis Machado <luis.machado@linaro.org>
2509 * config/aarch64/aarch64.c
2510 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
2511 (qdf24xx_tunings) <autoprefetcher_model>: Set to
2512 tune_params::AUTOPREFETCHER_WEAK.
2514 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2517 * config/arm/arm.c (arm_valid_target_attribute_rec):
2518 Parse "arch=" and "+<ext>".
2519 (arm_valid_target_attribute_tree): Re-init global options.
2520 (arm_option_override): Make non-static.
2521 (arm_options_perform_arch_sanity_checks): Make errors fatal.
2522 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
2523 (__ARM_FEATURE_CRC32): Support undef.
2524 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
2525 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
2527 2017-11-17 David Malcolm <dmalcolm@redhat.com>
2529 * gdbinit.in (break-on-diagnostic): New command.
2531 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2533 * config/i386/i386.c (ix86_expand_epilogue): Change simple
2534 return to indirect jump for EH return if control-flow
2535 protection is enabled. Change explicit 'false' argument in
2536 pro_epilogue_adjust_stack with a value of
2538 * config/i386/i386.md (simple_return_indirect_internal):
2539 Remove SImode restriction to support 64-bit.
2541 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2543 * combine.c (added_notes_insn): New.
2544 (try_combine): Handle added_notes_insn like added_links_insn.
2545 Rewrite return value code.
2546 (distribute_notes): Set added_notes_insn to the earliest insn we added
2549 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2551 PR rtl-optimization/82621
2552 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
2553 dest of one of those SETs is unused.
2555 2017-11-17 Richard Biener <rguenther@suse.de>
2558 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
2559 * tree-pretty-print.c (dump_generic_node): Handle
2560 annot_expr_parallel_kind.
2561 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
2562 * gimplify.c (gimple_boolify): Likewise.
2564 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2566 * config.gcc (extra_headers): Add cet.h for x86 targets.
2567 * config/i386/cet.h: New file.
2568 * doc/install.texi: Add --enable-cet/--disable-cet.
2570 2017-11-17 Richard Biener <rguenther@suse.de>
2572 PR tree-optimization/83017
2573 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
2574 (gen_parallel_loop): Properly count iterations.
2575 (parallelize_loops): Handle loop->can_be_parallel independent
2576 of flag_loop_parallelize_all. Make static profitability test match
2578 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
2579 * invoke.texi (parloops-min-per-thread): Document.
2581 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
2583 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
2584 upstreaming review comments.
2586 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2588 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
2589 with fast unaligned access.
2590 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
2592 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
2594 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
2595 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
2597 2017-11-17 Richard Biener <rguenther@suse.de>
2599 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
2600 folding of references.
2602 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
2605 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
2606 of replacing call to strncmp with corresponding call to strcmp when
2609 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
2611 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
2612 option prefer-avx256 for skylake-avx512 configuration.
2613 * config/i386/i386.c (ix86_option_override_internal): Ditto.
2614 (get_builtin_code_for_version): Ditto.
2616 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2617 Monk Chiang <sh.chiang04@gmail.com>
2619 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
2620 (FIXED_REGISTERS): Reserve more register numbers.
2621 (CALL_USED_REGISTERS): Likewise.
2622 (REG_ALLOC_ORDER): Likewise.
2623 (REG_CLASS_CONTENTS): Likewise.
2624 (REGISTER_NAMES): Likewise.
2626 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2627 Kito Cheng <kito.cheng@gmail.com>
2629 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
2631 * config/nds32/iterators.md: Add vector mode iterators and attributes.
2633 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
2635 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
2636 parameter list for vec_splats.
2638 2017-11-16 Joseph Myers <joseph@codesourcery.com>
2640 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
2642 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
2644 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
2647 * doc/install.texi: Document incompatibility of Perl >=5.6.26
2648 with the required version of automake 1.11.6.
2650 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
2652 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
2654 (DU_C2_3_power9): Correct reservation combinations.
2655 (FP_DIV_power9, VEC_DIV_power9): New.
2656 (power9-alu): Split out rotate/shift...
2657 (power9-rot): ...to here, correct dispatch resource.
2658 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
2660 (power9-fp): Correct latency.
2661 (power9-sdiv): Add div/sqrt resource.
2662 (power9-ddiv): Correct latency, add div/sqrt resource.
2663 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
2664 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
2666 (power9-qpdiv, power9-qpmul): Adjust resource usage.
2668 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2670 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
2671 switch statement mapping KF built-ins to TF built-ins if we don't
2672 have the proper ISA 3.0 assembler support.
2674 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2676 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
2677 (lower_emutls_function_body): Do not compute it.
2679 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2681 * ipa-split.c (split_bb_info): Turn time to sreal.
2682 (split_point): Likewise.
2683 (dump_split_point): Likewise.
2684 (fine_split_points): Likewise.
2685 (execute_split_functions): Only zero split_bbs; turn time to sreals.
2687 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2689 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
2692 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2694 * predict.c (combine_predictions_for_bb): Preserve zero predicted
2696 (expensive_function_p): Remove useless assert.
2697 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
2699 2017-11-16 Martin Sebor <msebor@redhat.com>
2701 PR tree-optimization/82588
2702 PR tree-optimization/82583
2703 * tree-vrp.c (check_array_ref): Handle flexible array members,
2704 string literals, and inner indices.
2705 (search_for_addr_array): Add detail to diagnostics.
2707 2017-11-16 Nathan Sidwell <nathan@acm.org>
2711 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
2712 (SET_DECL_ASSEMBLER_NAME): Forward to
2713 overwrite_decl_assembler_name.
2714 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
2715 (overwrite_decl_assembler_name): Declare.
2716 * tree.c (overwrite_decl_assembler_name): New.
2717 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
2718 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
2719 (LANG_HOOKS_INITIALIZER): Add it.
2720 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
2721 * langhooks.c (lhd_set_decl_assembler_name): Use
2722 SET_DECL_ASSEMBLER_NAME.
2723 (lhd_overwrite_decl_assembler_name): Default implementation.
2725 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2726 Jackson Woodruff <jackson.woodruff@arm.com>
2728 PR tree-optimization/71026
2729 * match.pd: Canonicalize constant multiplies in division.
2731 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2733 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
2734 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
2735 Remove OPT_fomit_frame_pointer entry.
2736 * common/config/arc/arc-common.c: Likewise.
2737 * common/config/arm/arm-common.c: Likewise.
2738 * common/config/avr/avr-common.c: Likewise.
2739 * common/config/c6x/c6x-common.c: Likewise.
2740 * common/config/cr16/cr16-common.c: Likewise.
2741 * common/config/cris/cris-common.c: Likewise.
2742 * common/config/epiphany/epiphany-common.c: Likewise.
2743 * common/config/fr30/fr30-common.c: Likewise.
2744 * common/config/frv/frv-common.c: Likewise.
2745 * common/config/ia64/ia64-common.c: Likewise.
2746 * common/config/iq2000/iq2000-common.c: Likewise.
2747 * common/config/lm32/lm32-common.c: Likewise.
2748 * common/config/m32r/m32r-common.c: Likewise.
2749 * common/config/mcore/mcore-common.c: Likewise.
2750 * common/config/microblaze/microblaze-common.c: Likewise.
2751 * common/config/mips/mips-common.c: Likewise.
2752 * common/config/mmix/mmix-common.c: Likewise.
2753 * common/config/mn10300/mn10300-common.c: Likewise.
2754 * common/config/nios2/nios2-common.c: Likewise.
2755 * common/config/pa/pa-common.c: Likewise.
2756 * common/config/pdp11/pdp11-common.c: Likewise.
2757 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
2758 * common/config/riscv/riscv-common.c: Likewise.
2759 * common/config/rs6000/rs6000-common.c: Likewise.
2760 * common/config/rx/rx-common.c: Likewise.
2761 * common/config/s390/s390-common.c: Likewise.
2762 * common/config/sh/sh-common.c: Likewise.
2763 * common/config/sparc/sparc-common.c: Likewise.
2764 * common/config/tilegx/tilegx-common.c: Likewise.
2765 * common/config/tilepro/tilepro-common.c: Likewise.
2766 * common/config/v850/v850-common.c: Likewise.
2767 * common/config/visium/visium-common.c: Likewise.
2768 * common/config/xstormy16/xstormy16-common.c: Likewise.
2769 * common/config/xtensa/xtensa-common.c: Likewise.
2770 * invoke.texi (-fomit-frame-pointer): Update documentation.
2772 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2774 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
2776 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2778 * cfg.c (scale_bbs_frequencies_int,
2779 cale_bbs_frequencies_gcov_type): Remove.
2780 * cfg.h (scale_bbs_frequencies_int,
2781 cale_bbs_frequencies_gcov_type): Remove.
2783 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2785 * tree-ssa-loop-manip.c
2786 (scale_dominated_blocks_in_loop): Update to profile counts.
2787 (tree_transform_and_unroll_loop): Likewise.
2789 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2791 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
2792 scale_bbs_frequencies_int.
2794 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2796 * final.c (compute_alignments): Use counts rather than frequencies.
2798 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2800 * cfgloopanal.c: Include sreal.h
2801 (average_num_loop_insns): Use counts and sreal for accounting.
2803 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2805 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
2808 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2809 Kito Cheng <kito.cheng@gmail.com>
2811 * config/nds32/constraints.md: Provide more constraints.
2812 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
2813 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
2814 support constraints usage.
2816 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2818 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
2820 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2821 Kito Cheng <kito.cheng@gmail.com>
2823 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
2824 * config/nds32/nds32.opt: Refine the layout.
2825 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
2826 TARGET_EXT_STRING): Support new options.
2827 * config/nds32/nds32.h: Likewise.
2828 * config/nds32/nds32.md: Likewise.
2829 * config/nds32/nds32-predicates.c: Likewise.
2830 * config/nds32/constraints.md: Likewise.
2831 * common/config/nds32/nds32-common.c: Likewise.
2833 2017-11-16 Julia Koval <julia.koval@intel.com>
2836 * config/i386/gfniintrin.h: Add sse check.
2837 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
2839 2017-11-16 Julia Koval <julia.koval@intel.com>
2841 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
2842 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
2843 (ix86_handle_option): Handle -mavx512vbmi2.
2844 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
2845 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2846 * config/i386/i386-c.c (__AVX512VBMI2__): New.
2847 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
2848 (ix86_valid_target_attribute_inner_p): Ditto.
2849 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
2850 * config/i386/i386.opt (mavx512vbmi2): New option.
2851 * doc/invoke.texi: Add new option.
2853 2017-11-16 Julia Koval <julia.koval@intel.com>
2855 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
2856 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
2857 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
2858 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
2859 _mm512_gf2p8mul_epi8): New intrinsics.
2860 * config/i386/i386-builtin-types.def
2861 (V64QI_FTYPE_V64QI_V64QI): New type.
2862 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
2863 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
2864 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
2865 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
2866 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
2867 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2869 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
2871 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
2872 explicitly as a stream of bytes.
2874 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2876 * config/rs6000/altivec.h (vec_xst_be): New #define.
2877 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
2878 and externalize from *altivec_vperm_<mode>_internal.
2879 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
2881 (XL_BE_V8HI): Likewise.
2882 (XL_BE_V4SI): Likewise.
2883 (XL_BE_V4SI): Likewise.
2884 (XL_BE_V2DI): Likewise.
2885 (XL_BE_V4SF): Likewise.
2886 (XL_BE_V2DF): Likewise.
2887 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
2888 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
2889 all array entries with these keys: VSX_BUILTIN_VEC_XL,
2890 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
2891 VSX_BUILTIN_VEC_XST_BE.
2892 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
2893 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
2895 (altivec_init_builtins): Replace conditional calls to def_builtin
2896 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
2897 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
2898 with unconditional calls. Remove calls to def_builtin for
2899 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
2900 __builtin_vec_xst_be.
2901 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
2902 to define_expand, and add alternate RTL generation for P8.
2903 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
2904 vsx_ld_elemrev_v8hi.
2905 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
2906 add alternate RTL generation for P8.
2907 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
2908 vsx_ld_elemrev_v16qi.
2909 (vsx_st_elemrev_v8hi): Convert define_insn
2910 to define_expand, and add alternate RTL generation for P8.
2911 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
2912 vsx_st_elemrev_v8hi.
2913 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
2914 add alternate RTL generation for P8.
2915 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
2916 vsx_st_elemrev_v16qi.
2918 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
2921 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
2922 TARGET_AVX512ER check.
2923 (ix86_option_override_internal): Set MASK_VZEROUPPER if
2924 neither -mzeroupper nor -mno-zeroupper is used and
2925 TARGET_EMIT_VZEROUPPER is set.
2926 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
2927 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
2929 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2931 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2932 folding of vector compares.
2933 (fold_build_vec_cmp): New helper function.
2934 (fold_compare_helper): New helper function.
2935 (builtin_function_type): Add compare builtins to the list of functions
2936 having unsigned arguments. Cosmetic updates to comment indentation.
2937 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
2938 the not+eq combination.
2940 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2942 PR tree-optimization/82726
2943 PR tree-optimization/70754
2944 * tree-predcom.c (order_drefs_by_pos): New function.
2945 (combine_chains): Move code setting has_max_use_after to...
2946 (try_combine_chains): ...here. New parameter. Sort combined chains
2947 according to position information.
2948 (tree_predictive_commoning_loop): Update call to above function.
2949 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
2951 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2953 PR tree-optimization/82726
2955 2017-01-23 Bin Cheng <bin.cheng@arm.com>
2957 PR tree-optimization/70754
2958 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
2959 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
2960 combined stmt before it if not NULL.
2961 (combine_chains): Process refs reversely and compute dominance point
2965 2017-02-23 Bin Cheng <bin.cheng@arm.com>
2967 PR tree-optimization/79663
2968 * tree-predcom.c (combine_chains): Process refs in reverse order
2969 only for ZERO length chains, and add explaining comment.
2971 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2973 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
2974 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
2975 * doc/invoke.texi (ARM Options): Add armv8.3-a.
2977 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2979 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
2981 2017-11-15 Martin Liska <mliska@suse.cz>
2983 * tree-cfg.c (pass_warn_function_return::execute):
2984 Compare warn_return_type for greater than zero.
2986 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
2990 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
2991 to return true on Xeon and not on Xeon Phi.
2992 (ix86_check_avx256_register): Changed to ...
2993 (ix86_check_avx_upper_register): ... this. Add extra check for
2994 VALID_AVX512F_REG_OR_XI_MODE.
2995 (ix86_avx_u128_mode_needed): Changed
2996 ix86_check_avx256_register to ix86_check_avx_upper_register.
2997 (ix86_check_avx256_stores): Changed to ...
2998 (ix86_check_avx_upper_stores): ... this. Changed
2999 ix86_check_avx256_register to ix86_check_avx_upper_register.
3000 (ix86_avx_u128_mode_after): Changed
3001 avx_reg256_found to avx_upper_reg_found. Changed
3002 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
3003 (ix86_avx_u128_mode_entry): Changed
3004 ix86_check_avx256_register to ix86_check_avx_upper_register.
3005 (ix86_avx_u128_mode_exit): Ditto.
3006 * config/i386/i386.h: (host_detect_local_cpu): New define.
3008 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3010 * config/arm/xgene1.md (xgene1): Split into automatons
3011 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
3012 (xgene1_f_load): Adjust reservations and/or types.
3013 (xgene1_f_store): Likewise.
3014 (xgene1_load_pair): Likewise.
3015 (xgene1_store_pair): Likewise.
3016 (xgene1_fp_load1): Likewise.
3017 (xgene1_load1): Likewise.
3018 (xgene1_store1): Likewise.
3019 (xgene1_move): Likewise.
3020 (xgene1_alu): Likewise.
3021 (xgene1_simd): Likewise.
3022 (xgene1_bfm): Likewise.
3023 (xgene1_neon_load1): Likewise.
3024 (xgene1_neon_store1): Likewise.
3025 (xgene1_neon_logic): Likewise.
3026 (xgene1_neon_st1): Likewise.
3027 (xgene1_neon_ld1r): Likewise.
3028 (xgene1_alu_cond): Added.
3029 (xgene1_shift_reg): Likwise.
3030 (xgene1_bfx): Likewise.
3031 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
3033 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3036 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
3038 (can_widen_mult_without_libcall): New function.
3039 (expand_mul_overflow): If only checking unsigned mul overflow,
3040 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
3041 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
3042 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
3043 (expand_DIVMOD): Formatting fix.
3044 * expmed.h (expand_mult): Add NO_LIBCALL argument.
3045 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
3046 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
3048 PR tree-optimization/82977
3049 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
3050 constructed temporary to strlen_to_stridx.put.
3052 2017-11-15 Martin Liska <mliska@suse.cz>
3054 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3055 * configure: Regenerate.
3057 2017-11-15 Martin Liska <mliska@suse.cz>
3060 * config/sh/sh-mem.cc: Use proper probability for
3063 2017-11-14 Jeff Law <law@redhat.com>
3065 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3066 the red zone for stack_clash_protection_final_dynamic_probe targets
3067 when the total dynamic stack size is zero bytes.
3069 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3070 blocks is post order.
3072 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3074 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3075 * final.c (rest_of_clean_state): Set it for the
3076 -fcompare-debug dump.
3077 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3078 class when TDF_COMPARE_DEBUG is set.
3080 * dwarf2out.c (gen_producer_string): Discard
3083 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3086 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3087 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3088 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3089 (__TGMATH_REAL_2_3): Remove macros.
3090 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3091 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3093 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3095 (remquo): Define using __TGMATH_REAL_3.
3097 2017-11-14 Jeff Law <law@redhat.com>
3099 * vr-values.c: New file with contents extracted from tree-vrp.c.
3100 * Makefile.in (OBJS): Add vr-values.o
3101 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3102 (set_value_range, set_and_canonicalize_value_range): Likewise.
3103 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3104 (value_range_constant_singleton, symbolic_range_p): Likewise.
3105 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3106 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3107 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3108 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3109 (find_case_label_range, find_case_label_index): Likewise.
3110 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3111 (range_int_cst_singleton_p, value_inside_range): Likewise.
3112 (get_single_symbol): Likewise.
3113 (switch_update): Move structure definition here.
3114 (to_remove_edges, to_update_switch_stmts): Provide externs.
3115 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
3116 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
3117 (vrp_val_is_min, set_value_range): Likewise.
3118 (set_and_canonicalize_value_range, copy_value_range): Likewise.
3119 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
3120 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
3121 (range_is_nonnull, range_int_cst_p): Likewwise.
3122 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
3123 (get_single_symbol, operand_less_p): Likewise
3124 (compare_values_warnv, compare_values): Likewise.
3125 (value_inside_range, value_range_constant_singleton): Likewise.
3126 (zero_nonzero_bitgs_from_vr): Likewise.
3127 (extract_range_from_binary_expr_1): Likewise.
3128 (overflow_comparison_p): Likewise.
3129 (to_remove_edges, to_update_switch_stmts): Likewise.
3130 (find_case_label-index, find_case_label_range): Likewise.
3131 (switch_update, set_value_range_to_nonnegative): Remove.
3132 (set_value_range_to_truthvalue): Likewise.
3133 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
3134 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
3135 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
3136 (find_case_label_ranges, test_for_singularity): Likewise.
3137 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
3138 (x_vr_values): Move to its remaining use site.
3140 2017-11-10 Jeff Law <law@redhat.com>
3142 * vr-values.h (VR_INITIALIZER): Move #define here.
3143 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
3144 * Makefile.in (OBJS): Add tree-evrp.o
3145 * tree-vrp.h (assert_info): Move structure definition here.
3146 (set_value_range_to_varying): Prototype.
3147 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3148 (infer_value_range, register_edge_assert_for): Likewise.
3149 (stmt_interesting_for_vrp): Likewise.
3150 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
3151 (set_value_range_to_varying): No longer static.
3152 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3153 (infer_value_range, register_edge_assert_for): Likewise.
3155 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3157 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
3158 generate the XXBRD instruction.
3160 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
3161 (rs6000_builtin_type_compatible): Treat _Float128 and long double
3162 as being compatible if -mabi=ieeelongdouble.
3163 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
3164 to setup float128 built-ins with hardware support.
3165 (BU_FLOAT128_HW_2): Likewise.
3166 (BU_FLOAT128_HW_3): Likewise.
3167 (BU_FLOAT128_HW_VSX_1): Likewise.
3168 (BU_FLOAT128_HW_VSX_2): Likewise.
3169 (scalar_extract_expq): Change float128 built-in functions to
3170 accommodate having both KFmode and TFmode functions. Use the
3171 KFmode variant as the default.
3172 (scalar_extract_sigq): Likewise.
3173 (scalar_test_neg_qp): Likewise.
3174 (scalar_insert_exp_q): Likewise.
3175 (scalar_insert_exp_qp): Likewise.
3176 (scalar_test_data_class_qp): Likewise.
3177 (sqrtf128_round_to_odd): Delete processing the round to odd
3178 built-in functions as special built-in functions, and define them
3179 as float128 built-ins. Use the KFmode variant as the default.
3180 (truncf128_round_to_odd): Likewise.
3181 (addf128_round_to_odd): Likewise.
3182 (subf128_round_to_odd): Likewise.
3183 (mulf128_round_to_odd): Likewise.
3184 (divf128_round_to_odd): Likewise.
3185 (fmaf128_round_to_odd): Likewise.
3186 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
3187 support for KFmode and TFmode xststdcqp calls.
3188 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
3189 point, switch the built-in handlers for the get/set float128
3190 exponent, get float128 mantissa, float128 test built-ins, and the
3191 float128 round to odd built-in functions. Eliminate creating the
3192 float128 round to odd built-in functions as special built-ins.
3193 (rs6000_init_builtins): Eliminate special creation of the float128
3194 round to odd built-in functions.
3195 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
3196 function insns to support both TFmode and KFmode variants.
3197 (xsxsigqp_<mode>): Likewise.
3198 (xsiexpqpf_<mode>): Likewise.
3199 (xsiexpqp_<mode>): Likewise.
3200 (xststdcqp_<mode>): Likewise.
3201 (xststdcnegqp_<mode>): Likewise.
3202 (xststdcqp_<mode>): Likewise.
3204 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3206 * tree-ssa-threadupdate.c (compute_path_counts): Remove
3207 unused path_in_freq_ptr parameter.
3208 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
3210 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3212 * ipa-inline.c (edge_badness): Dump sreal frequency.
3213 (compute_inlined_call_time): Match natural implementaiton ...
3214 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
3215 forgotten division by CGRAPH_FREQ_BASE.
3217 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3219 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
3220 Solaris 11. Update comment.
3221 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
3223 * config/sol2.h (STARTFILE_SPEC): Likewise.
3224 * configure: Regenerate.
3226 2017-11-14 Carl Love <cel@us.ibm.com>
3228 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
3229 le_ and be_ prefixes to swap* variables. Remove
3230 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
3233 2017-11-14 Jason Merrill <jason@redhat.com>
3235 Support GTY((cache)) on hash_map.
3236 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
3237 (ggc_cache_remove): Override it instead of ggc_mx.
3238 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
3239 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
3240 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
3241 (simple_cache_map_traits): Override maybe_mx.
3242 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
3243 (hash_map): Friend gt_cleare_cache.
3244 (gt_cleare_cache): New.
3245 * tree.h (tree_cache_traits): New hash_map traits class.
3246 (tree_cache_map): New typedef.
3248 2017-11-14 Richard Biener <rguenther@suse.de>
3250 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
3251 paramter and handling.
3252 (cleanup_control_flow_bb): Likewise.
3253 (cleanup_control_flow_pre): New helper performing a DFS walk
3254 to call cleanup_control_flow_bb in PRE order.
3255 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
3256 via cleanup_control_flow_pre.
3258 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
3260 * config/aarch64/aarch64-simd.md
3261 (aarch64_simd_bsl<mode>_internal): Remove DImode.
3262 (*aarch64_simd_bsl<mode>_alt): Likewise.
3263 (aarch64_simd_bsldi_internal): New.
3264 (aarch64_simd_bsldi_alt): Likewise.
3266 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3268 * tracer.c (better_p): Do not compare frequencies.
3269 * reg-stack.c (better_edge): Likewise.
3270 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
3273 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3275 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
3276 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
3277 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
3278 * lto-streamer-in.c (input_function): Use update_max_bb_count.
3279 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
3280 * predict.c (maybe_hot_frequency_p): Inline to ...
3281 (maybe_hot_count_p): ... here; rewrite to counts.
3282 (counts_to_freqs): Rename to ...
3283 (update_max_bb_count): ... this one.
3284 (expensive_function_p): Use counts.
3285 (estimate_bb_frequencies): Update.
3286 (rebuild_frequencies): Update.
3287 * predict.h (counts_to_freqs): Rename to ...
3288 (update_max_bb_count): ... this one.
3289 * profile.c (compute_branch_probabilities): Add debug info
3290 * tree-inline.c (expand_call_inline): Update debug info.
3291 (optimize_inline_calls): Use update_max_bb_count..
3292 (tree_function_versioning): Use update_max_bb_count..
3293 * value-prof.c (gimple_value_profile_transformations):
3294 Do not use update_max_bb_count.
3296 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3298 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
3299 always use frequencies.
3301 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3303 * bb-reorder.c: Remove frequencies from comments.
3304 (better_edge_p): Use profile counts.
3305 (find_traces): Dump profile counts.
3306 (rotate_loop): Use profile counts.
3307 (find_traces_1_round): Likewise.
3308 (connect_better_edge_p): Use counts instead of probabilities for
3310 (copy_bb_p): Drop early check for non-0 frequency.
3311 (sanitize_hot_paths): Update comments.
3313 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3315 * ipa-split.c (struct split_point): Add count.
3316 (consider_split): Do not compute incoming frequency; compute incoming
3317 count and store it to split_point.
3318 (split_function): Set count of the call to split part correctly.
3320 2017-11-13 Carl Love <cel@us.ibm.com>
3322 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
3324 2017-11-13 Tom Tromey <tom@tromey.com>
3326 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
3328 2017-11-13 Carl Love <cel@us.ibm.com>
3330 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3331 Add support for builtins:
3332 unsigned int vec_first_{,miss}_match_{,or_eos}index,
3333 vector {un,}signed {char,int,short},
3334 vector {un,}signed {char,int,short}) arguments.
3335 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
3336 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
3337 Add BU_P9V_AV_2 expansions for the builtins.
3338 * config/rs6000/altivec.h (vec_first_match_index,
3339 vec_first_mismatch_index, vec_first_match_or_eos_index,
3340 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
3341 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
3342 new extern declaration.
3343 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
3344 * config/rs6000/vsx.md (first_match_index_<mode>,
3345 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
3346 first_mismatch_or_eos_index_<mode>): Add define expand.
3347 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
3348 * doc/extend.texi: Update the built-in documenation file for the new
3351 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
3353 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
3354 into the min/max operations for _Float<N> and _Float<N>X types.
3356 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3359 * dwarf2out.c (do_eh_frame): New static variable.
3360 (dwarf2out_begin_prologue): Set it.
3361 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
3363 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3365 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
3367 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
3369 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3372 * debug.h (dwarf2out_do_eh_frame): Declare.
3373 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
3374 (dwarf2out_do_frame): Use it.
3375 (dwarf2out_do_cfi_asm): Likewise.
3376 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
3377 (dwarf2out_assembly_start): Likewise.
3378 (dwarf2out_begin_prologue): Fix comment.
3379 * toplev.c (compile_file): Always call dwarf2out_frame_finish
3380 if the target needs either debug or unwind DWARF2 info.
3381 * lto-opts.c (lto_write_options): Do not save -fexceptions,
3382 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
3383 -ftrapping-math, -ftrapv and -fwrapv.
3385 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3387 * cgraph.c (cgraph_edge::sreal_frequency): New function.
3388 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
3389 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
3390 (estimate_edge_size_and_time): Likewise.
3391 (ipa_merge_fn_summary_after_inlining): Likewise.
3392 * ipa-inline.c (cgraph_freq_base_rec): Remove.
3393 (compute_uninlined_call_time): Use sreal_frequency.
3394 (compute_inlined_call_time): Likewise.
3395 (ipa_inline): Do not initialize cgraph_freq_base_rec.
3396 * profile-count.c: Include sreal.h.
3397 (profile_count::to_sreal_scale): New.
3398 * profile-count.h: Forward declare sreal.
3399 (profile_count::to_sreal_scale): Declare.
3401 2017-11-13 Nathan Sidwell <nathan@acm.org>
3403 * diagnostic.c (maybe_line_and_column): New.
3404 (diagnostic_get_location_text): Use it.
3405 (diagnostic_report_current_module): Likewise.
3406 (test_diagnostic_get_location_text): Add tests.
3408 2017-11-13 Luis Machado <luis.machado@linaro.org>
3410 * doc/md.texi (Specifying processor pipeline description): Fix
3411 incorrect latency for the div instruction example.
3413 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3415 PR tree-optimization/78821
3416 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
3417 that bit_not_p is the same.
3418 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
3419 (split_group): Count precisely bit_not_p bits in each statement.
3420 (invert_op): New function.
3421 (imm_store_chain_info::output_merged_store): Use invert_op to
3422 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
3423 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
3425 2017-11-13 Martin Liska <mliska@suse.cz>
3427 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
3428 (struct source_info): Likewise.
3429 (add_branch_counts): Likewise.
3430 (add_line_counts): Likewise.
3431 (function_summary): Likewise.
3432 (output_intermediate_line): Likewise.
3433 (generate_results): Likewise.
3435 2017-11-13 Martin Liska <mliska@suse.cz>
3437 * gcov.c (struct block_info): Remove typedef for block_t.
3438 (struct line_info): Likewise.
3439 (line_info::has_block): Likewise.
3440 (EXIT_BLOCK): Likewise.
3441 (unblock): Likewise.
3442 (circuit): Likewise.
3443 (get_cycles_count): Likewise.
3444 (process_file): Likewise.
3445 (read_graph_file): Likewise.
3446 (solve_flow_graph): Likewise.
3447 (find_exception_blocks): Likewise.
3448 (add_line_counts): Likewise.
3449 (accumulate_line_info): Likewise.
3450 (output_line_details): Likewise.
3452 2017-11-13 Martin Liska <mliska@suse.cz>
3454 * gcov.c (struct arc_info): Remove typedef for arc_t.
3455 (struct line_info): Likewise.
3456 (add_branch_counts): Likewise.
3457 (output_branch_count): Likewise.
3458 (function_info::~function_info): Likewise.
3459 (circuit): Likewise.
3460 (output_intermediate_line): Likewise.
3461 (read_graph_file): Likewise.
3462 (solve_flow_graph): Likewise.
3463 (find_exception_blocks): Likewise.
3464 (add_line_counts): Likewise.
3465 (accumulate_line_info): Likewise.
3466 (output_line_details): Likewise.
3467 (output_function_details): Likewise.
3469 2017-11-13 Martin Liska <mliska@suse.cz>
3471 * gcov.c (struct function_info): Remove typedef for function_t.
3472 (struct source_info): Likewise.
3473 (source_info::get_functions_at_location): Likewise.
3474 (solve_flow_graph): Likewise.
3475 (find_exception_blocks): Likewise.
3476 (add_line_counts): Likewise.
3477 (output_intermediate_file): Likewise.
3478 (process_file): Likewise.
3479 (generate_results): Likewise.
3480 (release_structures): Likewise.
3481 (read_graph_file): Likewise.
3482 (read_count_file): Likewise.
3483 (accumulate_line_counts): Likewise.
3484 (output_lines): Likewise.
3486 2017-11-13 Martin Liska <mliska@suse.cz>
3488 * gcov.c (function_info::function_info): Remove num_counts
3489 and add vector<gcov_type>.
3490 (function_info::~function_info): Use the vector.
3491 (process_file): Likewise.
3492 (read_graph_file): Likewise.
3493 (read_count_file): Likewise.
3494 (solve_flow_graph): Likewise.
3496 2017-11-13 Martin Liska <mliska@suse.cz>
3498 * gcov.c (function_info::is_artificial): New function.
3499 (process_file): Erase all artificial early.
3500 (generate_results): Skip as all artificial are already
3503 2017-11-13 Martin Liska <mliska@suse.cz>
3505 * gcov.c (read_graph_file): Store to global vector of functions.
3506 (read_count_file): Iterate the vector.
3507 (process_file): Likewise.
3508 (generate_results): Likewise.
3509 (release_structures): Likewise.
3511 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3513 PR tree-optimization/82954
3514 * gimple-ssa-store-merging.c
3515 (imm_store_chain_info::coalesce_immediate_stores): If
3516 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
3518 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
3520 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
3521 Upddate call to ENDIAN_LANE_N.
3522 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
3523 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
3524 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
3525 and use aarch64_endian_lane_rtx.
3526 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
3528 2017-11-12 Tom de Vries <tom@codesourcery.com>
3530 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
3532 2017-11-12 Tom de Vries <tom@codesourcery.com>
3534 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
3536 2017-11-12 Tom de Vries <tom@codesourcery.com>
3538 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
3540 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3541 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3542 * defaults.h (ASM_OUTPUT_LABELREF): Same.
3544 2017-11-11 Martin Sebor <msebor@redhat.com>
3547 * doc/extend.texi (attribute nonstring): Remove spurious argument.
3550 * prefic.c (translate_name): Replace strncpy with memcpy to
3551 avoid -Wstringop-truncation.
3553 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3555 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
3557 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3559 * predict.c (maybe_hot_frequency_p): Do not use cfun.
3561 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3563 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
3566 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
3568 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
3569 enable generating XXBRH if the value is in a vector register.
3570 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
3571 value is in a vector register.
3572 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
3573 register to register bswap64's instead of doing the GPR sequence
3574 used on previous machines.
3575 (bswapdi2_xxbrd): New insn.
3576 (bswapdi2_reg): Disallow on ISA 3.0.
3577 (register to register bswap64 splitter): Do not split the insn on
3578 ISA 3.0 systems that use XXBRD.
3580 2017-11-10 Martin Sebor <msebor@redhat.com>
3583 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
3585 (find_subframework_file): Same.
3587 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3589 * auto-profile.c (afdo_indirect_call): Drop frequency.
3590 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
3591 (cgraph_node::create_edge): Drop frequency argument.
3592 (cgraph_node::create_indirect_edge): Drop frequency argument.
3593 (cgraph_edge::make_speculative): Drop frequency arguments.
3594 (cgraph_edge::resolve_speculation): Do not update frequencies
3595 (cgraph_edge::dump_edge_flags): Do not dump frequency.
3596 (cgraph_node::dump): Check consistency in IPA mode.
3597 (cgraph_edge::maybe_hot_p): Use IPA counter.
3598 (cgraph_edge::verify_count_and_frequency): Rename to ...
3599 (cgraph_edge::verify_count): ... this one; drop frequency checking.
3600 (cgraph_node::verify_node): Update.
3601 * cgraph.h (struct cgraph_edge): Drop frequency.
3602 (cgraph_edge::frequency): New function.
3603 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
3605 (cgraph_edge::rebuild_edges): Likewise.
3606 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
3607 (duplicate_thunk_for_node): Do not pass frequency.
3608 (cgraph_node::create_clone): Scale only counts.
3609 (cgraph_node::create_virtual_clone): Do not pass frequency.
3610 (cgraph_node::create_edge_including_clones): Do not pass frequency.
3611 (cgraph_node::create_version_clone): Do not pass frequency.
3612 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
3613 (cgraph_node::expand_thunk): Do not pass frequency.
3614 (cgraph_node::create_wrapper): Do not pass frequency.
3615 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
3617 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
3618 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
3619 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
3620 (gather_caller_stats): Use frequency function.
3621 (ipcp_cloning_candidate_p): Use frequency function.
3622 (ipcp_propagate_stage): Use frequency function.
3623 (get_info_about_necessary_edges): Use frequency function.
3624 (update_profiling_info): Update only IPA profile.
3625 (update_specialized_profile): Use frequency functoin.
3626 (perhaps_add_new_callers): Update only IPA profile.
3627 * ipa-devirt.c (ipa_devirt): Use IPA profile.
3628 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
3629 (dump_ipa_call_summary): Use frequency function.
3630 (estimate_edge_size_and_time): Use frequency function.
3631 (ipa_merge_fn_summary_after_inlining): Use frequency function.
3632 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
3633 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
3634 (update_noncloned_counts): ... ths one; scale counts only.
3635 (clone_inlined_nodes): Do not scale frequency.
3636 (inline_call): Do not pass frequency.
3637 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
3638 (compute_inlined_call_time): Use IPA profile.
3639 (want_inline_small_function_p): Use IPA profile.
3640 (want_inline_self_recursive_call_p): Use IPA profile.
3641 (edge_badness): Use IPA profile.
3642 (lookup_recursive_calls): Use IPA profile.
3643 (recursive_inlining): Do not pass frequency.
3644 (resolve_noninline_speculation): Do not update frequency.
3645 (inline_small_functions): Collect max of IPA profile.
3646 (dump_overall_stats): Dump IPA porfile.
3647 (dump_inline_stats): Dump IPA porfile.
3648 (ipa_inline): Collect IPA stats.
3649 * ipa-inline.h (clone_inlined_nodes): Update prototype.
3650 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
3651 (ipa_propagate_frequency): Use frequency function.
3652 (ipa_profile): Cleanup.
3653 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
3654 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
3655 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
3656 (input_node): Do not stream frequency.
3657 (input_edge): Do not stream frequency.
3658 (merge_profile_summaries): Scale only IPA profiles.
3659 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
3660 * predict.c (drop_profile): Do not recompute frequency.
3661 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
3662 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
3663 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
3664 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
3665 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
3666 * tree-inline.c (copy_bb): Do not scale frequency.
3667 (expand_call_inline): Do not scale frequency.
3668 (tree_function_versioning): Do not scale frequency.
3669 * ubsan.c (ubsan_create_edge): Do not pass frequency.
3671 2017-11-10 Julia Koval <julia.koval@intel.com>
3673 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
3674 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
3675 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
3676 (_mm256_maskz_gf2p8affine_epi64_epi8)
3677 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
3678 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
3679 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
3680 (__builtin_ia32_vgf2p8affineqb_v32qi)
3681 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
3682 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
3684 2017-11-10 Tamar Christina <tamar.christina@arm.com>
3688 (arm_option_override): Refactor.
3689 (arm_option_reconfigure_globals): New.
3690 (arm_options_perform_arch_sanity_checks): New.
3691 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
3693 (arm_options_perform_arch_sanity_checks): Likewise
3695 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
3697 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
3698 (power9-qpmul): New.
3699 * rs6000/rs6000.md ("type" attr): Add qmul.
3700 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
3701 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
3702 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
3704 2017-11-10 Martin Sebor <msebor@redhat.com>
3707 * builtins.c (compute_objsize): Handle arrays that
3708 compute_builtin_object_size likes to fail for. Make extern.
3709 * builtins.h (compute_objsize): Declare.
3710 (check_strncpy_sizes): New function.
3711 (expand_builtin_strncpy): Call check_strncpy_sizes.
3712 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
3713 -Wstringop-truncation.
3714 (gimple_fold_builtin_strncat): Same.
3715 * gimple.c (gimple_build_call_from_tree): Set call location.
3716 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
3717 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
3718 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
3719 (handle_builtin_strlen): Use strlen_to_stridx.
3720 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
3722 Use strlen_to_stridx.
3723 (pass_strlen::execute): Release strlen_to_stridx.
3724 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
3725 (-Wstringop-truncation): Document new option.
3727 2017-11-10 Martin Liska <mliska@suse.cz>
3729 PR gcov-profile/82702
3730 * gcov.c (main): Handle intermediate files in a different
3732 (get_gcov_intermediate_filename): New function.
3733 (output_gcov_file): Remove support of intermediate files.
3734 (generate_results): Allocate intermediate file.
3735 (release_structures): Clean-up properly fn_end.
3736 (output_intermediate_file): Start iterating with line 1.
3738 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3740 PR tree-optimization/82929
3741 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3742 ops_swapped_p non-static data member.
3743 (store_immediate_info::store_immediate_info): Clear it.
3744 (imm_store_chain_info::coalesce_immediate_stores): If swapping
3745 ops set ops_swapped_p.
3746 (count_multiple_uses): Handle ops_swapped_p.
3748 2017-11-10 Martin Liska <mliska@suse.cz>
3750 * coverage.c (coverage_init): Stream information about
3751 support of has_unexecuted_blocks.
3752 * doc/gcov.texi: Document that.
3753 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
3754 * gcov.c (read_graph_file): Likewise.
3755 (output_line_beginning): Fix a small issue with
3758 2017-11-10 Bin Cheng <bin.cheng@arm.com>
3760 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
3761 reference of trivial component.
3763 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3766 * gimple-ssa-store-merging.c
3767 (pass_store_merging::terminate_all_aliasing_chains): For
3768 gimple_store_p stmts also call refs_output_dependent_p.
3770 PR rtl-optimization/82913
3771 * compare-elim.c (try_merge_compare): Punt if def_insn is not
3774 2017-11-09 Jeff Law <law@redhat.com>
3776 * vr-values.h: New file with vr_values class.
3777 * tree-vrp.c: Include vr-values.h
3778 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
3779 data objects into the vr_values class.
3780 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
3781 (get_value_range): Make it a member function within vr_values class.
3782 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
3783 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
3784 (op_with_constant_singleton_value_range): Likewise.
3785 (extract_range_for_var_from_comparison_expr): Likewise.
3786 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
3787 (extract_range_from_binary_expr): Likewise.
3788 (extract_range_from_unary_expr): Likewise.
3789 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
3790 (check_for_binary_op_overflow, extract_range_basic): Likewise.
3791 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
3792 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
3793 (compare_name_with_value, compare_names): Likewise.
3794 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
3795 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
3796 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
3797 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
3798 (extract_range_from_phi_node): Likewise.
3799 (simplify_truth_ops_using_ranges): Likewise.
3800 (simplify_div_or_mod_using_ranges): Likewise.
3801 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
3802 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
3803 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
3804 (simplify_float_conversion_using_ranges): Likewise.
3805 (simplify_internal_call_using_ranges): Likewise.
3806 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
3807 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
3808 poitner via x_vr_values for calls into gimple folder.
3809 (vrp_initialize_lattice): Make this the vr_values ctor.
3810 (vrp_free_lattice): Make this the vr_values dtor.
3811 (set_vr_value): New function.
3812 (class vrp_prop): Add vr_values data member. Add various member
3813 functions as well as member functions that delegate to vr_values.
3814 (check_array_ref): Make a member function within vrp_prop class.
3815 (search_for_addr_array, vrp_initialize): Likewise.
3816 (vrp_finalize): Likewise. Revamp to avoid direct access to
3817 vr_value, values_propagated, etc.
3818 (check_array_bounds): Extract vrp_prop class instance pointer from
3819 walk info structure. Use it to call member functions.
3820 (check_all_array_refs): Make a member function within vrp_prop class.
3821 Smuggle class instance pointer via walk info structure.
3822 (x_vr_values): New local static.
3823 (vrp_valueize): Use x_vr_values to get class instance.
3824 (vr_valueize_1): Likewise.
3825 (class vrp_folder): Add vr_values data member. Add various member
3826 functions as well as member functions that delegate to vr_values.
3827 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
3828 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
3829 class instance from vr_values. Use it to call member functions.
3830 (vrp_dom_walker): Add vr_values data member.
3831 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
3832 instance via x_vr_values.
3833 (identify_jump_threads): Accept vr_values as argument. Store
3834 it into the walker structure.
3835 (evrp_dom_walker): Add vr_values class data member. Add various
3837 (evrp_dom_walker::try_find_new_range): Use vr_values data
3838 member to access the memory allocator.
3839 (evrp_dom_walker::before_dom_children): Store vr_values class
3840 instance into the vrp_folder class.
3841 (evrp_dom_walker::push_value_range): Rework to avoid direct
3842 access to num_vr_values and vr_value.
3843 (evrp_dom_walker::pop_value_range): Likewise.
3844 (execute_early_vrp): Remove call to vrp_initialize_lattice.
3845 Use vr_values to get to dump_all_value_ranges member function.
3846 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
3847 and simplify_cond_using_ranges_2 via vrp_prop class instance.
3848 Pass vr_values class instance down to identify_jump_threads.
3849 Remove call to vrp_free_lattice.
3850 (debug_all_value_ranges): Remove.
3852 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
3853 (vrp_folder): Likewise.
3855 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
3856 Get it from the existing bitmap instead.
3857 (vrp_intersect_ranges_1): Likewise.
3859 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3861 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3863 (store_immediate_info::store_immediate_info): Add bitnotp argument,
3864 set bit_not_p to it.
3865 (imm_store_chain_info::coalesce_immediate_stores): Break group
3866 if bit_not_p is different.
3867 (count_multiple_uses, split_group,
3868 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
3869 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
3870 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
3871 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
3872 to store_immediate_info ctor.
3874 2017-11-09 Jim Wilson <jimw@sifive.com>
3876 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
3877 (scan_prog_file): Likewise.
3879 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3881 * bb-reorder.c (max_entry_frequency): Remove.
3882 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
3883 connect_traces, push_to_next_round_p): Remove prototypes.
3884 (find_traces_1_round): Use counts only.
3885 (push_to_next_round_p): Likewise.
3886 (find_traces): Likewise.
3887 (rotate_loop): Likewise.
3888 (find_traces_1_round): Likewise.
3889 (connect_traces): Likewise.
3890 (edge_order): Likewise.
3892 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
3894 * config/arm/arm.c (output_return_instruction): Add comments to
3895 indicate requirement for cmse_nonsecure_entry return to account
3896 for the size of clearing instruction output here.
3897 (thumb_exit): Likewise.
3898 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
3899 return in hardfloat mode.
3901 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
3903 * config/rs6000/rs6000.c (machine_function): Add a bool,
3904 "toc_is_wrapped_separately".
3905 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
3906 if it wasn't explicitly set or unset, we are optimizing for speed, and
3907 doing separate shrink-wrapping.
3908 (rs6000_get_separate_components): Enable the TOC component if
3909 saving the TOC register in the prologue.
3910 (rs6000_components_for_bb): Handle the TOC component.
3911 (rs6000_emit_prologue_components): Store the TOC register where needed.
3912 (rs6000_set_handled_components): Mark TOC as handled, if handled.
3913 (rs6000_emit_prologue): Don't save the TOC if that is already done.
3915 2017-11-09 Martin Jambor <mjambor@suse.cz>
3917 * ipa-param-manipulation.c: New file.
3918 * ipa-param-manipulation.h: Likewise.
3919 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
3920 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
3921 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
3922 (ipa_parm_adjustment): Likewise.
3923 (ipa_parm_adjustment_vec): Likewise.
3924 (ipa_get_vector_of_formal_parms): Moved declaration to
3925 ipa-param-manipulation.h.
3926 (ipa_get_vector_of_formal_parm_types): Likewise.
3927 (ipa_modify_formal_parameters): Likewise.
3928 (ipa_modify_call_arguments): Likewise.
3929 (ipa_combine_adjustments): Likewise.
3930 (ipa_dump_param_adjustments): Likewise.
3931 (ipa_modify_expr): Likewise.
3932 (ipa_get_adjustment_candidate): Likewise.
3933 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
3934 ipa-param-manipulation.c.
3935 (ipa_get_vector_of_formal_parm_types): Likewise.
3936 (ipa_modify_formal_parameters): Likewise.
3937 (ipa_modify_call_arguments): Likewise.
3938 (ipa_modify_expr): Likewise.
3939 (get_ssa_base_param): Likewise.
3940 (ipa_get_adjustment_candidate): Likewise.
3941 (index_in_adjustments_multiple_times_p): Likewise.
3942 (ipa_combine_adjustments): Likewise.
3943 (ipa_dump_param_adjustments): Likewise.
3944 * tree-sra.c: Also include ipa-param-manipulation.h
3945 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
3948 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3949 Alan Hayward <alan.hayward@arm.com>
3950 David Sherwood <david.sherwood@arm.com>
3952 * doc/sourcebuild.texi (vect_masked_store): Document.
3954 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3955 Alan Hayward <alan.hayward@arm.com>
3956 David Sherwood <david.sherwood@arm.com>
3958 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
3960 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3961 Alan Hayward <alan.hayward@arm.com>
3962 David Sherwood <david.sherwood@arm.com>
3964 * doc/sourcebuild.texi (vect_variable_length): Document.
3966 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3967 Alan Hayward <alan.hayward@arm.com>
3968 David Sherwood <david.sherwood@arm.com>
3970 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
3972 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3973 Alan Hayward <alan.hayward@arm.com>
3974 David Sherwood <david.sherwood@arm.com>
3976 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
3978 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3979 Alan Hayward <alan.hayward@arm.com>
3980 David Sherwood <david.sherwood@arm.com>
3982 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
3983 previously undocumented selectors.
3984 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
3986 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3988 * doc/rtl.texi (const_vector): Say that elements can be
3989 const_wide_ints too.
3990 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
3991 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
3992 (gen_vec_duplicate): Use it instead of CONSTANT_P.
3993 * optabs.c (expand_vector_broadcast): Likewise.
3995 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3996 Alan Hayward <alan.hayward@arm.com>
3997 David Sherwood <david.sherwood@arm.com>
3999 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
4000 scaled index even if the unscaled address was invalid.
4001 Don't increase the complexity of using a scale in that case.
4003 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
4004 Alan Hayward <alan.hayward@arm.com>
4005 David Sherwood <david.sherwood@arm.com>
4007 * doc/rtl.texi: Rewrite the subreg rules so that they partition
4008 the inner register into REGMODE_NATURAL_SIZE bytes rather than
4009 UNITS_PER_WORD bytes.
4010 * emit-rtl.c (validate_subreg): Divide subregs into blocks
4011 based on REGMODE_NATURAL_SIZE of the inner mode.
4012 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
4013 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
4014 * expmed.c (lowpart_bit_field_p): Divide the value up into
4015 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
4016 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
4017 whether something is likely to occupy more than one register.
4019 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
4022 * ipa-inline-transform.c (update_noncloned_frequencies): Use
4023 profile_count::adjust_for_ipa_scaling.
4024 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
4025 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
4027 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
4029 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4031 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4032 (split_group): Add total_orig and total_new arguments, estimate the
4033 number of statements related to the store group without store merging
4034 and with store merging.
4035 (imm_store_chain_info::output_merged_store): Adjust split_group
4036 callers, punt if estimated number of statements with store merging
4037 is not smaller than estimated number of statements without it.
4039 (handled_load): Remove has_single_use checks.
4040 (pass_store_merging::process_store): Likewise.
4042 2017-11-09 Richard Biener <rguenther@suse.de>
4044 PR tree-optimization/82902
4045 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
4047 2017-11-09 Martin Liska <mliska@suse.cz>
4050 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
4053 2017-11-09 Martin Liska <mliska@suse.cz>
4055 PR tree-optimization/82669
4056 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4058 2017-11-09 Martin Liska <mliska@suse.cz>
4060 PR gcov-profile/48463
4061 * coverage.c (coverage_begin_function): Output also end locus
4062 of a function and information whether the function is
4064 * gcov-dump.c (tag_function): Parse and print the information.
4065 * gcov.c (INCLUDE_MAP): Add include.
4066 (INCLUDE_SET): Likewise.
4067 (struct line_info): Move earlier in the source file because
4068 of vector<line_info> in function_info structure.
4069 (line_info::line_info): Likewise.
4070 (line_info::has_block): Likewise.
4071 (struct source_info): Add new member index.
4072 (source_info::get_functions_at_location): New function.
4073 (function_info::group_line_p): New function.
4074 (output_intermediate_line): New function.
4075 (output_intermediate_file): Use the mentioned function.
4076 (struct function_start): New.
4077 (struct function_start_pair_hash): Likewise.
4078 (process_file): Add code that identifies group functions.
4079 Assign lines either to global or function scope.
4080 (generate_results): Skip artificial functions.
4081 (find_source): Assign index for each source file.
4082 (read_graph_file): Read new flag artificial and end_line.
4083 (add_line_counts): Assign it either to global of function scope.
4084 (accumulate_line_counts): Isolate core of the function to
4085 accumulate_line_info and call it for both function and global
4087 (accumulate_line_info): New function.
4088 (output_line_beginning): Fix GNU coding style.
4089 (print_source_line): New function.
4090 (output_line_details): Likewise.
4091 (output_function_details): Likewise.
4092 (output_lines): Iterate both source (global) scope and function
4094 (struct function_line_start_cmp): New class.
4095 * doc/gcov.texi: Reflect changes in documentation.
4097 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4100 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4101 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4102 and similarly for not instead of neg.
4104 2017-11-08 Andi Kleen <ak@linux.intel.com>
4106 * config/i386/i386.opt: Add -mforce-indirect-call.
4107 * config/i386/predicates.md: Check for flag_force_indirect_call.
4108 * doc/invoke.texi: Document -mforce-indirect-call
4110 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4112 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
4114 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
4115 config/riscv/riscv.c (predict.h): New include.
4116 (riscv_slow_unaligned_access_p): No longer static.
4117 (riscv_block_move_straight): Add require.
4118 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4120 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4123 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
4124 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
4125 nonimmediate_operand predicate for operand 1 instead of
4128 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4130 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
4132 * config/aarch64/constraints.md (Uml): New constraint.
4133 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
4136 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4138 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
4139 of two vec_duplicates into a vec_concat.
4141 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4143 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4144 Simplify vec_merge of vec_duplicate and vec_concat.
4145 * config/aarch64/constraints.md (Utq): New constraint.
4146 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
4149 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4151 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4152 Simplify vec_merge of vec_duplicate and const_vector.
4153 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
4155 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
4156 mode iterator. Update predicate on operand 1 to
4157 handle non-const_vec constants. Delete constraints.
4158 (*aarch64_combinez_be<mode>): Likewise for operand 2.
4160 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4162 PR tree-optimization/78821
4163 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
4165 (store_operand_info::store_operand_info): Initialize it to false.
4166 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
4167 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
4168 store in the group, and if chain_info is non-NULL, to ignore altogether
4170 (compatible_load_p): Fail if bit_not_p does not match.
4171 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
4172 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
4173 (pass_store_merging::process_store): Adjust
4174 terminate_all_aliasing_chains calls to pass NULL in all current spots,
4175 call terminate_all_aliasing_chains newly when adding a store into
4176 a chain with non-NULL chain_info.
4178 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4180 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
4182 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4184 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4186 (aarch64_layout_frame): Initialise emit_frame_chain.
4187 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
4188 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
4190 2017-11-08 Martin Liska <mliska@suse.cz>
4192 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
4193 of gimple_call_internal_p.
4195 2017-11-07 Tom de Vries <tom@codesourcery.com>
4197 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
4199 2017-11-07 Tom de Vries <tom@codesourcery.com>
4201 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4204 2017-11-08 Martin Liska <mliska@suse.cz>
4207 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
4209 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4211 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
4213 2017-11-07 Jakub Jelinek <jakub@redhat.com>
4216 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
4217 (*cmp<mode>_ccz_1): New insn with $k alternative.
4220 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
4221 RTX_COMM_COMPARE as commutative as well.
4222 (ix86_binary_operator_ok): Formatting fix.
4223 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
4224 *<code><mode>3<mask_name><round_saeonly_name>,
4225 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
4226 *<s>mul<mode>3_highpart<mask_name>,
4227 *vec_widen_umult_even_v16si<mask_name>,
4228 *vec_widen_umult_even_v8si<mask_name>,
4229 *vec_widen_umult_even_v4si<mask_name>,
4230 *vec_widen_smult_even_v16si<mask_name>,
4231 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
4232 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
4233 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
4234 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
4235 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
4236 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
4237 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
4238 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
4239 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
4240 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
4241 ix86_binary_operator_ok. Formatting fixes.
4242 (*<plusminus_insn><mode>3<mask_name><round_name>,
4243 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
4246 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
4248 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
4249 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
4250 eq and ne if TARGET_ISEL.
4251 (cmp): New code_iterator.
4252 (UNS, UNSU_, UNSIK): New code_attrs.
4253 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
4254 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
4256 ("eq<mode>3"): ... this.
4257 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
4259 ("ne<mode>3"): ... this.
4261 2017-11-07 Julia Koval <julia.koval@intel.com>
4264 * common/config/i386/i386-common.c
4265 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
4266 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
4267 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
4268 * config/i386/i386.opt: Ditto.
4269 * config/i386/i386.c (ix86_target_string): Ditto.
4270 (ix86_option_override_internal): Ditto.
4271 (ix86_init_mpx_builtins): Move MPX to args2.
4272 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
4273 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4274 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4275 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4276 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4277 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4278 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
4280 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
4283 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
4284 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
4285 (zero-extendsidi peephole2): Remove peephole.
4287 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4290 * stor-layout.c: Include attribs.h.
4291 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
4292 explicit lookup of "aligned" attribute.
4294 2017-11-07 Andrew Waterman <andrew@sifive.com>
4296 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
4297 (riscv_expand_block_move): Likewise.
4298 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
4300 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
4301 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
4302 * config/riscv/riscv.c (riscv_block_move_straight): New function.
4303 (riscv_adjust_block_mem): Likewise.
4304 (riscv_block_move_loop): Likewise.
4305 (riscv_expand_block_move): Likewise.
4306 * config/riscv/riscv.md (movmemsi): New pattern.
4308 2017-11-07 Michael Clark <michaeljclark@mac.com>
4310 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
4311 (MUSL_DYNAMIC_LINKER): Likewise.
4313 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
4315 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
4316 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
4319 2017-11-07 Richard Biener <rguenther@suse.de>
4321 * match.pd: Fix build.
4323 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
4324 Jackson Woodruff <jackson.woodruff@arm.com>
4326 PR tree-optimization/71026
4327 * match.pd: Canonicalize negate in division.
4329 2017-11-07 Sudakshina Das <sudi.das@arm.com>
4332 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
4334 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4336 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
4337 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
4339 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4341 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
4342 non-scalar integral types.
4343 * match.pd (negate_expr_p): Handle MINUS_EXPR.
4344 (-(A-B), -(~A)): New transformations.
4346 2017-11-07 Tom de Vries <tom@codesourcery.com>
4348 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
4349 semicolon after "do {} while (0)".
4350 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4351 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4352 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4353 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4354 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4356 2017-11-07 Tom de Vries <tom@codesourcery.com>
4358 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
4359 after "do {} while (0)".
4360 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4361 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4362 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4363 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4364 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4366 2017-11-07 Tom de Vries <tom@codesourcery.com>
4369 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
4371 (arm_rtx_costs_internal): Add missing semicolon after
4372 HANDLE_NARROW_SHIFT_ARITH call.
4374 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4376 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4377 disable isel if it was not set explicitly.
4379 2017-11-06 James Bowman <james.bowman@ftdichip.com>
4381 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
4382 (add_type_attribute) likewise.
4384 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
4386 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
4387 of struct ix86_frame.
4388 (ix86_initial_elimination_offset): Likewise.
4389 (ix86_expand_split_stack_prologue): Likewise.
4391 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
4393 * tree-vrp.h (enum value_range_type): Update stale comment.
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_expand_vec_perm)
4400 (aarch64_expand_vec_perm_const): Take the number of units too.
4401 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
4402 (aarch64_expand_vec_perm_const): Likewise.
4403 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
4404 (vec_perm<mode>): Update accordingly.
4406 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4407 Alan Hayward <alan.hayward@arm.com>
4408 David Sherwood <david.sherwood@arm.com>
4410 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
4411 Take the number of units too.
4412 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
4413 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
4414 but check for a vector mode before rather than after the call.
4415 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
4416 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
4417 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
4418 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
4419 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
4420 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
4421 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
4422 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
4423 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
4424 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
4425 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
4426 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
4427 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
4428 (aarch64_sqdmull2_n<mode>): Update accordingly.
4430 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4431 Alan Hayward <alan.hayward@arm.com>
4432 David Sherwood <david.sherwood@arm.com>
4434 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
4435 the number of units too.
4436 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
4437 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
4438 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
4439 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
4440 (vec_store_lanesxi<mode>): Update accordingly.
4442 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4443 Alan Hayward <alan.hayward@arm.com>
4444 David Sherwood <david.sherwood@arm.com>
4446 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
4447 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
4448 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
4449 of units rather than the mode.
4450 * config/aarch64/iterators.md (nunits): New mode attribute.
4451 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4452 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
4453 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
4454 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
4455 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
4456 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
4457 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
4458 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
4459 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
4460 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
4461 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4462 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
4463 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
4464 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
4465 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
4466 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
4467 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
4468 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
4469 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
4470 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
4471 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
4472 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
4473 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
4474 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4475 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
4476 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
4477 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4478 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
4479 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
4480 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
4481 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
4482 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
4483 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
4484 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
4485 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
4486 (aarch64_simd_vec_setv2di): Likewise.
4488 2017-11-06 Carl Love <cel@us.ibm.com>
4490 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
4492 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
4493 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
4495 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
4497 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
4498 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
4499 Add power 8 macro expansions.
4500 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
4501 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
4502 power 8 instructions. (VSX_XXBR): Add iterator.
4504 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
4506 * config/arm/arm.md (predicable_short_it): Change default to "no",
4507 improve documentation, remove uses that are identical to the default.
4508 (enabled_for_depr_it): Rename to enabled_for_short_it.
4509 * gcc/config/arm/arm-fixed.md (predicable_short_it):
4510 Remove default uses.
4511 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
4512 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
4513 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
4514 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
4516 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
4519 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
4520 float128 helper macros, which are no longer used after deleting
4521 the old 'q' built-in functions, and moving the round to odd
4522 built-in functions to being special built-in functions.
4523 (BU_FLOAT128_2): Likewise.
4524 (BU_FLOAT128_1_HW): Likewise.
4525 (BU_FLOAT128_2_HW): Likewise.
4526 (BU_FLOAT128_3_HW): Likewise.
4527 (FABSQ): Delete old 'q' built-in functions.
4528 (COPYSIGNQ): Likewise.
4529 (SQRTF128_ODD): Move round to odd built-in functions to be
4530 special built-in functions, so that we can handle
4531 -mabi=ieeelongdouble.
4532 (TRUNCF128_ODD): Likewise.
4533 (ADDF128_ODD): Likewise.
4534 (SUBF128_ODD): Likewise.
4535 (MULF128_ODD): Likewise.
4536 (DIVF128_ODD): Likewise.
4537 (FMAF128_ODD): Likewise.
4538 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
4539 built-in names to 'f128'.
4540 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
4541 old 'q' built-in functions, as the machine independent code for
4542 'f128' built-in functions handles this.
4543 (rs6000_expand_builtin): Add expansion for float128 round to odd
4544 functions, keying off on -mabi=ieeelongdouble of whether to use
4545 the KFmode or TFmode variant.
4546 (rs6000_init_builtins): Initialize the _Float128 round to odd
4548 * doc/extend.texi (PowerPC Built-in Functions): Document the old
4549 _Float128 'q' built-in functions are now mapped into the new
4550 'f128' built-in functions.
4552 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
4554 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
4555 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
4557 2017-11-06 David Malcolm <dmalcolm@redhat.com>
4560 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
4561 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
4562 * toplev.c: Include "ipa-fnsummary.h".
4563 (toplev::finalize): Call ipa_fnsummary_c_finalize.
4565 2017-11-06 Jakub Jelinek <jakub@redhat.com>
4567 PR tree-optimization/82838
4568 * gimple-ssa-store-merging.c
4569 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
4570 on a separate gimple_seq which is then appended to seq.
4572 2017-11-06 Jeff Law <law@redhat.com>
4575 * config/i386/i386.c (PROBE_INTERVAL): Remove.
4576 (get_probe_interval): New functions.
4577 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
4578 (ix86_adjust_stack_and_probe): Likewise.
4579 (output_adjust_stack_and_probe): Likewise.
4580 (ix86_emit_probe_stack_range): Likewise.
4581 (ix86_expand_prologue): Likewise.
4583 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4585 PR tree-optimization/82816
4586 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
4587 if the modes of the two types are the same.
4588 (convert_plusminus_to_widen): Likewise.
4590 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4592 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
4594 (usadv16qi): New define_expand.
4595 (usadv8hi): New define_expand.
4597 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
4600 * ipa-inline-transform.c (update_noncloned_frequencies): Always
4602 (inline_transform): Likewise.
4603 * predict.c (counts_to_freqs): Remove useless conditional.
4604 * profile-count.h (profile_count::apply_scale): Move sanity check.
4605 * tree-inline.c (copy_bb): Always scale.
4606 (copy_cfg_body): Likewise.
4608 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
4611 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
4613 (*cmp_ite0): Add enabled_for_depr_it attribute.
4614 (*cmp_ite1): Likewise.
4616 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4618 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
4621 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4623 * tree-vrp.c (vrp_int_const_binop): Return true on success and
4624 return the value by pointer.
4625 (extract_range_from_multiplicative_op_1): Update accordingly.
4626 Return as soon as an operation fails.
4628 2017-11-05 Tom de Vries <tom@codesourcery.com>
4631 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
4632 (DEF_SANITIZER_BUILTIN): ... here.
4633 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
4634 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
4636 2017-11-05 Tom de Vries <tom@codesourcery.com>
4639 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
4641 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
4642 ASM_OUTPUT_BEFORE_CASE_LABEL call.
4643 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
4645 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4646 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4647 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4649 2017-11-05 Tom de Vries <tom@codesourcery.com>
4652 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
4655 2017-11-04 Michael Clark <michaeljclark@mac.com>
4657 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
4658 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
4660 (*addsi3_extended): Likewise.
4661 (*addsi3_extended2): Likewise.
4662 (<optab>si3): Likewise.
4663 (<optab>di3): Likewise.
4664 (<optab><mode>3): Likewise.
4665 (<*optabe>si3_internal): Likewise.
4666 (zero_extendqi<SUPERQI:mode>2): Likewise.
4667 (*add<mode>hi3): Likewise.
4668 (*xor<mode>hi3): Likewise.
4669 (<optab>di3): Likewise.
4670 (*<optab>si3_extend): Likewise.
4671 (*sge<u>_<X:mode><GPR:mode>): Likewise.
4672 (*slt<u>_<X:mode><GPR:mode>): Likewise.
4673 (*sle<u>_<X:mode><GPR:mode>): Likewise.
4675 2017-11-04 Andrew Waterman <andrew@sifive.com>
4677 * config/riscv/riscv.c (riscv_option_override): Conditionally set
4678 TARGET_STRICT_ALIGN based upon -mtune argument.
4680 2017-11-04 Andrew Waterman <andrew@sifive.com>
4682 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
4684 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
4686 * config/i386/i386.c (choose_basereg): Use optional scratch
4687 register and add assertion.
4688 (x86_emit_outlined_ms2sysv_save): Use scratch register when
4689 needed, and don't allocate stack.
4690 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
4691 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
4692 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
4694 2017-11-03 Jeff Law <law@redhat.com>
4696 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
4697 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
4698 to probe at the start of a noreturn function.
4700 2017-11-03 Jakub Jelinek <jakub@redhat.com>
4702 PR tree-optimization/78821
4703 * gimple-ssa-store-merging.c: Update the file comment.
4704 (MAX_STORE_ALIAS_CHECKS): Define.
4705 (struct store_operand_info): New type.
4706 (store_operand_info::store_operand_info): New constructor.
4707 (struct store_immediate_info): Add rhs_code and ops data members.
4708 (store_immediate_info::store_immediate_info): Add rhscode, op0r
4709 and op1r arguments to the ctor, initialize corresponding data members.
4710 (struct merged_store_group): Add load_align_base and load_align
4712 (merged_store_group::merged_store_group): Initialize them.
4713 (merged_store_group::do_merge): Update them.
4714 (merged_store_group::apply_stores): Pick the constant for
4715 encode_tree_to_bitpos from one of the two operands, or skip
4716 encode_tree_to_bitpos if neither operand is a constant.
4717 (class pass_store_merging): Add process_store method decl. Remove
4718 bool argument from terminate_all_aliasing_chains method decl.
4719 (pass_store_merging::terminate_all_aliasing_chains): Remove
4720 var_offset_p argument and corresponding handling.
4721 (stmts_may_clobber_ref_p): New function.
4722 (compatible_load_p): New function.
4723 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
4724 if there is overlap and rhs_code is not INTEGER_CST. For
4725 non-overlapping stores terminate group if rhs is not mergeable.
4726 (get_alias_type_for_stmts): Change first argument from
4727 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
4728 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
4729 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
4731 (get_location_for_stmts): Change first argument from
4732 auto_vec<gimple *> & to vec<gimple *> &.
4733 (struct split_store): Remove orig_stmts data member, add orig_stores.
4734 (split_store::split_store): Create orig_stores rather than orig_stmts.
4735 (find_constituent_stmts): Renamed to ...
4736 (find_constituent_stores): ... this. Change second argument from
4737 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
4738 to info structures rather than the statements.
4739 (split_group): Rename ALLOW_UNALIGNED argument to
4740 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
4741 it. Adjust find_constituent_stores caller.
4742 (imm_store_chain_info::output_merged_store): Handle rhs_code other
4743 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
4744 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
4745 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
4746 (mem_valid_for_store_merging): New function.
4747 (handled_load): New function.
4748 (pass_store_merging::process_store): New method.
4749 (pass_store_merging::execute): Use process_store method. Adjust
4750 terminate_all_aliasing_chains caller.
4752 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4754 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4755 Return true for more constants, symbols and label references.
4756 (aarch64_valid_floating_const): Remove unused function.
4758 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4761 * config/aarch64/aarch64.c (aarch64_layout_frame):
4762 Undo forcing of LR at bottom of frame.
4764 2017-11-03 Jeff Law <law@redhat.com>
4767 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
4768 for int_registers_saved.
4770 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
4771 extracted from tree-ssa-dom.c.
4772 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
4773 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
4774 (record_equivalences_from_incoming_edge): Add additional argument
4775 to single_pred_edge_ignoring_loop_edges call.
4776 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
4777 (uncprop_dom_walker::before_dom_children): Add additional argument
4778 to single_pred_edge_ignoring_loop_edges call.
4779 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
4780 single_pred_edge_ignoring_loop_edges rather than open coding.
4781 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
4783 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
4785 * match.pd (-(-A)): Rewrite.
4787 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4789 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
4790 (rs6000_emit_int_cmove): New declaration.
4791 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
4792 (rs6000_emit_sISEL): Delete.
4793 (rs6000_emit_int_cmove): Make non-static.
4794 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
4795 instead of rs6000_emit_sISEL.
4797 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
4799 * asan.c (create_cond_insert_point): Maintain profile.
4800 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
4802 * basic-block.h (struct basic_block_def): Remove frequency.
4803 (EDGE_FREQUENCY): Use to_frequency
4804 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
4806 (find_traces): Update to use to_frequency.
4807 (find_traces_1_round): Likewise; use only IPA counts.
4808 (bb_to_key): Likewise.
4809 (connect_traces): Use IPA counts only.
4810 (copy_bb_p): Update to use to_frequency.
4811 (fix_up_crossing_landing_pad): Likewise.
4812 (sanitize_hot_paths): Likewise.
4813 * bt-load.c (basic_block_freq): Likewise.
4814 * cfg.c (init_flow): Set count_max to uninitialized.
4815 (check_bb_profile): Remove frequencies; check counts.
4816 (dump_bb_info): Do not dump frequencies.
4817 (update_bb_profile_for_threading): Update counts only.
4818 (scale_bbs_frequencies_int): Likewise.
4819 (MAX_SAFE_MULTIPLIER): Remove.
4820 (scale_bbs_frequencies_gcov_type): Update counts only.
4821 (scale_bbs_frequencies_profile_count): Update counts only.
4822 (scale_bbs_frequencies): Update counts only.
4823 * cfg.h (struct control_flow_graph): Add count-max.
4824 (update_bb_profile_for_threading): Update prototype.
4825 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
4826 (find_many_sub_basic_blocks): Likewise.
4827 * cfgcleanup.c (try_forward_edges): Likewise.
4828 (try_crossjump_to_edge): Likewise.
4829 * cfgexpand.c (expand_gimple_cond): Likewise.
4830 (expand_gimple_tailcall): Likewise.
4831 (construct_init_block): Likewise.
4832 (construct_exit_block): Likewise.
4833 * cfghooks.c (verify_flow_info): Check consistency of counts.
4834 (dump_bb_for_graph): Do not dump frequencies.
4835 (split_block_1): Do not update frequencies.
4836 (split_edge): Do not update frequencies.
4837 (make_forwarder_block): Do not update frequencies.
4838 (duplicate_block): Do not update frequencies.
4839 (account_profile_record): Do not update frequencies.
4840 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
4841 for global heuristics.
4842 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
4843 (expected_loop_iterations_unbounded): Use counts only.
4844 * cfgloopmanip.c (scale_loop_profile): Simplify.
4845 (create_empty_loop_on_edge): Simplify
4847 (duplicate_loop_to_header_edge): Simplify
4848 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
4849 (update_br_prob_note): Take care of removing note when profile
4851 (relink_block_chain): Do not dump frequency.
4852 (rtl_account_profile_record): Use to_frequency.
4853 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
4854 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
4855 (cgraph_update_edges_for_call_stmt_node): Likewise.
4856 (cgraph_edge::verify_count_and_frequency): Update.
4857 (cgraph_node::verify_node): Temporarily disable frequency verification.
4858 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
4859 to_cgraph_frequency.
4860 (cgraph_edge::rebuild_edges): Convert to ipa counts.
4861 * cgraphunit.c (init_lowered_empty_function): Do not initialize
4863 (cgraph_node::expand_thunk): Update profile.
4864 * except.c (dw2_build_landing_pads): Do not update frequency.
4865 * final.c (compute_alignments): Use to_frequency.
4866 (dump_basic_block_info): Do not dump frequency.
4867 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
4868 (dump_gimple_bb_header): Do not dump frequency.
4869 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
4871 * gimple-streamer-in.c (input_bb): Do not stream frequency.
4872 * gimple-streamer-out.c (output_bb): Do not stream frequency.
4873 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
4874 (init_before_recovery): Do not update frequency.
4875 (sched_create_recovery_edges): Do not update frequency.
4876 * hsa-gen.c (convert_switch_statements): Do not update frequency.
4877 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
4878 (ipa_cp_c_finalize): Set max_count to uninitialized.
4879 * ipa-fnsummary.c (get_minimal_bb): Use counts.
4880 (param_change_prob): Use counts.
4881 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
4883 * ipa-split.c (consider_split): Use to_frequency.
4884 (split_function): Use to_frequency.
4885 * ira-build.c (loop_compare_func): Likewise.
4886 (mark_loops_for_removal): Likewise.
4887 (mark_all_loops_for_removal): Likewise.
4888 * loop-doloop.c (doloop_modify): Do not update frequency.
4889 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
4891 * lto-streamer-in.c (input_function): Update count_max.
4892 * omp-expand.c (expand_omp_taskreg): Update count_max.
4893 * omp-simd-clone.c (simd_clone_adjust): Update profile.
4894 * predict.c (maybe_hot_frequency_p): Use to_frequency.
4895 (maybe_hot_count_p): Use ipa counts only.
4896 (maybe_hot_bb_p): Simplify.
4897 (maybe_hot_edge_p): Simplify.
4898 (probably_never_executed): Do not take frequency argument.
4899 (probably_never_executed_bb_p): Do not pass frequency.
4900 (probably_never_executed_edge_p): Likewise.
4901 (combine_predictions_for_bb): Check that profile is nonzero.
4902 (propagate_freq): Do not set frequency.
4903 (drop_profile): Simplify.
4904 (counts_to_freqs): Simplify.
4905 (expensive_function_p): Use to_frequency.
4906 (propagate_unlikely_bbs_forward): Simplify.
4907 (determine_unlikely_bbs): Simplify.
4908 (estimate_bb_frequencies): Add hack to silence graphite issues.
4909 (compute_function_frequency): Use ipa counts.
4910 (pass_profile::execute): Update.
4911 (rebuild_frequencies): Use counts only.
4912 (force_edge_cold): Use counts only.
4913 * profile-count.c (profile_count::dump): Dump new count types.
4914 (profile_count::differs_from_p): Check compatiblity.
4915 (profile_count::to_frequency): New function.
4916 (profile_count::to_cgraph_frequency): New function.
4917 * profile-count.h (struct function): Declare.
4918 (enum profile_quality): Add profile_guessed_local and
4919 profile_guessed_global0.
4920 (class profile_proability): Decrease number of bits to 29;
4921 update from_reg_br_prob_note and to_reg_br_prob_note.
4922 (class profile_count: Update comment; decrease number of bits
4923 to 61. Check compatibility.
4924 (profile_count::compatible_p): New private member function.
4925 (profile_count::ipa_p): New member function.
4926 (profile_count::operator<): Handle global zero correctly.
4927 (profile_count::operator>): Handle global zero correctly.
4928 (profile_count::operator<=): Handle global zero correctly.
4929 (profile_count::operator>=): Handle global zero correctly.
4930 (profile_count::nonzero_p): New member function.
4931 (profile_count::force_nonzero): New member function.
4932 (profile_count::max): New member function.
4933 (profile_count::apply_scale): Handle IPA scalling.
4934 (profile_count::guessed_local): New member function.
4935 (profile_count::global0): New member function.
4936 (profile_count::ipa): New member function.
4937 (profile_count::to_frequency): Declare.
4938 (profile_count::to_cgraph_frequency): Declare.
4939 * profile.c (OVERLAP_BASE): Delete.
4940 (compute_frequency_overlap): Delete.
4941 (compute_branch_probabilities): Do not use compute_frequency_overlap.
4942 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
4943 * sched-ebb.c (rank): Use counts only.
4944 * shrink-wrap.c (handle_simple_exit): Use counts only.
4945 (try_shrink_wrapping): Use counts only.
4946 (place_prologue_for_one_component): Use counts only.
4947 * tracer.c (find_best_predecessor): Use to_frequency.
4948 (find_trace): Use to_frequency.
4949 (tail_duplicate): Use to_frequency.
4950 * trans-mem.c (expand_transaction): Do not update frequency.
4951 * tree-call-cdce.c: Do not update frequency.
4952 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
4953 (gimple_merge_blocks): Likewise.
4954 (gimple_split_edge): Likewise.
4955 (gimple_duplicate_sese_region): Likewise.
4956 (gimple_duplicate_sese_tail): Likewise.
4957 (move_sese_region_to_fn): Likewise.
4958 (gimple_account_profile_record): Likewise.
4959 (insert_cond_bb): Likewise.
4960 * tree-complex.c (expand_complex_div_wide): Likewise.
4961 * tree-eh.c (lower_resx): Update profile.
4962 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
4964 (initialize_cfun): Do not initialize frequencies
4965 (freqs_to_counts): Delete.
4966 (copy_cfg_body): Ignore count parameter.
4967 (copy_body): Update.
4968 (expand_call_inline): Update count_max.
4969 (optimize_inline_calls): Update count_max.
4970 (tree_function_versioning): Update count_max.
4971 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
4972 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
4974 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
4975 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
4976 (try_peel_loop): Likewise.
4977 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
4979 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
4980 (tree_transform_and_unroll_loop): Do not use frequencies
4981 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
4982 Use reliable prediction only.
4983 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
4984 * tree-ssa-sink.c (select_best_block): Use to_frequency.
4985 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
4986 probability scaling.
4987 * tree-ssa-threadupdate.c (create_block_for_threading): Do
4988 not update frequency
4989 (any_remaining_duplicated_blocks): Likewise.
4990 (update_profile): Likewise.
4991 (estimated_freqs_path): Delete.
4992 (freqs_to_counts_path): Delete.
4993 (clear_counts_path): Delete.
4994 (ssa_fix_duplicate_block_edges): Likewise.
4995 (duplicate_thread_path): Likewise.
4996 * tree-switch-conversion.c (gen_inbound_check): Use counts.
4997 * tree-tailcall.c (decrease_profile): Do not update frequency.
4998 (eliminate_tail_call): Likewise.
4999 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
5000 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
5001 (optimize_mask_stores): Likewise.
5002 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5003 * ubsan.c (ubsan_expand_null_ifn): Update profile.
5004 (ubsan_expand_ptr_ifn): Update profile.
5005 * value-prof.c (gimple_ic): Simplify.
5006 * value-prof.h (gimple_ic): Update prototype.
5007 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
5008 * ipa-inline.c (compute_uninlined_call_time): Be sure that
5010 (want_inline_self_recursive_call_p): Likewise.
5011 (resolve_noninline_speculation): Only cummulate defined counts.
5012 (inline_small_functions): Use nonzero_p.
5013 (ipa_inline): Do not access freed node.
5015 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
5017 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5018 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
5020 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
5022 * config/riscv/riscv.c (riscv_legitimize_move): Handle
5023 non-legitimate address.
5025 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5027 * config/rs6000/rs6000.md (*lt0_disi): Delete.
5028 (*lt0_<mode>di, *lt0_<mode>si): New.
5030 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5032 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
5033 TARGET_PAIRED_FLOAT.
5035 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
5036 Jim Wilson <jim.wilson@linaro.org>
5038 * config/aarch64/aarch64-cores.def (saphira): New CPU.
5039 * config/aarch64/aarch64-tune.md: Regenerated.
5040 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
5041 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
5043 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
5045 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
5046 (arc_expand_prologue): Restore blink for millicode.
5047 * config/arc/linux.h (LINK_EH_SPEC): Defined.
5049 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
5052 * config/i386/i386.c (ix86_vector_duplicate_value): Use
5053 gen_vec_duplicate after forcing the scalar into a register.
5055 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5057 * combine (try_combine): Print the insns input to try_combine to the
5060 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5063 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5064 Remove second argument from aarch64_process_target_attr call.
5065 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5067 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5069 (aarch64_handle_attr_arch): Remove second argument.
5070 (aarch64_handle_attr_cpu): Ditto.
5071 (aarch64_handle_attr_tune): Ditto.
5072 (aarch64_handle_attr_isa_flags): Ditto.
5073 (aarch64_process_one_target_attr): Ditto.
5074 (aarch64_process_target_attr): Ditto.
5075 (aarch64_option_valid_attribute_p): Remove second argument.
5076 on aarch64_process_target_attr call.
5078 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5080 * diagnostic.c: Include "selftest-diagnostic.h".
5081 (selftest::assert_location_text): New function.
5082 (selftest::test_diagnostic_get_location_text): New function.
5083 (selftest::diagnostic_c_tests): Call it.
5085 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5087 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5088 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5089 (class selftest::test_diagnostic_context): Move to...
5090 * selftest-diagnostic.c: New file.
5091 * selftest-diagnostic.h: New file.
5093 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5095 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5096 offset range for FT32B.
5097 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5098 * config/ft32/ft32.md: Add TARGET_NOPM.
5099 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5100 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5102 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5104 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5106 2017-11-02 Jeff Law <law@redhat.com>
5108 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5109 virtual keyword on FINAL OVERRIDE members.
5111 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5112 virtuals together. Add virtual destructor.
5113 (substitute_and_fold_engine): Similarly.
5115 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
5117 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5119 2017-11-02 Richard Biener <rguenther@suse.de>
5121 PR tree-optimization/82795
5122 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
5124 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5126 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
5128 (gcc_GAS_CHECK_FEATURE): Remove.
5129 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
5130 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
5131 * configure: Regenerate.
5133 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
5135 * config/arc/arc.c (hwloop_optimize): Account for empty
5138 2017-11-02 Richard Biener <rguenther@suse.de>
5141 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
5142 Truncate ARRAY_REF index and element size.
5144 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5146 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
5148 2017-11-01 Jeff Law <law@redhat.com>
5150 * tree-ssa-ccp.c (ccp_folder): New class derived from
5151 substitute_and_fold_engine.
5152 (ccp_folder::get_value): New member function.
5153 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
5154 (ccp_fold_stmt): Remove prototype.
5155 (ccp_finalize): Call substitute_and_fold from the ccp_class.
5156 * tree-ssa-copy.c (copy_folder): New class derived from
5157 substitute_and_fold_engine.
5158 (copy_folder::get_value): Renamed from get_value.
5159 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
5160 * tree-vrp.c (vrp_folder): New class derived from
5161 substitute_and_fold_engine.
5162 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
5163 (vrp_folder::get_value): New member function.
5164 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
5165 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
5166 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
5167 provide a class interface to folder/substitute routines.
5168 (ssa_prop_fold_stmt_fn): Remove typedef.
5169 (ssa_prop_get_value_fn): Likewise.
5170 (subsitute_and_fold): Remove prototype.
5171 (replace_uses_in): Likewise.
5172 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
5173 Renamed from replace_uses_in. Call the virtual member function
5174 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
5175 (substitute_and_fold_dom_walker): Remove initialization of
5176 data member entries for calbacks. Add substitute_and_fold_engine
5177 member and initialize it.
5178 (substitute_and_fold_dom_walker::before_dom_children0: Use the
5179 member functions for get_value, replace_phi_args_in c
5180 replace_uses_in, and fold_stmt calls.
5181 (substitute_and_fold_engine::substitute_and_fold): Renamed from
5182 substitute_and_fold. Remove assert. Update ctor call.
5184 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
5185 (ssa_prop_visit_phi_fn): Likewise.
5186 (class ssa_propagation_engine): New class to provide an interface
5188 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
5190 (ssa_prop_visit_phi): Likewise.
5191 (ssa_propagation_engine::simulate_stmt): Moved into class.
5192 Call visit_phi/visit_stmt from the class rather than via
5193 file scoped static variables.
5194 (ssa_propagation_engine::simulate_block): Moved into class.
5195 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
5196 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
5197 set file scoped statics for the visit_stmt/visit_phi callbacks.
5198 * tree-complex.c (complex_propagate): New class derived from
5199 ssa_propagation_engine.
5200 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
5201 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
5202 (tree_lower_complex): Call ssa_propagate via the complex_propagate
5204 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
5205 ssa_propagation_engine.
5206 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
5207 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
5208 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
5209 * tree-ssa-copy.c (copy_prop): New class derived from
5210 ssa_propagation_engine.
5211 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
5212 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
5213 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
5214 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
5215 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
5216 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
5217 (execute_vrp): Call ssa_propagate from the vrp_prop class.
5219 2017-11-01 Jakub Jelinek <jakub@redhat.com>
5221 PR rtl-optimization/82778
5222 PR rtl-optimization/82597
5223 * compare-elim.c (struct comparison): Add in_a_setter field.
5224 (find_comparison_dom_walker::before_dom_children): Remove killed
5225 bitmap and df_simulate_find_defs call, instead walk the defs.
5226 Compute last_setter and initialize in_a_setter. Merge definitions
5227 with first initialization for a few variables.
5228 (try_validate_parallel): Use insn_invalid_p instead of
5229 recog_memoized. Return insn rather than just the pattern.
5230 (try_merge_compare): Fix up comment. Don't uselessly test if
5231 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
5233 (execute_compare_elim_after_reload): Remove df_chain_add_problem
5236 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5237 Alan Hayward <alan.hayward@arm.com>
5238 David Sherwood <david.sherwood@arm.com>
5240 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
5241 aarch64_hard_regno_nregs to get the number of registers
5244 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5245 Alan Hayward <alan.hayward@arm.com>
5246 David Sherwood <david.sherwood@arm.com>
5248 * config/aarch64/constraints.md (Upl): Rename to...
5250 * config/aarch64/aarch64.md
5251 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
5254 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5255 Alan Hayward <alan.hayward@arm.com>
5256 David Sherwood <david.sherwood@arm.com>
5258 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
5259 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
5262 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5263 Alan Hayward <alan.hayward@arm.com>
5264 David Sherwood <david.sherwood@arm.com>
5266 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
5267 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
5268 (aarch64_evpc_dup): Generate rtl direcly, rather than using
5270 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
5271 of a single element.
5272 * config/aarch64/iterators.md: Add a comment above the permute
5273 unspecs to say that they are generated directly by
5274 aarch64_expand_vec_perm_const.
5275 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
5277 2017-11-01 Nathan Sidwell <nathan@acm.org>
5279 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
5281 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5283 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
5284 and medany code models, and describe what they do.
5286 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5288 Revert accidental duplicate:
5290 * combine.c (can_change_dest_mode): Reject changes in
5291 REGMODE_NATURAL_SIZE.
5293 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
5295 PR rtl-optimization/64682
5296 PR rtl-optimization/69567
5297 PR rtl-optimization/69737
5298 PR rtl-optimization/82683
5299 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
5300 register mentioned in the note, drop the note, unless it came from I3,
5301 in which case it should go to I3 again.
5303 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5305 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
5306 and return false if not.
5307 (clear_bytes_written_by, live_bytes_read): Update accordingly.
5309 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5311 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
5312 range is known to be empty.
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 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
5319 and CONST_VECTOR_NUNITS instead of computing the number of units from
5320 the byte sizes of the vector and element.
5321 (simplify_binary_operation_1): Likewise.
5322 (simplify_const_binary_operation): Likewise.
5323 (simplify_ternary_operation): Likewise.
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 * var-tracking.c (INT_MEM_OFFSET): Replace with...
5330 (int_mem_offset): ...this new function.
5331 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
5332 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
5333 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
5336 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5337 Alan Hayward <alan.hayward@arm.com>
5338 David Sherwood <david.sherwood@arm.com>
5340 * lower-subreg.c (interesting_mode_p): New function.
5341 (compute_costs, find_decomposable_subregs, decompose_register)
5342 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
5343 (resolve_clobber, dump_choices): Use it.
5345 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5346 Alan Hayward <alan.hayward@arm.com>
5347 David Sherwood <david.sherwood@arm.com>
5349 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
5351 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5352 Alan Hayward <alan.hayward@arm.com>
5353 David Sherwood <david.sherwood@arm.com>
5355 * alias.c (find_base_value, find_base_term): Only process integer
5356 truncations. Check the precision rather than the size.
5358 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5359 Alan Hayward <alan.hayward@arm.com>
5360 David Sherwood <david.sherwood@arm.com>
5362 * machmode.h (is_narrower_int_mode): New function
5363 * optabs.c (expand_float, expand_fix): Use it.
5364 * dwarf2out.c (rotate_loc_descriptor): Likewise.
5366 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5367 Alan Hayward <alan.hayward@arm.com>
5368 David Sherwood <david.sherwood@arm.com>
5370 * rtl.h (narrower_subreg_mode): New function.
5371 * ira-color.c (update_costs_from_allocno): Use it.
5373 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5374 Alan Hayward <alan.hayward@arm.com>
5375 David Sherwood <david.sherwood@arm.com>
5377 * optabs-query.h (convert_optab_p): New function, split out from...
5378 (convert_optab_handler): ...here.
5379 (widening_optab_handler): Delete.
5380 (find_widening_optab_handler): Remove permit_non_widening parameter.
5381 (find_widening_optab_handler_and_mode): Likewise. Provide an
5382 override that operates on mode class wrappers.
5383 * optabs-query.c (widening_optab_handler): Delete.
5384 (find_widening_optab_handler_and_mode): Remove permit_non_widening
5385 parameter. Assert that the two modes are the same class and that
5386 the "from" mode is narrower than the "to" mode. Use
5387 convert_optab_handler instead of widening_optab_handler.
5388 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
5389 instead of widening_optab_handler.
5390 * expr.c (expand_expr_real_2): Update calls to
5391 find_widening_optab_handler.
5392 * optabs.c (expand_widen_pattern_expr): Likewise.
5393 (expand_binop_directly): Take the insn_code as a parameter.
5394 (expand_binop): Only call find_widening_optab_handler for
5395 conversion optabs; use optab_handler otherwise. Update calls
5396 to find_widening_optab_handler and expand_binop_directly.
5397 Use convert_optab_handler instead of widening_optab_handler.
5398 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
5399 find_widening_optab_handler and use scalar_mode rather than
5401 (convert_plusminus_to_widen): Likewise.
5403 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5404 Alan Hayward <alan.hayward@arm.com>
5405 David Sherwood <david.sherwood@arm.com>
5407 * machmode.h (fixed_size_mode): New class.
5408 * rtl.h (get_pool_mode): Return fixed_size_mode.
5409 * gengtype.c (main): Add fixed_size_mode.
5410 * target.def (get_raw_result_mode): Return a fixed_size_mode.
5411 (get_raw_arg_mode): Likewise.
5412 * doc/tm.texi: Regenerate.
5413 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
5414 * targhooks.c (default_get_reg_raw_mode): Likewise.
5415 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
5416 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
5417 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
5418 (msp430_get_raw_result_mode): Likewise.
5419 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
5420 * dbxout.c (dbxout_parms): Require fixed-size modes.
5421 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
5422 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5423 * omp-low.c (lower_oacc_reductions): Likewise.
5424 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
5425 (simplify_subreg): Update accordingly.
5426 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
5427 (force_const_mem): Update accordingly. Return NULL_RTX for modes
5428 that aren't fixed-size.
5429 (get_pool_mode): Return a fixed_size_mode.
5430 (output_constant_pool_2): Take a fixed_size_mode.
5432 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5433 Alan Hayward <alan.hayward@arm.com>
5434 David Sherwood <david.sherwood@arm.com>
5436 * doc/rtl.texi (vec_series): Document.
5437 (const): Say that the operand can be a vec_series.
5438 * rtl.def (VEC_SERIES): New rtx code.
5439 * rtl.h (const_vec_series_p_1): Declare.
5440 (const_vec_series_p): New function.
5441 * emit-rtl.h (gen_const_vec_series): Declare.
5442 (gen_vec_series): Likewise.
5443 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
5444 (gen_vec_series): Likewise.
5445 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
5446 * simplify-rtx.c (simplify_unary_operation): Handle negations
5448 (simplify_binary_operation_series): New function.
5449 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
5450 (test_vector_ops_series): New function.
5451 (test_vector_ops): Call it.
5452 * config/powerpcspe/altivec.md (altivec_lvsl): Use
5453 gen_const_vec_series.
5454 (altivec_lvsr): Likewise.
5455 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
5457 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5458 Alan Hayward <alan.hayward@arm.com>
5459 David Sherwood <david.sherwood@arm.com>
5461 * doc/rtl.texi (const): Update description of address constants.
5462 Say that vector constants are allowed too.
5463 * common.md (E, F): Use CONSTANT_P instead of checking for
5465 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
5466 checking for CONST_VECTOR.
5467 * expmed.c (make_tree): Use build_vector_from_val for a CONST
5469 * expr.c (expand_expr_real_2): Check for vector modes instead
5470 of checking for CONST_VECTOR.
5471 * rtl.h (const_vec_p): New function.
5472 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
5473 (unwrap_const_vec_duplicate): Handle them here too.
5475 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5476 David Malcolm <dmalcolm@redhat.com>
5477 Alan Hayward <alan.hayward@arm.com>
5478 David Sherwood <david.sherwood@arm.com>
5480 * rtl.h (vec_duplicate_p): New function.
5481 * selftest-rtl.c (assert_rtx_eq_at): New function.
5482 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
5483 (assert_rtx_eq_at): Declare.
5484 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
5485 * selftest-run-tests.c (selftest::run_tests): Call it.
5486 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
5487 (simplify_unary_operation_1): Recursively handle vector duplicates.
5488 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
5490 (simplify_subreg): Handle subregs of vector duplicates.
5491 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
5492 (selftest::simplify_rtx_c_tests): New functions.
5494 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5495 Alan Hayward <alan.hayward@arm.com>
5496 David Sherwood <david.sherwood@arm.com>
5498 * emit-rtl.h (gen_const_vec_duplicate): Declare.
5499 (gen_vec_duplicate): Likewise.
5500 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
5502 (gen_const_vector): ...here.
5503 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
5504 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
5505 whose elements are all equal.
5506 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
5507 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
5508 (simplify_relational_operation): Likewise.
5509 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
5511 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
5512 (aarch64_expand_vector_init): Likewise.
5513 * config/arm/arm.c (neon_vdup_constant): Likewise.
5514 (neon_expand_vector_init): Likewise.
5515 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
5516 (arm_block_set_unaligned_vect): Likewise.
5517 (arm_block_set_aligned_vect): Likewise.
5518 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
5519 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
5520 (expand_vec_perm_even_odd_pack): Likewise.
5521 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
5522 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
5523 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
5524 gen_const_vec_duplicate.
5525 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
5526 * config/mips/mips.c (mips_gen_const_int_vector): Use
5527 gen_const_vec_duplicate.
5528 (mips_expand_vector_init): Use CONST0_RTX.
5529 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
5530 (define_split): Use gen_const_vec_duplicate.
5531 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
5532 (define_split): Use gen_const_vec_duplicate.
5533 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
5534 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
5535 * config/spu/spu.c (spu_const): Likewise.
5537 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5538 Alan Hayward <alan.hayward@arm.com>
5539 David Sherwood <david.sherwood@arm.com>
5541 * combine.c (can_change_dest_mode): Reject changes in
5542 REGMODE_NATURAL_SIZE.
5544 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
5546 * configure.ac (--enable-libssp): New.
5547 (gcc_cv_libc_provides_ssp): Check for explicit setting before
5548 trying to determine target-specific default. Adjust indentation.
5549 * configure: Regenerated.
5550 * doc/install.texi (Configuration): Expand --disable-libssp
5553 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
5555 config/i386/i386.c (ix86_expand_epilogue): Correct stack
5558 2017-10-31 Martin Jambor <mjambor@suse.cz>
5561 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
5563 2017-10-31 David Malcolm <dmalcolm@redhat.com>
5565 * auto-profile.c (autofdo_source_profile::read): Use
5566 UNKNOWN_LOCATION rather than 0.
5567 * diagnostic-core.h (warning_at_rich_loc): Rename to...
5568 (warning_at): ...this overload.
5569 (warning_at_rich_loc_n): Rename to...
5570 (warning_n): ...this overload.
5571 (error_at_rich_loc): Rename to...
5572 (error_at): ...this overload.
5573 (pedwarn_at_rich_loc): Rename to...
5574 (pedwarn): ...this overload.
5575 (permerror_at_rich_loc): Rename to...
5576 (permerror): ...this overload.
5577 (inform_at_rich_loc): Rename to...
5578 (inform): ...this overload.
5579 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
5580 (diagnostic_n_impl_richloc): Rename to...
5581 (diagnostic_n_impl): ...this rich_location *-based decl.
5582 (inform_at_rich_loc): Rename to...
5583 (inform): ...this, and add an assertion.
5584 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
5585 (warning_at_rich_loc): Rename to...
5586 (warning_at): ...this, and add an assertion.
5587 (warning_at_rich_loc_n): Rename to...
5588 (warning_n): ...this, and add an assertion.
5589 (warning_n): Update location_t-based implementation for removal of
5590 location_t-based diagnostic_n_impl.
5591 (pedwarn_at_rich_loc): Rename to...
5592 (pedwarn): ...this, and add an assertion.
5593 (permerror_at_rich_loc): Rename to...
5594 (permerror): ...this, and add an assertion.
5595 (error_n): Update for removal of location_t-based diagnostic_n_impl.
5596 (error_at_rich_loc): Rename to...
5597 (error_at): ...this, and add an assertion.
5598 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
5599 (driver::do_spec_on_infiles): Likewise.
5600 * substring-locations.c (format_warning_va): Update for renaming
5601 of inform_at_rich_loc.
5603 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
5605 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
5606 _Float<N>X built-in functions so that the variant without the
5607 "__builtin_" prefix is only enabled for the GNU C and Objective C
5608 languages when they are in non-strict ANSI/ISO mode.
5609 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
5610 * target.def (floatn_builtin_p): Add a target hook to control
5611 whether _Float<N> and _Float<N>X built-in functions without the
5612 "__builtin_" prefix are enabled, and return true for C and
5613 Objective C in the default hook. Include langhooks.h in
5615 * targhooks.h (default_floatn_builtin_p): Likewise.
5616 * targhooks.c (default_floatn_builtin_p): Likewise.
5617 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
5618 floatn_builtin_p target hook.
5619 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
5621 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
5622 Eric Botcazou <ebotcazou@adacore.com>
5624 PR rtl-optimization/81803
5625 * lra-constraints.c (curr_insn_transform): Also reload the whole
5626 register for a strict subreg no wider than a word if this is for
5627 a WORD_REGISTER_OPERATIONS target.
5629 2017-10-31 Jason Merrill <jason@redhat.com>
5631 * gdbinit.in: Skip over inlines from timevar.h.
5633 2017-10-31 Martin Liska <mliska@suse.cz>
5635 * doc/gcov.texi: Document new option.
5636 * gcov.c (print_usage): Likewise print it.
5637 (process_args): Support the argument.
5638 (format_count): New function.
5639 (format_gcov): Use the function.
5641 2017-10-31 Martin Liska <mliska@suse.cz>
5643 * gcov.c (struct name_map): do not use typedef.
5644 Define operator== and operator<.
5645 (name_search): Remove.
5646 (name_sort): Remove.
5647 (main): Do not allocate names.
5648 (process_file): Add vertical space.
5649 (generate_results): Use std::find.
5650 (release_structures): Do not release memory.
5651 (find_source): Use std::find.
5653 2017-10-31 Martin Liska <mliska@suse.cz>
5655 * gcov.c (struct line_info): Remove it's typedef.
5656 (line_info::line_info): Add proper ctor.
5657 (line_info::has_block): Do not use a typedef.
5658 (struct source_info): Do not use typedef.
5659 (circuit): Likewise.
5660 (get_cycles_count): Likewise.
5661 (output_intermediate_file): Iterate via vector iterator.
5662 (add_line_counts): Use std::vector methods.
5663 (accumulate_line_counts): Likewise.
5664 (output_lines): Likewise.
5666 2017-10-31 Martin Liska <mliska@suse.cz>
5668 * gcov.c (struct source_info): Remove typedef.
5669 (source_info::source_info): Add proper ctor.
5670 (accumulate_line_counts): Use struct, not it's typedef.
5671 (output_gcov_file): Likewise.
5672 (output_lines): Likewise.
5673 (main): Do not allocate an array.
5674 (output_intermediate_file): Use size of vector container.
5675 (process_file): Resize the vector.
5676 (generate_results): Do not preallocate, use newly added vector
5678 (release_structures): Do not release sources.
5679 (find_source): Use vector methods.
5680 (add_line_counts): Do not use typedef.
5682 2017-10-31 Martin Liska <mliska@suse.cz>
5684 * doc/gcov.texi: Document that.
5685 * gcov.c (add_line_counts): Mark lines with a non-executed
5687 (output_line_beginning): Handle such lines.
5688 (output_lines): Pass new argument.
5689 (output_intermediate_file): Print it in intermediate format.
5691 2017-10-31 Martin Liska <mliska@suse.cz>
5693 * color-macros.h: New file.
5694 * diagnostic-color.c: Factor out color related to macros to
5696 * doc/gcov.texi: Document -k option.
5697 * gcov.c (INCLUDE_STRING): Include string.h.
5698 (print_usage): Add -k option.
5699 (process_args): Parse it.
5700 (pad_count_string): New function.
5701 (output_line_beginning): Likewise.
5702 (DEFAULT_LINE_START): New macro.
5703 (output_lines): Support color output.
5705 2017-10-31 Martin Liska <mliska@suse.cz>
5707 PR gcov-profile/82633
5708 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
5709 their interaction with GCOV infrastructure.
5710 * configure.ac: Add -fkeep-{inline,static}-functions to
5712 * configure: Regenerate.
5714 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
5717 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
5719 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
5722 * config/rs6000/rs6000.md (allocate_stack): Force update interval
5723 into a register if it does not fit into an immediate offset field.
5725 2017-10-31 Olivier Hainque <hainque@adacore.com>
5727 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
5729 2017-10-31 Julia Koval <julia.koval@intel.com>
5731 * config.gcc: Add gfniintrin.h.
5732 * config/i386/gfniintrin.h: New.
5733 * config/i386/i386-builtin-types.def
5734 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5735 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5736 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5737 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5738 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5739 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
5740 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
5741 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
5742 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
5743 V64QI_FTYPE_V64QI_V64QI_INT): New types.
5744 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5745 * config/i386/immintrin.h: Include gfniintrin.h.
5746 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
5748 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
5750 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
5752 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5754 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
5755 (arm_ashldi3_1bit): Remove pattern.
5756 (ashrdi3): Remove shift by 1 expansion.
5757 (arm_ashrdi3_1bit): Remove pattern.
5758 (lshrdi3): Remove shift by 1 expansion.
5759 (arm_lshrdi3_1bit): Remove pattern.
5760 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
5761 cost of ashldi3 by 1.
5762 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
5763 (<shift>di3_neon): Likewise.
5765 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5767 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
5768 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
5769 and (*aarch64_simd_mov<VQ:mode>).
5770 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
5771 pattern alternative.
5772 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
5773 attributes to match pattern alternative.
5775 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
5777 * config.gcc (powerpc*-*-*): Add emmintrin.h.
5778 * config/rs6000/emmintrin.h: New file.
5779 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
5781 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5783 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
5784 * (movdi_vfp_cortexa8): Remove pattern.
5786 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5788 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
5790 (Specific, alpha*-dec-osf5.1): Remove.
5791 (Specific, mips-sgi-irix5): Remove.
5792 (Specific, mips-sgi-irix6): Remove.
5794 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5797 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
5798 arguments to clear_bit_region_be.
5800 2017-10-30 Jim Wilson <wilson@tuliptree.org>
5802 * gimplify.c: Include memmodel.h.
5804 2017-10-30 Martin Jambor <mjambor@suse.cz>
5806 * omp-grid.c (grid_attempt_target_gridification): Also insert a
5807 condition whether loop should be executed at all.
5809 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
5811 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5812 gimple folding of vec_madd() intrinsics.
5813 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
5814 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
5815 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
5817 2017-10-30 Richard Biener <rguenther@suse.de>
5819 PR tree-optimization/82762
5821 2017-10-23 Richard Biener <rguenther@suse.de>
5823 PR tree-optimization/82129
5825 2017-08-01 Richard Biener <rguenther@suse.de>
5827 PR tree-optimization/81181
5828 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5829 (compute_antic): ... end of iteration here.
5831 2017-10-30 Joseph Myers <joseph@codesourcery.com>
5833 * doc/invoke.texi (C Dialect Options): Document -std=c17,
5834 -std=iso9899:2017 and -std=gnu17.
5835 * doc/standards.texi (C Language): Document C17 support.
5836 * doc/cpp.texi (Overview): Mention -std=c17.
5837 (Standard Predefined Macros): Document C11 and C17 values of
5838 __STDC_VERSION__. Do not refer to C99 support as incomplete.
5839 * doc/extend.texi (Inline): Do not list individual options for
5840 standards newer than C99.
5841 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
5843 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
5846 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
5848 * asan.c (asan_finish_file): Align asan globals array by shadow
5851 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5854 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
5855 (struct store_immediate_info): Add bitregion_start and bitregion_end
5857 (store_immediate_info::store_immediate_info): Add brs and bre
5858 arguments and initialize bitregion_{start,end} from those.
5859 (struct merged_store_group): Add bitregion_start, bitregion_end,
5860 align_base and mask fields. Drop unnecessary struct keyword from
5861 struct store_immediate_info. Add do_merge method.
5862 (clear_bit_region_be): Use memset instead of loop storing zeros.
5863 (merged_store_group::do_merge): New method.
5864 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
5865 stores as long as the surrounding bitregions have no gaps.
5866 (merged_store_group::merge_overlapping): Use do_merge.
5867 (merged_store_group::apply_stores): Test that bitregion_{start,end}
5868 is byte aligned, rather than requiring that start and width are
5869 byte aligned. Drop unnecessary struct keyword from
5870 struct store_immediate_info. Allocate and populate also mask array.
5871 Make start of the arrays relative to bitregion_start rather than
5872 start and size them according to bitregion_{end,start} difference.
5873 (struct imm_store_chain_info): Drop unnecessary struct keyword from
5874 struct store_immediate_info.
5875 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
5876 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
5877 struct keyword from struct store_immediate_info.
5878 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
5879 between stores as long as the surrounding bitregions have no gaps.
5881 (struct split_store): Add orig non-static data member.
5882 (split_store::split_store): Initialize orig to false.
5883 (find_constituent_stmts): Return store_immediate_info *, non-NULL
5884 if there is exactly a single original stmt. Change stmts argument
5885 to pointer from reference, if NULL, don't push anything to it. Add
5886 first argument, use it to optimize skipping over orig stmts that
5887 are known to be before bitpos already. Simplify.
5888 (split_group): Return unsigned int count how many stores are or
5889 would be needed rather than a bool. Add allow_unaligned argument.
5890 Change split_stores argument from reference to pointer, if NULL,
5891 only do a dry run computing how many stores would be produced.
5892 Rewritten algorithm to use both alignment and misalign if
5893 !allow_unaligned and handle bitfield stores with gaps.
5894 (imm_store_chain_info::output_merged_store): Set start_byte_pos
5895 from bitregion_start instead of start. Compute allow_unaligned
5896 here, if true, do 2 split_group dry runs to compute which one
5897 produces fewer stores and prefer aligned if equal. Punt if
5898 new count is bigger or equal than original before emitting any
5899 statements, rather than during that. Remove no longer needed
5900 new_ssa_names tracking. Replace num_stmts with
5901 split_stores.length (). Use 32-bit stack allocated entries
5902 in split_stores auto_vec. Try to reuse original store lhs/rhs1
5903 if possible. Handle bitfields with gaps.
5904 (pass_store_merging::execute): Ignore bitsize == 0 stores.
5905 Compute bitregion_{start,end} for the stores and construct
5906 store_immediate_info with that. Formatting fixes.
5908 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
5911 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
5912 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
5914 2017-10-29 Jim Wilson <wilson@tuliptree.org>
5916 * gimplify.c: Include tm_p.h.
5918 * common.opt (gcoff): Re-add as ignored option.
5919 (gcoff1, gcoff2, gcoff3): Likewise.
5921 * Makefile.in (OBJS): Delete sdbout.o.
5922 (GTFILES): Delete $(srcdir)/sdbout.c.
5923 * debug.h: Delete sdb_debug_hooks.
5924 * final.c: Delete sdbout.h include.
5925 (final_scan_insn): Delete SDB_DEBUG check.
5926 (rest_of_clean_state): Likewise.
5927 * output.h: Delete sdb_begin_function_line.
5930 * toplev.c: Delete sdbout.h include.
5931 (process_options): Delete SDB_DEBUG check.
5932 * tree-core.h (tree_type_common): Delete pointer field of
5934 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
5935 TYPE_SYMTAB_POINTER.
5936 * tree.h (TYPE_SYMTAB_POINTER): Delete.
5937 (TYPE_SYMTAB_IS_POINTER): Delete.
5938 (TYPE_SYMTAB_IS_DIE): Renumber.
5939 * xcoffout.c: Refer to former sdbout.c file.
5940 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
5942 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
5943 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
5944 (Debugging Options): Delete -gcoff.
5945 (-gstabs): Delete SDB reference.
5947 (-gcoff@var{level}): Delete.
5948 * doc/passes.texi (Debugging information output): Delete SDB and
5949 sdbout.c references.
5950 * doc/tm.texi: Regenerate.
5951 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
5952 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
5954 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
5955 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
5956 (SDB_DEBUGGING_INFO): Delete.
5957 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
5958 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
5959 * target.def (output_source_filename): Delete COFF reference.
5961 * common.opt (gcoff): Delete.
5962 (gxcoff+): Update Negative chain.
5963 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
5965 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
5966 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
5967 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
5969 (expand_function_start): Change sdb reference to past tense.
5970 (expand_function_end): Change sdb reference to past tense.
5971 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
5972 * opts.c (debug_type_names): Delete coff entry.
5973 (common_handle_option): Delete OPT_gcoff case.
5974 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
5976 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
5977 * config/cris/cris.h: Delete SDB reference in comment.
5978 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
5979 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
5980 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
5981 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
5983 (ix86_expand_prologue): Likewise.
5984 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
5985 * config/ia64/ia64.h: Likewise.
5986 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
5987 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
5989 * config/mmix/mmix.h: Likewise.
5990 * config/nds32/nds32.c: Likewise.
5991 * config/stormy/storym16.h: Likewise.
5992 * config/visium/visium.h: Likewise.
5993 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
5995 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
5997 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
5998 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
5999 FRAME_POINTER_REGNUM point at high end of local var area.
6001 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
6003 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
6004 Move comment around. Do not reset best_edge for a copiable
6005 destination if the copy would cause a partition change.
6006 (better_edge_p): Remove redundant check.
6008 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6010 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
6012 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
6014 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
6015 for math functions that have _Float<N> and _Float<N>X variants.
6016 (mathfn_built_in_2): Add support for math functions that have
6017 _Float<N> and _Float<N>X variants.
6018 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
6019 (expand_builtin_mathfn_ternary): Add support for fma with
6020 _Float<N> and _Float<N>X variants.
6021 (expand_builtin): Likewise.
6022 (fold_builtin_3): Likewise.
6023 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
6024 create math function _Float<N> and _Float<N>X variants as external
6026 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
6027 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
6028 the __builtin_ prefix and if not strict ansi, without the prefix.
6029 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
6030 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
6031 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
6032 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
6033 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
6034 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
6035 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
6036 function signatures for fma _Float<N> and _Float<N>X variants.
6037 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
6038 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
6039 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
6040 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
6041 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
6042 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
6043 * gencfn-macros.c (print_case_cfn): Add support for math functions
6044 that have _Float<N> and _Float<N>X variants.
6045 (print_define_operator_list): Likewise.
6046 (fltfn_suffixes): Likewise.
6048 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
6049 for math functions that have _Float<N> and _Float<N>X variants.
6050 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
6051 and _Float<N>X variants.
6052 (COPYSIGN): Likewise.
6055 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6056 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6058 (integer_valued_read_call_p): Likewise.
6059 * fold-const-call.c (fold_const_call_ss): Likewise.
6060 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6061 _Float<N> and _Float<N>X variants.
6062 (fold_const_call_ssss): Add support for fma _Float<N> and
6063 _Float<N>X variants.
6064 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6065 support for copysign and fma _Float<N> and _Float<N>X variants.
6066 (backprop::process_builtin_call_use): Likewise.
6067 * tree-call-cdce.c (can_test_argument_range); Add support for
6068 sqrt _Float<N> and _Float<N>X variants.
6069 (edom_only_function): Likewise.
6070 (get_no_error_domain): Likewise.
6071 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6072 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6073 copysign _Float<N> and _Float<N>X variants.
6074 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6075 handled by machine independent code.
6076 (FMAF128): Likewise.
6077 * doc/cpp.texi (Common Predefined Macros): Document defining
6078 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6079 fma _Float<N> and _Float<N>X variants.
6081 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6084 * config/i386/i386-modes.def (CCFPU): Remove definition.
6085 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6086 (ix86_cc_modes_compatible): Ditto.
6087 (ix86_expand_carry_flag_compare): Ditto.
6088 (ix86_expand_int_movcc): Ditto.
6089 (ix86_expand_int_addcc): Ditto.
6090 (ix86_reverse_condition): Ditto.
6091 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6092 Return true/false for unordered/ordered fp comparisons.
6093 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6094 (ix86_prepare_fp_compare_args): Update for rename.
6095 (ix86_expand_fp_compare): Update for rename. Generate unordered
6096 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6097 (ix86_expand_sse_compare_and_jump): Ditto.
6098 * config/i386/predicates.md (fcmov_comparison_operator):
6099 Remove CCFPU mode handling.
6100 (ix86_comparison_operator): Ditto.
6101 (ix86_carry_flag_operator): Ditto.
6102 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6103 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6104 (*cmpu<mode>_cc_i387): Ditto.
6105 (FPCMP): Remove mode iterator.
6106 (unord): Remove mode attribute.
6107 (unord_subst): New define_subst transformation
6108 (unord): New define_subst attribute.
6110 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6111 (*cmpi<unord>xf_i387): Ditto.
6112 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6113 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
6114 using unord_subst transformation.
6115 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
6116 (round_saeonly): Also handle CCFP mode.
6117 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
6118 Remove UNSPEC_SAHF unspec handling.
6120 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6122 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
6124 2017-10-27 Jeff Law <law@redhat.com>
6126 * gimple-ssa-sprintf.c: Include domwalk.h.
6127 (class sprintf_dom_walker): New class, derived from dom_walker.
6128 (sprintf_dom_walker::before_dom_children): New function.
6129 (struct call_info): Moved into sprintf_dom_walker class
6130 (compute_formath_length, handle_gimple_call): Likewise.
6131 (sprintf_length::execute): Call the dominator walker rather
6132 than walking the statements.
6134 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
6135 gimple statement locations.
6136 (check_array_bounds): Corresponding changes. Get the statement's
6137 location directly from wi->stmt.
6139 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
6142 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
6144 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6146 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
6147 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
6149 2017-10-27 Jakub Jelinek <jakub@redhat.com>
6152 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
6153 * config/i386/i386.c (maybe_get_pool_constant): Removed.
6154 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
6155 maybe_get_pool_constant.
6156 * config/i386/predicates.md (zero_extended_scalar_load_operand):
6159 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6161 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
6162 2.26 caveat. Update gas and gld versions.
6163 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
6166 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6168 * cgraph.h (set_malloc_flag): Declare.
6169 * cgraph.c (set_malloc_flag_1): New function.
6170 (set_malloc_flag): Likewise.
6171 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
6172 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
6174 (read_ipa_call_summary): Add support for reading is_return_callee.
6175 (write_ipa_call_summary): Stream is_return_callee.
6176 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
6177 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
6178 ipa-prop.h, ipa-fnsummary.h.
6179 (pure_const_names): Change to static.
6180 (malloc_state_e): Define.
6181 (malloc_state_names): Define.
6182 (funct_state_d): Add field malloc_state.
6183 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
6184 (check_retval_uses): New function.
6185 (malloc_candidate_p): Likewise.
6186 (analyze_function): Add support for malloc attribute.
6187 (pure_const_write_summary): Stream malloc_state.
6188 (pure_const_read_summary): Add support for reading malloc_state.
6189 (dump_malloc_lattice): New function.
6190 (propagate_malloc): New function.
6191 (warn_function_malloc): New function.
6192 (ipa_pure_const::execute): Call propagate_malloc and
6193 ipa_free_fn_summary.
6194 (pass_local_pure_const::execute): Add support for malloc attribute.
6195 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
6196 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
6198 2017-10-27 Martin Liska <mliska@suse.cz>
6200 PR gcov-profile/82457
6201 * doc/invoke.texi: Document that one needs a non-strict ISO mode
6202 for fork-like functions to be properly instrumented.
6204 2017-10-27 Richard Biener <rguenther@suse.de>
6207 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
6208 info when we redirected EH.
6210 2017-10-26 Michael Collison <michael.collison@arm.com>
6212 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
6214 (<optab>_trunchf<GPI:mode>2: New pattern.
6215 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
6216 * config/aarch64/iterators.md (wv): New mode attribute.
6217 (vf, VF): New mode attributes.
6218 (vgp, VGP): New mode attributes.
6219 (s): Update attribute with SImode and DImode prefixes.
6221 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6223 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
6224 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
6225 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
6226 (nios2_option_overide): Initialize it. Don't allow R0-relative
6227 addressing with PIC.
6228 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
6229 (nios2_symbolic_constant_p): Likewise.
6230 (nios2_legitimate_address_p): Likewise.
6231 (nios2_r0rel_section_name_p): New.
6232 (nios2_symbol_ref_in_r0rel_data_p): New.
6233 (nios2_emit_move_sequence): Handle r0rel_constant_p.
6234 (r0rel_constant_p): New.
6235 (nios2_print_operand_address): Handle r0rel_constant_p.
6236 (nios2_cdx_narrow_form_p): Likewise.
6237 * config/nios2/nios2.opt (mr0rel-sec=): New option.
6238 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
6239 (Nios II Options): Document -mr0rel-sec.
6241 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6243 * config/nios2/nios2.c: Include xregex.h.
6244 (nios2_gprel_sec_regex): New.
6245 (nios2_option_overide): Initialize it. Don't allow GP-relative
6246 addressing with PIC.
6247 (nios2_small_section_name_p): Check for regex match.
6248 * config/nios2/nios2.opt (mgprel-sec=): New option.
6249 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
6250 (Nios II Options): Document -mgprel-sec.
6252 2017-10-26 Jim Wilson <wilson@tuliptree.org>
6254 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
6256 2017-10-26 Tom de Vries <tom@codesourcery.com>
6258 PR tree-optimization/82707
6259 * gimple.c (gimple_copy): Fix unsharing of
6260 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
6262 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
6264 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
6265 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
6266 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
6267 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
6268 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
6269 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
6270 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
6271 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
6272 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
6273 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
6274 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
6275 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
6276 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
6277 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
6278 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
6279 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
6280 _mm512_mask_cmpunord_ps_mask): New intrinsics.
6282 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6284 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
6286 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
6287 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
6288 warning to rs6000.c. Remove the Undocumented flag, since it has
6290 (-mabi=ibmlongdouble): Likewise.
6291 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
6292 already set, set the default format for long double.
6293 (rs6000_debug_reg_global): Print whether long double is IBM or
6295 (rs6000_option_override_internal): Rework setting long double
6296 format. Only warn if the user is changing the long double default
6297 and they did not use -Wno-psabi.
6298 * doc/invoke.texi (PowerPC options): Update the documentation for
6299 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
6301 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6302 Alan Hayward <alan.hayward@arm.com>
6303 David Sherwood <david.sherwood@arm.com>
6305 * rtl.h (wider_subreg_mode): New function.
6306 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
6307 rather than an unsigned int *.
6308 * ira-color.c (regno_max_ref_width): Replace with...
6309 (regno_max_ref_mode): ...this new variable.
6310 (coalesced_pseudo_reg_slot_compare): Update accordingly.
6311 Use wider_subreg_mode.
6312 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
6313 rather than an unsigned int *.
6314 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
6315 (process_alt_operands): Likewise.
6316 (invariant_p): Likewise.
6317 * lra-spills.c (assign_mem_slot): Likewise.
6318 (add_pseudo_to_slot): Likewise.
6319 * lra.c (collect_non_operand_hard_regs): Likewise.
6320 (add_regs_to_insn_regno_info): Likewise.
6321 * reload1.c (regno_max_ref_width): Replace with...
6322 (regno_max_ref_mode): ...this new variable.
6323 (reload): Update accordingly. Update call to
6324 ira_sort_regnos_for_alter_reg.
6325 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
6326 (init_eliminable_invariants): Update to use regno_max_ref_mode.
6327 (scan_paradoxical_subregs): Likewise.
6329 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6331 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
6332 (aarch64_frame): Add emit_frame_chain boolean.
6333 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6334 Move eh_return case to aarch64_layout_frame.
6335 (aarch64_layout_frame): Initialize emit_frame_chain.
6336 (aarch64_expand_prologue): Use emit_frame_chain.
6338 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6340 * config/aarch64/aarch64.c (aarch64_layout_frame):
6341 Ensure LR is always stored at the bottom of the callee-saves.
6342 Remove rarely used frame layout which saves callee-saves at top of
6343 frame, so the store of LR can be used as a valid probe in all cases.
6345 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6347 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
6348 Improve unaligned TImode/TFmode base/offset split.
6350 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6351 Alan Hayward <alan.hayward@arm.com>
6352 David Sherwood <david.sherwood@arm.com>
6354 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
6355 * combine.c (find_single_use_1): Likewise.
6356 (expand_field_assignment): Likewise.
6357 (move_deaths): Likewise.
6358 * lra-constraints.c (simplify_operand_subreg): Likewise.
6359 (curr_insn_transform): Likewise.
6360 * lra.c (collect_non_operand_hard_regs): Likewise.
6361 (add_regs_to_insn_regno_info): Likewise.
6362 * rtlanal.c (reg_referenced_p): Likewise.
6363 (covers_regno_no_parallel_p): Likewise.
6365 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6367 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
6368 Don't print any bits outside the precision of the value.
6369 * wide-int.cc (test_printing): Add some new tests.
6371 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6373 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
6374 supports -xbrace_comment option.
6375 * configure: Regenerate.
6376 * config.in: Regenerate.
6377 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
6378 (ASM_CPU_SPEC): Use it.
6380 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6382 * target.def (static_rtx_alignment): New hook.
6383 * targhooks.h (default_static_rtx_alignment): Declare.
6384 * targhooks.c (default_static_rtx_alignment): New function.
6385 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
6386 * doc/tm.texi: Regenerate.
6387 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
6388 instead of targetm.constant_alignment. Remove call to
6390 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6391 (cris_preferred_mininum_alignment): New function, split out from...
6392 (cris_constant_alignment): ...here.
6393 (cris_static_rtx_alignment): New function.
6394 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
6396 (ix86_constant_alignment): ...here.
6397 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6398 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6399 (mmix_static_rtx_alignment): New function.
6400 * config/spu/spu.c (spu_static_rtx_alignment): New function.
6401 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6403 2017-10-26 Tamar Christina <tamar.christina@arm.com>
6406 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
6407 Add flag_trapping_math and flag_fp_int_builtin_inexact.
6409 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
6411 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
6412 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
6413 mark as a sign-extending load.
6414 (local_pic_load_u): Define.
6416 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
6419 * fold-const.c (operand_equal_for_comparison_p): Also return true
6420 if ARG0 is a simple variant of ARG1 with narrower precision.
6421 (fold_ternary_loc): Always pass unstripped operands to the predicate.
6423 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
6425 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
6427 * i386.h (processor_costs): Add gather_static, gather_per_elt,
6428 scatter_static, scatter_per_elt.
6429 * x86-tune-costs.h: Add new cost entries.
6431 2017-10-25 Richard Biener <rguenther@suse.de>
6433 * tree-ssa-sccvn.h (vn_eliminate): Declare.
6434 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
6435 class pass_fre): Move to ...
6436 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
6437 class pass_fre): ... here and adjust for statistics.
6439 2017-10-25 Jakub Jelinek <jakub@redhat.com>
6442 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
6443 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
6444 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
6445 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
6448 2017-10-25 Richard Biener <rguenther@suse.de>
6450 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
6451 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
6452 eliminate_push_avail, eliminate_insert): Move inside...
6453 (class eliminate_dom_walker): ... this class in preparation
6455 (fini_eliminate): Remove by merging with ...
6456 (eliminate): ... this function. Adjust for class changes.
6457 (pass_pre::execute): Remove fini_eliminate call.
6458 (pass_fre::execute): Likewise.
6460 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6463 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
6464 (VPERMI2, VPERMI2I): New mode iterators.
6465 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
6466 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
6468 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
6469 mode iterator. Remove 3 old define_insn patterns.
6470 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
6471 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
6472 VPERMI2 mode iterator, remove the other two expanders.
6473 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
6474 to use VPERMI2 mode iterator, add another alternative for vpermi2*
6475 instructions, remove the other two patterns.
6476 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
6477 mode iterator, remove the other two patterns.
6478 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
6479 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
6480 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
6481 and adjust argument order accordingly.
6482 (ix86_expand_vec_perm): Adjust caller.
6483 (expand_vec_perm_1): Likewise.
6484 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
6485 (expand_vec_perm_vpermt2_vpshub2): ... this.
6486 (ix86_expand_vec_perm_const_1): Adjust caller.
6487 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
6490 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
6491 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
6492 (vec_shl_<mode>): Remove unused expander.
6493 (avx512bw_<shift_insn><mode>3): New define_insn.
6494 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
6495 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
6497 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
6500 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
6503 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6505 PR rtl-optimization/82396
6506 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
6507 (autopref_multipass_init): Simplify initialization.
6508 (autopref_rank_data): Simplify sort order.
6509 * gcc/sched-int.h (autopref_multipass_data_): Remove
6510 multi_mem_insn_p, min_offset and max_offset.
6512 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6515 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6516 Check special value of flag_omit_frame_pointer.
6517 (aarch64_can_eliminate): Likewise.
6518 (aarch64_override_options_after_change_1): Simplify handling of
6519 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
6521 2017-10-24 Richard Biener <rguenther@suse.de>
6523 PR tree-optimization/82697
6524 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
6525 zero for conditional load and unconditional store.
6527 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6529 * doc/install.texi: Document bootstrap-cet.
6531 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6534 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
6535 ENDBR instruction at function entrance if function is only
6538 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6541 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
6542 patterns to better describe from which operation the CF is computed.
6543 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
6544 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
6545 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
6546 is 0, use _0 suffixed expanders instead of emitting a comparison
6549 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
6551 * config/i386/i386.md(*movsf_internal, *movdf_internal):
6552 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
6554 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
6557 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
6558 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
6559 * loop-iv.c (iv_get_reaching_def): Likewise.
6560 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
6561 variable is promoted and the partition contains undefined values.
6563 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6565 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
6567 (nios2_address_cost): Define.
6568 (nios2_legitimize_address): Recognize (exp + constant) directly.
6569 (TARGET_ADDRESS_COST): Define.
6571 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6573 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
6574 (nios2_symbolic_memory_operand_p): Declare.
6575 (nios2_split_large_constant): Declare.
6576 (nios2_split_symbolic_memory_operand): Declare.
6577 * config/nios2/nios2.c: Adjust includes.
6578 (nios2_symbolic_constant_allowed): New.
6579 (nios2_symbolic_constant_p): New.
6580 (nios2_plus_symbolic_constant_p): New.
6581 (nios2_valid_addr_expr_p): Recognize addresses involving
6583 (nios2_legitimate_address_p): Likewise, also LO_SUM.
6584 (nios2_symbolic_memory_operand_p): New.
6585 (nios2_large_constant_p): New.
6586 (nios2_split_large_constant): New.
6587 (nios2_split_plus_large_constant): New.
6588 (nios2_split_symbolic_memory_operand): New.
6589 (nios2_legitimize_address): Code refactoring. Handle addresses
6590 involving symbolic constants.
6591 (nios2_emit_move_sequence): Likewise.
6592 (nios2_print_operand): Improve error output.
6593 (nios2_print_operand_address): Handle LO_SUM.
6594 (nios2_cdx_narrow_form_p): Likewise.
6595 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
6596 operands involving symbolic constants.
6597 (movhi_internal, movsi_internal): Likewise.
6598 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
6599 (extendhisi2, extendqi<mode>2): Likewise.
6601 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6603 * tree-pass.h (PROP_rtl_split_insns): Define.
6604 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
6606 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6608 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
6610 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6613 * target.def (const_not_ok_for_debug_p): Default to
6614 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
6615 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
6616 * targhooks.c (default_const_not_ok_for_debug_p): New function.
6617 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
6618 which targetm.const_not_ok_for_debug_p returned true.
6619 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
6621 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
6623 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
6624 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
6625 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
6627 (ix86_const_not_ok_for_debug_p): New function.
6628 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
6629 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
6631 2017-10-23 David Malcolm <dmalcolm@redhat.com>
6634 * system.h: Conditionally include "unique-ptr.h" if
6635 INCLUDE_UNIQUE_PTR is defined.
6636 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
6637 of defining INCLUDE_UNIQUE_PTR before including "system.h".
6639 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
6641 * config/rl78/rl78.md: New define_expand "subdi3".
6643 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
6646 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
6647 DF_REF_INSN if DF_REF_INSN_INFO is false.
6649 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
6651 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
6652 xmm_move instead of sse_move.
6653 (sse_store_index): New function.
6654 (ix86_register_move_cost): Be more sensible about mismatch stall;
6655 model AVX moves correctly; make difference between sse->integer and
6657 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
6658 moves; make difference between SSE and AVX.
6659 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
6660 and zmm_move. Increase size of sse load and store tables;
6661 add unaligned load and store tables; add ssemmx_to_integer.
6662 * x86-tune-costs.h: Update all entries according to real
6663 move latencies from Agner Fog's manual and chip documentation.
6665 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6668 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
6669 * config/i386/constraints.md (Wf): New constraint.
6670 * config/i386/i386.md (UNSPEC_SBB): New unspec.
6671 (cmp<dwi>_doubleword): Removed.
6672 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
6673 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
6674 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
6675 expand with cmp<dwi>_doubleword. For LTU and GEU use
6676 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
6678 * common.opt (gcolumn-info): Enable by default.
6679 * doc/invoke.texi (gcolumn-info): Document new default.
6681 2017-10-23 Richard Biener <rguenther@suse.de>
6683 PR tree-optimization/82672
6684 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
6685 Fold the stmt if we propagated into it.
6687 2017-10-23 Richard Biener <rguenther@suse.de>
6689 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
6690 (bitmap_remove_expr_from_set): ... this. All callers call this
6691 for non-constant values.
6692 (bitmap_set_subtract): Rename to...
6693 (bitmap_set_subtract_expressions): ... this. Adjust and
6695 (bitmap_set_contains_value): Remove superfluous check.
6696 (bitmap_set_replace_value): Inline into single caller ...
6697 (bitmap_value_replace_in_set): ... here and simplify.
6698 (dependent_clean): Merge into ...
6699 (clean): ... this using an overload. Adjust.
6700 (prune_clobbered_mems): Adjust.
6701 (compute_antic_aux): Likewise.
6702 (compute_partial_antic_aux): Likewise.
6704 2017-10-23 Richard Biener <rguenther@suse.de>
6706 PR tree-optimization/82129
6708 2017-08-01 Richard Biener <rguenther@suse.de>
6710 PR tree-optimization/81181
6711 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6712 (compute_antic): ... end of iteration here.
6714 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6716 * target.def (starting_frame_offset): New hook.
6717 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
6718 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
6719 * doc/tm.texi.in: Regenerate.
6720 * hooks.h (hook_hwi_void_0): Declare.
6721 * hooks.c (hook_hwi_void_0): New function.
6722 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
6723 STARTING_FRAME_OFFSET.
6724 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
6725 * reload1.c (reload): Likewise.
6726 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
6727 instead of STARTING_FRAME_OFFSET.
6728 * function.c (try_fit_stack_local): Likewise.
6729 (assign_stack_local_1): Likewise
6730 (instantiate_virtual_regs): Likewise.
6731 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
6732 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
6733 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
6734 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
6735 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
6736 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
6737 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
6738 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
6739 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
6740 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
6741 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
6742 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
6743 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
6744 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
6745 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
6746 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
6747 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
6748 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
6749 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
6750 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
6751 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
6752 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
6753 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
6754 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
6755 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
6756 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
6757 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
6758 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
6759 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
6760 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
6761 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
6762 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
6763 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
6764 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
6765 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
6766 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
6767 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
6768 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
6769 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
6770 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
6771 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
6772 * config/avr/avr.c (avr_starting_frame_offset): Make static and
6773 return a HOST_WIDE_INT.
6774 (avr_builtin_setjmp_frame_value): Use it instead of
6775 STARTING_FRAME_OFFSET.
6776 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6777 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
6778 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
6780 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6781 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
6782 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
6783 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6784 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
6785 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
6786 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6787 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
6788 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
6789 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6790 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
6791 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
6793 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6794 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
6795 * config/mips/mips.c (mips_compute_frame_info): Refer to
6796 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6797 (mips_starting_frame_offset): New function.
6798 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6799 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
6800 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
6801 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
6802 and return a HOST_WIDE_INT.
6803 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6804 (mmix_initial_elimination_offset): Refer to
6805 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6806 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
6807 * config/pa/pa.c (pa_starting_frame_offset): New function.
6808 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
6809 (pa_expand_prologue): Likewise.
6810 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6811 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
6812 !FRAME_GROWS_DOWNWARD handling to...
6813 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6814 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
6815 !FRAME_GROWS_DOWNWARD handling to...
6816 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6817 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
6818 !FRAME_GROWS_DOWNWARD handling to...
6819 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6820 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
6822 (rs6000_starting_frame_offset): New function.
6823 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
6824 !FRAME_GROWS_DOWNWARD handling to...
6825 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6826 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
6827 !FRAME_GROWS_DOWNWARD handling to...
6828 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6829 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
6830 !FRAME_GROWS_DOWNWARD handling to...
6831 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6832 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
6833 (rs6000_starting_frame_offset): New function.
6834 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
6835 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
6836 * config/vax/vax.c (vax_starting_frame_offset): New function.
6837 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
6838 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6839 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
6840 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
6841 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6842 * system.h (STARTING_FRAME_OFFSET): Poison.
6844 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6846 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
6847 SCALAR_TYPE_MODE instead of TYPE_MODE.
6849 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6850 Alan Hayward <alan.hayward@arm.com>
6851 David Sherwood <david.sherwood@arm.com>
6853 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
6855 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6856 Alan Hayward <alan.hayward@arm.com>
6857 David Sherwood <david.sherwood@arm.com>
6859 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
6861 2017-10-23 Richard Biener <rguenther@suse.de>
6863 PR tree-optimization/82129
6864 * tree-ssa-pre.c (bitmap_set_and): Remove.
6865 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
6866 canonicalizing expressions in the set to those with lowest
6867 ID rather than taking that from the first edge.
6869 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6871 * combine.c (rtx_equal_for_field_assignment_p): Use
6872 byte_lowpart_offset.
6874 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6875 Alan Hayward <alan.hayward@arm.com>
6876 David Sherwood <david.sherwood@arm.com>
6878 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
6879 to a SUBREG_PROMOTED_VAR.
6881 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6882 Alan Hayward <alan.hayward@arm.com>
6883 David Sherwood <david.sherwood@arm.com>
6885 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
6886 (expand_debug_source_expr): Likewise.
6887 * combine.c (combine_simplify_rtx): Likewise.
6888 * cse.c (fold_rtx): Likewise.
6889 * optabs.c (expand_float): Likewise.
6890 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6891 (simplify_binary_operation_1): Likewise.
6893 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6894 Alan Hayward <alan.hayward@arm.com>
6895 David Sherwood <david.sherwood@arm.com>
6897 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
6898 (record_promoted_value): Likewise.
6899 * expr.c (expand_expr_real_2): Likewise.
6900 * ree.c (update_reg_equal_equiv_notes): Likewise.
6901 (combine_set_extension): Likewise.
6902 * rtlanal.c (low_bitmask_len): Likewise.
6903 * simplify-rtx.c (neg_const_int): Likewise.
6904 (simplify_binary_operation_1): Likewise.
6906 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6907 Alan Hayward <alan.hayward@arm.com>
6908 David Sherwood <david.sherwood@arm.com>
6910 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
6911 * regcprop.c (maybe_mode_change): Likewise.
6912 * reload1.c (alter_reg): Likewise.
6914 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6916 * inchash.h (inchash::hash::add_wide_int): New function.
6917 * lto-streamer-out.c (hash_tree): Use it.
6919 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6921 * inchash.h (inchash::hash::add_wide_int): Rename to...
6922 (inchash::hash::add_hwi): ...this.
6923 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
6924 (polymorphic_call_target_hasher::hash): Likewise.
6925 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
6926 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
6927 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
6928 * lto-streamer-out.c (hash_tree): Likewise.
6929 * optc-save-gen.awk: Likewise.
6930 * tree.c (add_expr): Likewise.
6932 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6935 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
6936 for ordered inequality comparisons even with TARGET_IEEE_FP.
6938 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6941 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
6942 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6943 Expand with cmp<dwi>_doubleword.
6945 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6947 * extend.texi: Add x86 specific to 'nocf_check' attribute.
6948 List CET intrinsics.
6949 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
6950 specific to -fcf-protection option.
6952 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6954 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
6955 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
6956 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
6957 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
6958 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
6959 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
6960 (extra_objs): Add cet.o for Linux/x86 targets.
6961 (tmake_file): Add i386/t-cet for Linux/x86 targets.
6962 * config/i386/cet.c: New file.
6963 * config/i386/cetintrin.h: Likewise.
6964 * config/i386/t-cet: Likewise.
6965 * config/i386/cpuid.h (bit_SHSTK): New.
6966 (bit_IBT): Likewise.
6967 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
6968 pass IBT and SHSTK bits.
6969 * config/i386/i386-builtin-types.def
6970 (VOID_FTYPE_UNSIGNED_PVOID): New.
6971 (VOID_FTYPE_UINT64_PVOID): Likewise.
6972 * config/i386/i386-builtin.def: Add CET intrinsics.
6973 * config/i386/i386-c.c (ix86_target_macros_internal): Add
6974 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
6975 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
6976 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
6978 * config/i386/i386.c (rest_of_insert_endbranch): New.
6979 (pass_data_insert_endbranch): Likewise.
6980 (pass_insert_endbranch): Likewise.
6981 (make_pass_insert_endbranch): Likewise.
6982 (ix86_notrack_prefixed_insn_p): Likewise.
6983 (ix86_target_string): Add -mibt, -mshstk flags.
6984 (ix86_option_override_internal): Add flag_cf_protection
6986 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
6987 (ix86_print_operand): Add 'notrack' prefix output.
6988 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
6989 (ix86_expand_builtin): Expand CET intrinsics.
6990 (x86_output_mi_thunk): Add 'endbranch' instruction.
6991 * config/i386/i386.h (TARGET_IBT): New.
6992 (TARGET_IBT_P): Likewise.
6993 (TARGET_SHSTK): Likewise.
6994 (TARGET_SHSTK_P): Likewise.
6995 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
6996 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
6997 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
6998 (builtin_setjmp_setup): New pattern.
6999 (builtin_longjmp): Likewise.
7000 (rdssp<mode>): Likewise.
7001 (incssp<mode>): Likewise.
7002 (saveprevssp): Likewise.
7003 (rstorssp): Likewise.
7004 (wrss<mode>): Likewise.
7005 (wruss<mode>): Likewise.
7006 (setssbsy): Likewise.
7007 (clrssbsy): Likewise.
7008 (nop_endbr): Likewise.
7009 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
7011 * config/i386/immintrin.h: Include <cetintrin.h>.
7012 * config/i386/linux-common.h
7013 (file_end_indicate_exec_stack_and_cet): New prototype.
7014 (TARGET_ASM_FILE_END): New.
7016 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7018 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
7019 latencies instead of having separate table; make difference between
7020 integer and float costs.
7021 * i386.h (processor_costs): Remove scalar_stmt_cost,
7022 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
7023 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
7025 * x86-tune-costs.h: Remove entries which has been removed in
7026 procesor_costs from all tables; make cond_taken_branch_cost
7027 and cond_not_taken_branch_cost COST_N_INSNS based.
7029 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7031 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
7033 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7035 * config/i386/i386.md (isa): Remove fma_avx512f.
7036 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
7037 <avx512>_fmadd_<mode>_mask3<round_name>,
7038 <avx512>_fmsub_<mode>_mask<round_name>,
7039 <avx512>_fmsub_<mode>_mask3<round_name>,
7040 <avx512>_fnmadd_<mode>_mask<round_name>,
7041 <avx512>_fnmadd_<mode>_mask3<round_name>,
7042 <avx512>_fnmsub_<mode>_mask<round_name>,
7043 <avx512>_fnmsub_<mode>_mask3<round_name>,
7044 <avx512>_fmaddsub_<mode>_mask<round_name>,
7045 <avx512>_fmaddsub_<mode>_mask3<round_name>,
7046 <avx512>_fmsubadd_<mode>_mask<round_name>,
7047 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
7048 (*vec_widen_umult_even_v16si<mask_name>,
7049 *vec_widen_smult_even_v16si<mask_name>): Likewise.
7050 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
7052 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7054 * extend.texi: Add 'nocf_check' documentation.
7055 * gimple.texi: Add second parameter to
7056 gimple_build_call_from_tree.
7057 * invoke.texi: Add -fcf-protection documentation.
7058 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7060 2017-10-20 Richard Biener <rguenther@suse.de>
7062 PR tree-optimization/82473
7063 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7064 the largest input type.
7066 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7068 * c-attribs.c (handle_nocf_check_attribute): New function.
7069 (c_common_attribute_table): Add 'nocf_check' handling.
7070 * gimple-parser.c: Add second argument NULL to
7071 gimple_build_call_from_tree.
7072 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7073 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7075 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7076 * common.opt: Add fcf-protection flag.
7077 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7078 * flag-types.h: Add enum cf_protection_level.
7079 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7080 Add 'nocf_check' attribute propagation to gimple call.
7081 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7082 (gimple_build_call_from_tree): Update prototype.
7083 (gimple_call_nocf_check_p): New function.
7084 (gimple_call_set_nocf_check): Likewise.
7085 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7086 * ipa-icf.c: Add nocf_check attribute in statement hash.
7087 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7088 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7089 * toplev.c (process_options): Add flag_cf_protection handling.
7091 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7093 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7095 2017-10-20 Richard Biener <rguenther@suse.de>
7097 PR tree-optimization/82603
7098 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7099 remove false predicated stores.
7101 2017-10-20 Richard Biener <rguenther@suse.de>
7103 * graphite-isl-ast-to-gimple.c
7104 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7105 Remove return value and simplify, dump copied stmt after lhs
7107 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7108 Reduce dump verbosity.
7109 (gsi_insert_earliest): Likewise.
7110 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7111 * graphite.c (print_global_statistics): Adjust dumping.
7112 (print_graphite_scop_statistics): Likewise.
7113 (print_graphite_statistics): Do not dump loops here.
7114 (graphite_transform_loops): But here.
7116 2017-10-20 Nicolas Roche <roche@adacore.com>
7118 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
7119 * configure: Regenerate.
7121 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7124 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
7125 functions when optimizing replace GIMPLE_RETURN stmts with
7126 calls to __builtin_unreachable ().
7129 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
7130 for -fsanitize=thread link of executables.
7131 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
7132 link of executables.
7135 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
7137 (<shift_insn><mode>3<mask_name>): Change the last of the 3
7138 define_insns for logical vector shifts to use VI248_AVX512BW
7139 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
7140 condition, useless isa and prefix attributes. Change the first
7141 2 of these define_insns to ...
7142 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
7143 define_insn for avx512vl.
7144 (<shift_insn><mode>3): ... and this, new define_insn without
7145 masking for non-avx512vl.
7148 * config/i386/sse.md (*andnot<mode>3,
7149 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
7150 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
7151 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
7152 not applied use empty suffix even for TARGET_AVX512VL.
7153 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
7154 is applied, supply evex,evex or evex,evex,evex instead of just
7157 2017-10-20 Julia Koval <julia.koval@intel.com>
7159 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
7160 (OPTION_MASK_ISA_GFNI_UNSET): New.
7161 (ix86_handle_option): Handle OPT_mgfni.
7162 * config/i386/cpuid.h (bit_GFNI): New.
7163 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
7164 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
7165 * config/i386/i386.c (ix86_target_string): Add -mgfni.
7166 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
7167 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
7168 * config/i386/i386.opt: Add mgfni.
7170 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
7172 * config/msp430/msp430.c (msp430_option_override): Disable
7173 -fdelete-null-pointer-checks.
7174 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
7176 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7178 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
7179 of x87 and SSE instructions.
7181 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7183 * asan.c (create_cond_insert_point): Do not update edge count.
7184 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
7185 (afdo_propagate_circuit): Likewise.
7186 (afdo_calculate_branch_prob): Likewise.
7187 (afdo_annotate_cfg): Likewise.
7188 * basic-block.h (struct edge_def): Remove count.
7189 (edge_def::count): New accessor.
7190 * bb-reorder.c (rotate_loop): Update.
7191 (find_traces_1_round): Update.
7192 (connect_traces): Update.
7193 (sanitize_hot_paths): Update.
7194 * cfg.c (unchecked_make_edge): Update.
7195 (make_single_succ_edge): Update.
7196 (check_bb_profile): Update.
7197 (dump_edge_info): Update.
7198 (update_bb_profile_for_threading): Update.
7199 (scale_bbs_frequencies_int): Update.
7200 (scale_bbs_frequencies_gcov_type): Update.
7201 (scale_bbs_frequencies_profile_count): Update.
7202 (scale_bbs_frequencies): Update.
7203 * cfganal.c (connect_infinite_loops_to_exit): Update.
7204 * cfgbuild.c (compute_outgoing_frequencies): Update.
7205 (find_many_sub_basic_blocks): Update.
7206 * cfgcleanup.c (try_forward_edges): Update.
7207 (try_crossjump_to_edge): Update
7208 * cfgexpand.c (expand_gimple_cond): Update
7209 (expand_gimple_tailcall): Update
7210 (construct_exit_block): Update
7211 * cfghooks.c (verify_flow_info): Update
7212 (redirect_edge_succ_nodup): Update
7213 (split_edge): Update
7214 (make_forwarder_block): Update
7215 (duplicate_block): Update
7216 (account_profile_record): Update
7217 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
7218 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
7219 * cfgloopmanip.c (scale_loop_profile): Update.
7221 (lv_adjust_loop_entry_edge): Update.
7222 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7223 (force_nonfallthru_and_redirect): Update.
7224 (purge_dead_edges): Update.
7225 (rtl_flow_call_edges_add): Update.
7226 * cgraphunit.c (init_lowered_empty_function): Update.
7227 (cgraph_node::expand_thunk): Update.
7228 * gimple-pretty-print.c (dump_probability): Update.
7229 (dump_edge_probability): Update.
7230 * gimple-ssa-isolate-paths.c (isolate_path): Update.
7231 * haifa-sched.c (sched_create_recovery_edges): Update.
7232 * hsa-gen.c (convert_switch_statements): Update.
7233 * ifcvt.c (dead_or_predicable): Update.
7234 * ipa-inline-transform.c (inline_transform): Update.
7235 * ipa-split.c (split_function): Update.
7236 * ipa-utils.c (ipa_merge_profiles): Update.
7237 * loop-doloop.c (add_test): Update.
7238 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
7239 * lto-streamer-in.c (input_cfg): Update.
7240 (input_function): Update.
7241 * lto-streamer-out.c (output_cfg): Update.
7242 * modulo-sched.c (sms_schedule): Update.
7243 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
7244 * predict.c (maybe_hot_edge_p): Update.
7245 (unlikely_executed_edge_p): Update.
7246 (probably_never_executed_edge_p): Update.
7247 (dump_prediction): Update.
7248 (drop_profile): Update.
7249 (propagate_unlikely_bbs_forward): Update.
7250 (determine_unlikely_bbs): Update.
7251 (force_edge_cold): Update.
7252 * profile.c (compute_branch_probabilities): Update.
7253 * reg-stack.c (better_edge): Update.
7254 * shrink-wrap.c (handle_simple_exit): Update.
7255 * tracer.c (better_p): Update.
7256 * trans-mem.c (expand_transaction): Update.
7257 (split_bb_make_tm_edge): Update.
7258 * tree-call-cdce.c: Update.
7259 * tree-cfg.c (gimple_find_sub_bbs): Update.
7260 (gimple_split_edge): Update.
7261 (gimple_duplicate_sese_region): Update.
7262 (gimple_duplicate_sese_tail): Update.
7263 (gimple_flow_call_edges_add): Update.
7264 (insert_cond_bb): Update.
7265 (execute_fixup_cfg): Update.
7266 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
7267 * tree-complex.c (expand_complex_div_wide): Update.
7268 * tree-eh.c (lower_resx): Update.
7269 (unsplit_eh): Update.
7270 (cleanup_empty_eh_move_lp): Update.
7271 * tree-inline.c (copy_edges_for_bb): Update.
7272 (freqs_to_counts): Update.
7273 (copy_cfg_body): Update.
7274 * tree-ssa-dce.c (remove_dead_stmt): Update.
7275 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7276 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7277 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7278 (unloop_loops): Update.
7279 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7280 * tree-ssa-loop-split.c (connect_loops): Update.
7281 (split_loop): Update.
7282 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
7283 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7284 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7285 * tree-ssa-reassoc.c (branch_fixup): Update.
7286 * tree-ssa-tail-merge.c (replace_block_by): Update.
7287 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7288 (compute_path_counts): Update.
7289 (update_profile): Update.
7290 (recompute_probabilities): Update.
7291 (update_joiner_offpath_counts): Update.
7292 (estimated_freqs_path): Update.
7293 (freqs_to_counts_path): Update.
7294 (clear_counts_path): Update.
7295 (ssa_fix_duplicate_block_edges): Update.
7296 (duplicate_thread_path): Update.
7297 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7298 (case_bit_test_cmp): Update.
7299 (collect_switch_conv_info): Update.
7300 (gen_inbound_check): Update.
7301 (do_jump_if_equal): Update.
7302 (emit_cmp_and_jump_insns): Update.
7303 * tree-tailcall.c (decrease_profile): Update.
7304 (eliminate_tail_call): Update.
7305 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7306 (vect_do_peeling): Update.
7307 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7308 * ubsan.c (ubsan_expand_null_ifn): Update.
7309 (ubsan_expand_ptr_ifn): Update.
7310 * value-prof.c (gimple_divmod_fixed_value): Update.
7311 (gimple_mod_pow2): Update.
7312 (gimple_mod_subtract): Update.
7313 (gimple_ic): Update.
7314 (gimple_stringop_fixed_value): Update.
7316 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7319 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
7321 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
7323 PR rtl-optimization/82395
7324 * ira-color.c (allocno_priority_compare_func): Fix comparison step
7325 based on non_spilled_static_chain_regno_p.
7327 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7329 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
7330 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
7331 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
7333 2017-10-19 Martin Sebor <msebor@redhat.com>
7335 PR tree-optimization/82596
7336 * tree.c (array_at_struct_end_p): Handle STRING_CST.
7338 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7340 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
7341 (get_mem_refs_of_builtin_call): Likewise.
7342 * builtins.c (expand_builtin_apply): Adjust call to
7343 allocate_dynamic_stack_space.
7344 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
7345 the third argument to allocate_dynamic_stack_space, otherwise -1.
7346 (expand_builtin): Deal with all alloca variants.
7347 (is_inexpensive_builtin): Likewise.
7348 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
7349 * calls.c (special_function_p): Deal with all alloca variants.
7350 (initialize_argument_information): Adjust call to
7351 allocate_dynamic_stack_space.
7352 (expand_call): Likewise.
7353 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
7354 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
7355 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
7356 use it for the stack usage computation.
7357 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
7358 * function.c (gimplify_parameters): Call build_alloca_call_expr.
7359 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
7360 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
7361 (in_loop_p): Remove first argument and useless check.
7362 (pass_walloca::execute): Remove useless test and adjust call to above.
7363 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
7364 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
7365 (gimplify_call_expr): Deal with all alloca variants.
7366 * hsa-gen.c (gen_hsa_alloca): Likewise.
7367 (gen_hsa_insns_for_call): Likewise.
7368 * ipa-pure-const.c (special_builtin_state): Likewise.
7369 * tree-chkp.c (chkp_build_returned_bound): Likewise.
7370 * tree-object-size.c (alloc_object_size): Likewise.
7371 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
7372 (call_may_clobber_ref_p_1): Likewise.
7373 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
7374 (ccp_fold_stmt): Likewise.
7375 (optimize_stack_restore): Likewise.
7376 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
7377 (mark_all_reaching_defs_necessary_1): Likewise.
7378 (propagate_necessity): Likewise.
7379 (eliminate_unnecessary_stmts): Likewise.
7380 * tree.c (build_common_builtin_nodes): Build
7381 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
7382 (build_alloca_call_expr): New function.
7383 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
7384 (CASE_BUILT_IN_ALLOCA): Likewise.
7385 (build_alloca_call_expr): Declare.
7386 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
7388 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7391 * dwarf2out.c (new_die_raw): New static inline function.
7392 (new_die): Use it to create the DIE.
7393 (add_AT_external_die_ref): Likewise.
7394 (clone_die): Likewise.
7395 (clone_as_declaration): Likewise.
7396 (dwarf2out_vms_debug_main_pointer): Likewise.
7397 (base_type_die): Likewise. Remove early return for corner cases.
7398 Do not call add_pubtype on the DIE here.
7399 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
7400 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
7401 typedefs for base types with DW_AT_endianity. Make sure a DIE with
7402 native order exists for base types, attach the DIE manually and call
7403 add_pubtype on it. Do not equate a reverse order DIE to the type.
7405 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
7407 * config/arm/arm.c (align_ok_ldrd_strd): New function.
7408 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
7410 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
7412 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7414 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
7415 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
7416 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
7417 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
7418 * opts.c (sanitizer_opts): Add builtin.
7419 * ubsan.c (instrument_builtin): New function.
7420 (pass_ubsan::execute): Call it.
7421 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
7422 * doc/invoke.texi: Document -fsanitize=builtin.
7424 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
7425 builtins, store max (log2 (align), 0) into uchar field instead of
7426 align into uptr field.
7427 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
7428 store uchar 0 field instead of uptr 0 field.
7429 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
7430 instead of passing one address of struct with 2 locations pass
7431 two addresses of structs with 1 location each.
7432 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
7433 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
7434 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
7435 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
7436 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
7437 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
7438 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
7439 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
7441 2017-10-19 Martin Liska <mliska@suse.cz>
7444 * file-find.c (remove_prefix): Remove.
7445 * file-find.h (remove_prefix): Likewise.
7446 * gcc-ar.c: Remove smartness of lookup.
7448 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
7450 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
7451 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
7452 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
7454 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7457 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
7458 (setcc + and to xor + setcc): New peephole2.
7460 2017-10-19 Tom de Vries <tom@codesourcery.com>
7462 * doc/sourcebuild.texi (Test Directives, Variants of
7463 dg-require-support): Add dg-require-stack-size.
7465 2017-10-19 Martin Liska <mliska@suse.cz>
7468 * gimplify.c (gimplify_decl_expr): Do not instrument variables
7469 that have a large alignment.
7470 (gimplify_target_expr): Likewise.
7472 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
7474 PR rtl-optimization/82602
7475 * ira.c (rtx_moveable_p): Return false for volatile asm.
7477 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
7480 * config/i386/i386-modes.def (CCGZ): New CC mode.
7481 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
7482 * config/i386/predicates.md (ix86_comparison_operator):
7484 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7485 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
7486 with double-word subtraction.
7487 (put_condition_code): Handle CCGZmode.
7489 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
7491 * wide-int.cc (debug (const wide_int &)): New.
7492 (debug (const wide_int *)): New.
7493 (debug (const widest_int &)): New.
7494 (debug (const widest_int *)): New.
7496 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
7499 * lra-constraints.c (curr_insn_transform): Use non-input operand
7500 instead of output one for matched reload.
7502 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7504 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
7505 (tree-ssa-loop-ivopts.h): New header file.
7506 (struct builtin_info): New fields.
7507 (classify_builtin_1): Compute and record base and offset parts for
7508 memset builtin partition by calling strip_offset.
7509 (offset_cmp, fuse_memset_builtins): New functions.
7510 (finalize_partitions): Fuse adjacent memset partitions by calling
7512 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
7513 Expose the interface.
7514 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
7516 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7518 PR tree-optimization/82574
7519 * tree-loop-distribution.c (find_single_drs): New parameter. Check
7520 that data reference must be executed exactly once per iteration
7521 against the outermost loop in nest.
7522 (classify_partition): Update call to above function.
7524 2017-10-18 Richard Biener <rguenther@suse.de>
7526 PR tree-optimization/82591
7527 * graphite.c (graphite_transform_loops): Move code gen message
7529 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7530 Here. Handle scop_to_isl_ast failing.
7531 (scop_to_isl_ast): Limit the number of ISL operations.
7533 2017-10-18 Richard Biener <rguenther@suse.de>
7535 * graphite-isl-ast-to-gimple.c
7536 (translate_isl_ast_to_gimple::set_rename): Simplify.
7537 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
7538 (graphite_copy_stmts_from_block): ... here.
7539 (copy_bb_and_scalar_dependences): Simplify.
7540 (add_parameters_to_ivs_params): Canonicalize.
7541 (generate_entry_out_of_ssa_copies): Simplify.
7542 * graphite-sese-to-poly.c (extract_affine_name): Simplify
7543 by passing in ISL dimension.
7544 (parameter_index_in_region_1): Rename to ...
7545 (parameter_index_in_region): ... this.
7546 (extract_affine): Adjust assert, pass down parameter index.
7547 (add_param_constraints): Use range-info when available.
7548 (build_scop_context): Adjust.
7549 * sese.c (new_sese_info): Adjust.
7550 (free_sese_info): Likewise.
7551 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
7552 Remove unused typedefs.
7553 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
7555 2017-10-18 Martin Liska <mliska@suse.cz>
7557 * combine.c (simplify_compare_const): Add gcc_fallthrough.
7559 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7561 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
7562 (s390_sched_init): Do not reset s390_sched_state if we entered the
7563 current basic block via a fallthru edge and all others are unlikely.
7565 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7567 * config/s390/s390.c (NUM_SIDES): New variable.
7568 (LONGRUNNING_THRESHOLD): New variable.
7569 (LATENCY_FACTOR): New variable.
7570 (s390_sched_score): Decrease score for long-running instructions on
7572 (s390_sched_variable_issue): Perform bookkeeping for long-running
7575 2017-10-18 Richard Biener <rguenther@suse.de>
7577 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
7578 Simplify with removal of the parameter rename map.
7579 (set_rename): Likewise.
7580 (should_copy_to_new_region): Likewise.
7581 (graphite_copy_stmts_from_block): Likewise.
7582 (copy_bb_and_scalar_dependences): Remove initialization of
7583 unused copied_bb_map.
7585 (copy_internal_parameters): Likewise.
7586 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
7587 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
7588 Use INTEGRAL_TYPE_P.
7589 (parameter_index_in_region_1): Rename to ...
7590 (assign_parameter_index_in_region): ... this. Assert we have
7591 a parameter we handle.
7592 (scan_tree_for_params): Adjust.
7593 * sese.h (parameter_rename_map_t): Remove.
7594 (struct sese_info_t): Remove unused parameter_rename_map and
7595 copied_bb_map members.
7596 * sese.c (new_sese_info): Adjust.
7597 (free_sese_info): Likewise.
7599 2017-10-18 Martin Liska <mliska@suse.cz>
7602 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
7603 on an abnormal edge.
7605 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
7607 * doc/invoke.texi (ffunction-sections and fdata-sections):
7610 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7612 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
7613 the use statement can throw internally.
7615 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7617 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
7618 any RTX present on the RHS of a SET.
7619 * compare-elim.c (try_eliminate_compare): Restore comment.
7621 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7623 * langhooks.h (struct lang_hooks): Document that tree_size langhook
7624 may be also called on tcc_type nodes.
7625 * langhooks.c (lhd_tree_size): Likewise.
7627 2017-10-17 David Malcolm <dmalcolm@redhat.com>
7629 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
7630 format_warning_at_substring.
7631 (maybe_warn): Convert source_range * param to a location_t. Pass
7632 UNKNOWN_LOCATION rather than NULL to fmtwarn.
7633 (format_directive): Remove code to extract source_ranges and
7634 source_range * in favor of just a location_t.
7635 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
7637 * substring-locations.c (format_warning_va): Convert
7638 source_range * param to a location_t.
7639 (format_warning_at_substring): Likewise.
7640 * substring-locations.h (format_warning_va): Likewise.
7641 (format_warning_at_substring): Likewise.
7643 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
7645 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
7647 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
7648 and scatter/gather ops.
7650 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
7651 vec_gather_load and vec_scatter_store.
7652 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
7653 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
7654 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
7655 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
7656 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
7657 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
7659 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
7661 * reg-stack.c (compare_for_stack_reg): Add bool argument.
7662 Detect FTST instruction and handle its register pops. Only pop
7663 second operand if can_pop_second_op is true.
7664 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
7665 set can_pop_second_op to false in the compare_for_stack_reg call.
7667 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
7668 output_fp_compare for stack register operands.
7669 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
7670 instructions here. Do not emit stack register pops here. Assert
7671 that FCOMPP pops next to top stack register. Rewrite function.
7673 2017-10-17 Nathan Sidwell <nathan@acm.org>
7676 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
7677 use DECL_ASSEMBLER_NAME_RAW.
7680 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
7683 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
7684 Wilco Dijkstra <wilco.dijkstra@arm.com>
7686 * builtins.c (expand_builtin_update_setjmp_buf): Add a
7687 converstion to Pmode from the buf_addr.
7689 2017-10-17 Richard Biener <rguenther@suse.de>
7691 * graphite-dependences.c (scop_get_reads_and_writes): Change
7692 output parameters to references.
7694 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
7696 PR 71026/tree-optimization
7697 * fold-const.c (distribute_real_division): Removed.
7698 (fold_binary_loc): Remove calls to distribute_real_divison.
7700 2017-10-17 Richard Biener <rguenther@suse.de>
7702 * graphite-scop-detection.c
7703 (scop_detection::stmt_has_simple_data_refs_p): Always use
7704 the full nest as region.
7705 (try_generate_gimple_bb): Likewise.
7706 * sese.c (scalar_evolution_in_region): Simplify now that
7707 SCEV can handle instantiation in regions.
7708 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
7709 in the non-loop part of a function if requested.
7711 2017-10-17 Richard Biener <rguenther@suse.de>
7713 PR tree-optimization/82563
7714 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
7716 (graphite_regenerate_ast_isl): Call it.
7717 * graphite-scop-detection.c (build_scops): Remove entry edge split.
7719 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7721 PR tree-optimization/82549
7722 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
7723 Formatting fixes. Instead of calling make_bit_field_ref with negative
7726 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
7728 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
7729 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
7730 _mm_maskz_reduce_ss): New.
7731 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
7732 __builtin_ia32_reducess_mask): Ditto..
7733 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
7734 * config/i386/sse.md (reduces<mode>): Renamed to ...
7735 (reduces<mode><mask_scalar_name>): ... this.
7736 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
7738 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
7739 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
7741 2017-10-16 David Malcolm <dmalcolm@redhat.com>
7743 * Makefile.in (OBJS): Add unique-ptr-tests.o.
7744 * selftest-run-tests.c (selftest::run_tests): Call
7745 selftest::unique_ptr_tests_cc_tests.
7746 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
7747 * unique-ptr-tests.cc: New file.
7749 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
7752 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7754 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7755 (make_hard_regno_born, make_hard_regno_dead): Update
7756 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
7758 2017-10-16 Jeff Law <law@redhat.com>
7760 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7762 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
7764 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
7766 2017-10-16 Olivier Hainque <hainque@adacore.com>
7768 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
7769 with_cpu if we were configured for an e500v2 target cpu name.
7771 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
7773 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
7774 * doc/invoke.texi: Document +nodsp as a valid extension for
7777 2017-10-16 Martin Liska <mliska@suse.cz>
7779 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
7780 (test_set_range): Likewise.
7781 (test_range_functions): Rename to ...
7782 (test_bit_in_range): ... this.
7783 (sbitmap_c_tests): Add new test.
7785 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7787 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
7789 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
7790 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
7792 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7794 * config/aarch64/aarch64-builtins.c
7795 (aarch64_types_quadopu_lane_qualifiers): New.
7796 (TYPES_QUADOPU_LANE): New.
7797 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
7798 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
7799 (aarch64_<sur>dot_laneq<vsi2qi>): New.
7800 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
7801 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
7802 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
7803 (Vdottype, DOTPROD): New.
7804 (sur): Add SDOT and UDOT.
7806 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7808 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
7809 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
7810 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
7812 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
7813 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
7814 Enable TARGET_DOTPROD.
7815 (cortex-a75.cortex-a55): Likewise.
7816 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
7818 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7820 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
7821 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
7823 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
7825 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
7826 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
7827 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
7828 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
7829 * config/arm/types.md (neon_dot, neon_dot_q): New.
7830 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
7832 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7834 * config/arm/arm.h (TARGET_DOTPROD): New.
7835 * config/arm/arm.c (arm_arch_dotprod): New.
7836 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
7837 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
7838 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
7839 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
7840 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
7841 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
7842 * doc/invoke.texi (armv8.2-a): Document dotprod
7844 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7846 * i386.c (ix86_vec_cost): New function.
7847 (ix86_rtx_costs): Handle vector operations better.
7848 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
7849 * x86-tune-costs.h: Add new costs to all tables.
7851 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7853 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
7855 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
7856 divsd, sqrtss and sqrtsd
7857 * x86-tune-costs.h: Add new entries to all costs.
7858 (znver1_cost): Fix to match real instruction latencies.
7860 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7861 Michael Collison <michael.collison@arm.com>
7863 * compare-elim.c: Include emit-rtl.h.
7864 (can_merge_compare_into_arith): New function.
7865 (try_validate_parallel): Likewise.
7866 (try_merge_compare): Likewise.
7867 (try_eliminate_compare): Call the above when no previous clobber
7869 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
7872 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7876 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
7877 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
7881 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
7882 to be any operand_equal_p operands. For & (B - 1) require
7883 B to be power of 2. Recognize
7884 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
7886 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7889 * optabs.c (expand_memory_blockage): Fix call of
7890 targetm.have_memory_blockage.
7892 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7895 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
7896 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
7897 objects to extra_objs instead of overwriting it.
7899 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7901 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
7902 Use any_fp_register_operand as operand[3] predicate. Simplify
7903 equality test for operands[2] and operands[4] memory location.
7904 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
7905 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
7906 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
7907 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
7908 any_fp_register_operand as operand[1] predicate. Simplify
7909 equality test for operands[0] and operands[3] memory location.
7910 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
7911 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
7912 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
7914 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7916 * target-insns.def: Add memory_blockage.
7917 * optabs.c (expand_memory_blockage): New function.
7918 (expand_asm_memory_barrier): Rename ...
7919 (expand_asm_memory_blockage): ... to this.
7920 (expand_mem_thread_fence): Call expand_memory_blockage
7921 instead of expand_asm_memory_barrier.
7922 (expand_mem_singnal_fence): Ditto.
7923 (expand_atomic_load): Ditto.
7924 (expand_atomic_store): Ditto.
7925 * doc/md.texi (Standard Pattern Names For Generation):
7926 Document memory_blockage instruction pattern.
7928 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
7930 * config/rl78/rl78.c (rl78_emit_libcall): New function.
7931 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
7932 * config/rl78/rl78.md: New define_expand "adddi3".
7934 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7936 * cfghooks.c (verify_flow_info): Disable check that all probabilities
7939 2017-10-13 Jeff Law <law@redhat.com>
7941 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
7943 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7946 * internal-fn.c (expand_mul_overflow): If both operands have
7947 the same highpart of -1 or 0 and the topmost bit of lowpart
7948 is different, overflow is if res <= 0 rather than res < 0.
7950 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
7952 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
7953 TARGET_P9_VECTOR code for unaligned_load case.
7955 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7957 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
7959 2017-10-13 Nathan Sidwell <nathan@acm.org>
7961 * tree-core.h (tree_contains_struct): Make bool.
7962 * tree.c (tree_contains_struct): Likewise.
7963 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
7964 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
7965 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
7966 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
7967 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
7969 2017-10-13 Richard Biener <rguenther@suse.de>
7971 * graphite-isl-ast-to-gimple.c
7972 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
7973 parameters and dominance check.
7974 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
7975 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
7976 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
7977 Do not update SSA form here or do intermediate IL verification.
7978 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
7979 (graphite_initialize): Remove check on the number of loops in
7980 the function and inline into graphite_transform_loops.
7981 (graphite_finalize): Inline into graphite_transform_loops.
7982 (graphite_transform_loops): Perform SSA update and IL verification
7984 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
7986 2017-10-13 Richard Biener <rguenther@suse.de>
7988 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
7989 graphite_expression_type_precision): Avoid global constructor
7991 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
7992 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
7993 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
7994 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
7995 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
7996 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
7998 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
8001 * config/i386/i386.h (ix86_red_zone_size): New.
8002 * config/i386/i386.md (push peephole2s): Replace
8003 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
8005 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8006 Alan Hayward <alan.hayward@arm.com>
8007 David Sherwood <david.sherwood@arm.com>
8009 * combine.c (can_change_dest_mode): Reject changes in
8010 REGMODE_NATURAL_SIZE.
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 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
8017 (expand_debug_source_expr): Likewise.
8018 * combine.c (combine_simplify_rtx): Likewise.
8019 * cse.c (fold_rtx): Likewise.
8020 * fwprop.c (canonicalize_address): Likewise.
8021 * targhooks.c (default_shift_truncation_mask): Likewise.
8023 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8024 Alan Hayward <alan.hayward@arm.com>
8025 David Sherwood <david.sherwood@arm.com>
8027 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
8028 (widened_mode): Likewise.
8029 (expand_unop): Likewise.
8030 * ree.c (transform_ifelse): Likewise.
8031 (merge_def_and_ext): Likewise.
8032 (combine_reaching_defs): Likewise.
8033 * simplify-rtx.c (simplify_unary_operation_1): 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 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
8040 * combine.c (gen_lowpart_for_combine): Likewise.
8041 * dwarf2out.c (rtl_for_decl_location): Likewise.
8042 * final.c (alter_subreg): Likewise.
8043 * rtlhooks.c (gen_lowpart_general): Likewise.
8044 (gen_lowpart_if_possible): Likewise.
8046 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8047 Alan Hayward <alan.hayward@arm.com>
8048 David Sherwood <david.sherwood@arm.com>
8050 * calls.c (expand_call): Use subreg_lowpart_offset.
8051 * cse.c (cse_insn): Likewise.
8052 * regcprop.c (copy_value): Likewise.
8053 (copyprop_hardreg_forward_1): Likewise.
8055 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8058 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8059 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8060 =Q constraints to +Q and into insn condition add check
8061 that operands[0] and operands[1] are equal.
8062 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8063 =Q constraints to +Q and into insn condition add check
8064 that operands[0] is equal to either operands[1] or operands[2].
8067 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8068 instead of handling MINUS_EXPR twice (once for each argument),
8069 canonicalize operand order and handle just once, use rtype where
8070 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8073 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8074 any values of __C while still being pattern recognizable as a simple
8077 2017-10-13 Richard Biener <rguenther@suse.de>
8079 PR tree-optimization/82451
8081 2017-10-02 Richard Biener <rguenther@suse.de>
8083 PR tree-optimization/82355
8084 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8085 a mapping for the enclosing loop but avoid generating one for
8087 (copy_bb_and_scalar_dependences): Remove premature codegen
8088 error on PHIs in blocks duplicated into multiple places.
8089 * graphite-scop-detection.c
8090 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8091 in the region use it as loop and nest to analyze the DR in.
8092 (try_generate_gimple_bb): Likewise.
8093 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8094 (add_loop_constraints): For blocks in a loop not in the region
8095 create a dimension with a single iteration.
8096 * sese.h (gbb_loop_at_index): Remove assert.
8098 * cfgloop.c (loop_preheader_edge): For the loop tree root
8099 return the single successor of the entry block.
8100 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8101 Reset the SCEV hashtable and niters.
8102 * graphite-scop-detection.c
8103 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8104 assert that we only have POLYNOMIAL_CHREC that vary in loops
8105 contained in the region.
8106 (scop_detection::graphite_can_represent_expr): Adjust.
8107 (scop_detection::stmt_has_simple_data_refs_p): For loops
8108 not in the region set loop to NULL. The nest is now the
8109 entry edge to the region.
8110 (try_generate_gimple_bb): Likewise.
8111 * sese.c (scalar_evolution_in_region): Adjust for
8112 instantiate_scev change.
8113 * tree-data-ref.h (graphite_find_data_references_in_stmt):
8114 Make nest parameter the edge into the region.
8115 (create_data_ref): Likewise.
8116 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
8117 entry edge into a region and adjust instantiate_scev calls.
8118 (create_data_ref): Likewise.
8119 (graphite_find_data_references_in_stmt): Likewise.
8120 (find_data_references_in_stmt): Pass the loop preheader edge
8121 from the nest argument.
8122 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
8123 parameter the edge into the region.
8124 (instantiate_parameters): Use the loop preheader edge as entry.
8125 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
8127 (get_instantiated_value_entry): Make instantiate_below parameter
8128 the edge into the region.
8129 (instantiate_scev_name): Likewise. Adjust dominance checks,
8130 when we cannot use loop-based instantiation instantiate by
8131 walking use-def chains.
8132 (instantiate_scev_poly): Adjust.
8133 (instantiate_scev_binary): Likewise.
8134 (instantiate_scev_convert): Likewise.
8135 (instantiate_scev_not): Likewise.
8136 (instantiate_array_ref): Remove.
8137 (instantiate_scev_3): Likewise.
8138 (instantiate_scev_2): Likewise.
8139 (instantiate_scev_1): Likewise.
8140 (instantiate_scev_r): Do not blindly handle N-operand trees.
8141 Do not instantiate array-refs. Handle all constants and invariants.
8142 (instantiate_scev): Make instantiate_below parameter
8143 the edge into the region.
8144 (resolve_mixers): Use the loop preheader edge for the region
8145 parameter to instantiate_scev_r.
8146 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
8148 2017-10-13 Richard Biener <rguenther@suse.de>
8150 PR tree-optimization/82525
8151 * graphite-isl-ast-to-gimple.c
8152 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
8154 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
8155 Fail code generation when we cannot represent the isl integer.
8156 (binary_op_to_tree): Elide modulo operations that are no-ops
8157 in the type we code generate. Remove now superfluous code
8160 2017-10-13 Richard Biener <rguenther@suse.de>
8162 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
8163 (scop_detection::harmful_loop_in_region): Remove premature
8164 IV type restriction.
8165 (scop_detection::graphite_can_represent_scev): We can handle
8166 pointer IVs just fine.
8168 2017-10-13 Alan Modra <amodra@gmail.com>
8170 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
8171 "Clobbers and Scratch Registers". Add paragraph on
8172 alternative to clobbers for scratch registers and OpenBLAS
8175 2017-10-13 Alan Modra <amodra@gmail.com>
8177 * doc/extend.texi (Clobbers): Correct vax example. Delete old
8178 example of a memory input for a string of known length. Move
8179 commentary out of table. Add a number of new examples
8180 covering array memory inputs.
8182 2017-10-12 Martin Liska <mliska@suse.cz>
8184 PR tree-optimization/82493
8185 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
8186 (test_range_functions): New function.
8187 (sbitmap_c_tests): Likewise.
8188 * selftest-run-tests.c (selftest::run_tests): Run new tests.
8189 * selftest.h (sbitmap_c_tests): New function.
8191 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8193 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8195 * config/rs6000/amo.h: Fix spacing issue.
8197 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8200 * config/i386/i386.md (*ashl<mode>3_mask_1,
8201 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
8202 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
8205 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8207 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
8208 (profile_probability): Set max_probability
8209 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
8211 * profile-count.c (profile_probability::differs_from_p): Do not
8212 rely on max_probaiblity == 10000
8214 2017-10-12 Jeff Law <law@redhat.com>
8216 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
8219 2017-10-12 Martin Sebor <msebor@redhat.com>
8223 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
8224 (handle_alias_pairs): Call it.
8225 * common.opt (-Wattribute-alias): New option.
8226 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
8227 * doc/invoke.texi (-Wattribute-alias): Document.
8229 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
8232 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8234 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8236 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8237 (make_hard_regno_born, make_hard_regno_dead): Update
8238 bb_killed_pseudos and bb_gen_pseudos.
8240 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8242 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
8244 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
8246 * config/alpha/alpha.c (alpha_split_conditional_move):
8247 Use std::swap instead of manually swapping.
8248 (alpha_stdarg_optimize_hook): Ditto.
8249 (alpha_canonicalize_comparison): Ditto.
8251 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8253 * tree-loop-distribution.c (struct builtin_info): New struct.
8254 (struct partition): Refactor fields into struct builtin_info.
8255 (partition_free): Free struct builtin_info.
8256 (build_size_arg_loc, build_addr_arg_loc): Delete.
8257 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
8258 information from struct builtin_info.
8259 (find_single_drs): New function refactored from classify_partition.
8260 Also moved builtin validity checks to this function.
8261 (compute_access_range, alloc_builtin): New functions.
8262 (classify_builtin_st, classify_builtin_ldst): New functions.
8263 (classify_partition): Refactor code into functions find_single_drs,
8264 classify_builtin_st and classify_builtin_ldst.
8265 (distribute_loop): Don't do runtime alias check when distributing
8267 (find_seed_stmts_for_distribution): New function.
8268 (pass_loop_distribution::execute): Refactor code finding seed
8269 stmts into above function. Support distribution for the innermost
8270 two-level loop nest. Adjust dump information.
8272 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8274 * tree-loop-distribution.c: Adjust the general comment.
8275 (NUM_PARTITION_THRESHOLD): New macro.
8276 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
8277 (classify_partition): Skip builtin pattern of loop nest's inner loop.
8278 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
8279 in call to build_partition_graph.
8280 (finalize_partitions): New parameter. Make loop distribution more
8281 conservative by fusing more partitions.
8282 (distribute_loop): Don't do runtime alias check in case of loop nest
8284 (find_seed_stmts_for_distribution): New function.
8285 (prepare_perfect_loop_nest): New function.
8286 (pass_loop_distribution::execute): Refactor code finding seed stmts
8287 and loop nest into above functions. Support loop nest distribution.
8288 Adjust dump information accordingly.
8290 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8292 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
8293 and set PTYPE_SEQUENTIAL for merged partition.
8295 2017-10-12 Richard Biener <rguenther@suse.de>
8297 PR tree-optimization/69728
8299 2017-09-19 Richard Biener <rguenther@suse.de>
8301 PR tree-optimization/69728
8302 * graphite-sese-to-poly.c (schedule_error): New global.
8303 (add_loop_schedule): Handle empty domain by failing the
8305 (build_original_schedule): Handle schedule_error.
8307 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
8308 domain by returning an unchanged schedule.
8310 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8312 * genrecog.c (validate_pattern): For VEC_SELECT verify that
8313 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
8315 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
8317 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
8320 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8323 * expr.c (store_field): Don't optimize away bitsize == 0 store
8324 from CALL_EXPR with addressable return type.
8326 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8328 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
8329 * config/rs6000/rs6000.md (sel): Delete mode attribute.
8330 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
8331 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
8332 TARGET_ISEL instead of TARGET_ISEL<sel>.
8334 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
8336 * config/rs6000/rs6000.c
8337 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
8339 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8341 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
8343 (reg_or_cint_operand): Fix comment.
8344 (reg_or_zero_operand): New predicate.
8345 * config/rs6000/rs6000-protos.h (output_isel): Delete.
8346 * config/rs6000/rs6000.c (output_isel): Delete.
8347 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
8348 instead of reg_or_cint_operand. Output instruction directly (not via
8350 (isel_unsigned_<mode>): Ditto.
8351 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
8352 gpc_reg_operand. Add an instruction alternative for this. Output
8353 instruction directly.
8354 (*isel_reversed_unsigned_<mode>): Ditto.
8356 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8358 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
8359 (TARGET_CANONICALIZE_COMPARISON): Define.
8361 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
8364 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8365 Check whether the dest is REG before adding REG_EQUIV note.
8367 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8370 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8372 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8373 (make_hard_regno_born, make_hard_regno_dead): Update
8374 bb_killed_pseudos and bb_gen_pseudos.
8376 2017-10-11 Nathan Sidwell <nathan@acm.org>
8378 * incpath.h (enum incpath_kind): Name enum, prefix values.
8379 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
8380 * incpath.c (heads, tails): Use INC_MAX.
8381 (add_env_var_paths, add_standard_paths): Use incpath_kind.
8382 (merge_include_chains, split_quote_chain,
8383 register_include_chains): Update incpath_kind names.
8384 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
8385 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
8387 (add_framework_path, darwin_register_objc_includes): Likewise.
8388 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
8390 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8392 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
8393 Do not use float_operator operator predicate.
8394 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
8395 * config/i386/predicates.md (float_operator): Remove predicate.
8397 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8399 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
8400 (*jccxf_i387): Ditto.
8401 (*jcc<mode>_i387): Ditto.
8402 (*jccu<mode>_i387): Ditto.
8403 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
8404 (*jcc_*_i387 splitters): Remove.
8405 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
8406 * config/i386/i386.c (ix86_split_fp_branch): Remove.
8407 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
8410 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8412 * profile-count.h (slow_safe_scale_64bit): New function.
8413 (safe_scale_64bit): New inline.
8414 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
8415 * profile-count.c: Include wide-int.h
8416 (slow_safe_scale_64bit): New.
8418 2017-10-11 Nathan Sidwell <nathan@acm.org>
8420 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
8421 HAS_DECL_ASSEMBLER_NAME_P.
8422 * gimple-expr.c (gimple_decl_printable_name: Check
8423 HAS_DECL_ASSEMBLER_NAME_P too.
8424 * ipa-utils.h (type_in_anonymous_namespace_p): Check
8425 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
8426 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
8427 * passes.c (rest_of_decl_compilation): Check
8428 HAS_DECL_ASSEMBLER_NAME_P too.
8429 * recog.c (verify_changes): Likewise.
8430 * tree-pretty-print.c (dump_decl_name): Likewise.
8431 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
8433 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
8434 (SET_DECL_ASSEMBLER_NAME): Use it.
8435 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
8436 (COPY_DECL_ASSEMBLER_NAME): Likewise.
8437 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
8439 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8441 * config.gcc (i386, x86_64): Add extra objects.
8442 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
8443 (ix86_min_insn_size): Declare.
8444 (ix86_issue_rate): Declare.
8445 (ix86_adjust_cost): Declare.
8446 (ia32_multipass_dfa_lookahead): Declare.
8447 (ix86_macro_fusion_p): Declare.
8448 (ix86_macro_fusion_pair_p): Declare.
8449 (ix86_bd_has_dispatch): Declare.
8450 (ix86_bd_do_dispatch): Declare.
8451 (ix86_core2i7_init_hooks): Declare.
8452 (ix86_atom_sched_reorder): Declare.
8453 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
8454 (COSTS_N_BYTES): Move to x86-tune-costs.h.
8455 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
8456 (rip_relative_addr_p): Rename to ...
8457 (ix86_rip_relative_addr_p): ... this one; export.
8458 (memory_address_length): Update.
8459 (ix86_issue_rate): Move to x86-tune-sched.c.
8460 (ix86_flags_dependent): Move to x86-tune-sched.c.
8461 (ix86_agi_dependent): Move to x86-tune-sched.c.
8462 (exact_dependency_1): Move to x86-tune-sched.c.
8463 (exact_store_load_dependency): Move to x86-tune-sched.c.
8464 (ix86_adjust_cost): Move to x86-tune-sched.c.
8465 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
8466 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
8467 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
8468 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
8469 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
8470 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
8471 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
8472 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
8473 (min_insn_size): Rename to ...
8474 (ix86_min_insn_size): ... this one; export.
8475 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
8476 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
8477 (core2i7_first_cycle_multipass_backtrack): Move to
8478 x86-tune-sched-core.c.
8479 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
8480 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
8481 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
8482 (ix86_avoid_jump_mispredicts): Update.
8483 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
8484 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
8485 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
8486 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
8487 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
8488 (MAX_INSN): Move to ix86-tune-sched-bd.c.
8489 (MAX_IMM): Move to ix86-tune-sched-bd.c.
8490 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
8491 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
8492 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
8493 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
8494 (MAX_STORE): Move to ix86-tune-sched-bd.c.
8495 (BIG): Move to ix86-tune-sched-bd.c.
8496 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
8497 (enum insn_path): Move to ix86-tune-sched-bd.c.
8498 (get_mem_group): Move to ix86-tune-sched-bd.c.
8499 (is_cmp): Move to ix86-tune-sched-bd.c.
8500 (dispatch_violation): Move to ix86-tune-sched-bd.c.
8501 (is_branch): Move to ix86-tune-sched-bd.c.
8502 (is_prefetch): Move to ix86-tune-sched-bd.c.
8503 (init_window): Move to ix86-tune-sched-bd.c.
8504 (allocate_window): Move to ix86-tune-sched-bd.c.
8505 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
8506 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
8507 (process_end_window): Move to ix86-tune-sched-bd.c.
8508 (allocate_next_window): Move to ix86-tune-sched-bd.c.
8509 (find_constant): Move to ix86-tune-sched-bd.c.
8510 (get_num_immediates): Move to ix86-tune-sched-bd.c.
8511 (has_immediate): Move to ix86-tune-sched-bd.c.
8512 (get_insn_path): Move to ix86-tune-sched-bd.c.
8513 (get_insn_group): Move to ix86-tune-sched-bd.c.
8514 (count_num_restricted): Move to ix86-tune-sched-bd.c.
8515 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
8516 (add_insn_window): Move to ix86-tune-sched-bd.c.
8517 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
8518 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
8519 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
8520 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
8521 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
8522 (do_dispatch): Move to ix86-tune-sched-bd.c.
8523 (has_dispatch): Move to ix86-tune-sched-bd.c.
8524 * config/i386/t-i386: Add new object files.
8525 * config/i386/x86-tune-costs.h: New file.
8526 * config/i386/x86-tune-sched-atom.c: New file.
8527 * config/i386/x86-tune-sched-bd.c: New file.
8528 * config/i386/x86-tune-sched-core.c: New file.
8529 * config/i386/x86-tune-sched.c: New file.
8531 2017-10-11 Liu Hao <lh_mouse@126.com>
8533 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
8534 the generic version below instead.
8535 (should_colorize): Recognize Windows consoles as terminals
8537 * pretty-print.c [__MINGW32__] (write_all): New function.
8538 [__MINGW32__] (find_esc_head): Likewise.
8539 [__MINGW32__] (find_esc_terminator): Likewise.
8540 [__MINGW32__] (eat_esc_sequence): Likewise.
8541 [__MINGW32__] (mingw_ansi_fputs): New function that handles
8543 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
8546 2017-10-11 Richard Biener <rguenther@suse.de>
8548 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
8549 Properly call analyze_scalar_evolution with the loop of the stmt.
8551 2017-10-11 Richard Biener <rguenther@suse.de>
8553 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
8554 * tree-core.h (tree_base): Add chrec_var union member.
8555 * tree.h (CHREC_VAR): Remove.
8556 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
8557 * tree-chrec.h (build_polynomial_chrec): Adjust.
8558 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
8559 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
8561 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
8563 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
8564 * match.pd: ... here.
8565 ((T) X == (T) Y): Relax condition.
8567 2017-10-11 Bin Cheng <bin.cheng@arm.com>
8569 PR tree-optimization/82472
8570 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
8572 (break_alias_scc_partitions): Update postorder number.
8574 2017-10-11 Martin Liska <mliska@suse.cz>
8577 * opts.c (parse_no_sanitize_attribute): Do not use error_value
8579 * opts.h (parse_no_sanitize_attribute): Remove last argument.
8581 2017-10-11 Martin Liska <mliska@suse.cz>
8583 * print-rtl.c (print_insn): Move declaration of idbuf
8584 to same scope as name.
8586 2017-10-11 Martin Liska <mliska@suse.cz>
8591 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8594 2017-10-11 Martin Liska <mliska@suse.cz>
8597 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8600 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
8602 * config.gcc (default_gnu_indirect_function): Default to yes for
8603 arm*-*-linux* with glibc.
8605 2017-10-11 Richard Biener <rguenther@suse.de>
8607 * tree-scalar-evolution.c (get_scalar_evolution): Handle
8608 default-defs and types we do not want to analyze.
8609 (interpret_loop_phi): Replace unreachable code with an assert.
8610 (compute_scalar_evolution_in_loop): Remove and inline ...
8611 (analyze_scalar_evolution_1): ... here, replacing condition with
8612 what makes the intent clearer. Remove handling of cases
8613 get_scalar_evolution now handles.
8615 2017-10-10 Jim Wilson <wilson@tuliptree.org>
8617 PR rtl-optimization/81434
8618 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
8619 comment for main loop. In sched_group_found if, also add checks for
8620 pass and min_cost_group.
8622 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
8624 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
8625 (rs6000_insn_cost): New function.
8626 * config/rs6000/rs6000.md (cost): New attribute.
8628 2017-10-10 Jakub Jelinek <jakub@redhat.com>
8629 H.J. Lu <hongjiu.lu@intel.com>
8633 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
8634 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
8635 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
8636 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
8637 to OPTION_MASK_ISA_AVX512VL - builtins that have both
8638 OPTION_MASK_ISA_MMX and some other bit set require both
8639 mmx and the ISAs without the mmx bit.
8640 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
8641 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
8642 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
8643 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
8644 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
8645 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
8646 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
8647 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
8648 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
8649 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
8650 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
8651 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
8652 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
8653 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
8654 Add OPTION_MASK_ISA_MMX.
8656 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
8658 * config.gcc (armv7*-*-freebsd*): New target.
8659 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
8661 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
8663 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
8664 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
8667 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
8669 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
8670 a template parameter.
8671 (WIDE_INT_REF_FOR): Update accordingly.
8672 * tree.h (wi::int_traits <const_tree>): Delete.
8673 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
8674 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
8675 (wi::tree_to_wide_ref): New typedef.
8676 (wi::to_wide): New function.
8677 * calls.c (get_size_range): Use wi::to_wide when operating on
8679 * cgraph.c (cgraph_node::create_thunk): Likewise.
8680 * config/i386/i386.c (ix86_data_alignment): Likewise.
8681 (ix86_local_alignment): Likewise.
8682 * dbxout.c (stabstr_O): Likewise.
8683 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
8684 * expr.c (const_vector_from_tree): Likewise.
8685 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
8686 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
8687 (fold_negate_expr_1, int_const_binop_1, const_binop)
8688 (fold_convert_const_int_from_real, optimize_bit_field_compare)
8689 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
8690 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
8691 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
8692 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
8693 (fold_not_const, round_up_loc): Likewise.
8694 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
8695 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
8696 (alloca_call_type): Likewise.
8697 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
8698 * godump.c (go_output_typedef): Likewise.
8699 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
8700 * internal-fn.c (get_min_precision): Likewise.
8701 * ipa-cp.c (ipcp_store_vr_results): Likewise.
8702 * ipa-polymorphic-call.c
8703 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
8704 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
8705 (ipa_modify_call_arguments): Likewise.
8706 * match.pd: Likewise.
8707 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
8708 * print-tree.c (print_node_brief, print_node): Likewise.
8709 * stmt.c (expand_case): Likewise.
8710 * stor-layout.c (layout_type): Likewise.
8711 * tree-affine.c (tree_to_aff_combination): Likewise.
8712 * tree-cfg.c (group_case_labels_stmt): Likewise.
8713 * tree-data-ref.c (dr_analyze_indices): Likewise.
8714 (prune_runtime_alias_test_list): Likewise.
8715 * tree-dump.c (dequeue_and_dump): Likewise.
8716 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
8717 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
8718 * tree-pretty-print.c (dump_generic_node): Likewise.
8719 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
8720 (simple_iv_with_niters): Likewise.
8721 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
8722 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
8723 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
8724 * tree-ssa-loop-niter.c (split_to_var_and_offset)
8725 (refine_value_range_using_guard, number_of_iterations_ne_max)
8726 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
8727 (get_cst_init_from_scev, record_nonwrapping_iv)
8728 (scev_var_range_cant_overflow): Likewise.
8729 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
8730 * tree-ssa-pre.c (compute_avail): Likewise.
8731 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
8732 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
8733 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
8734 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
8735 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
8736 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
8737 (dump_case_nodes, try_switch_expansion): Likewise.
8738 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
8739 (vect_do_peeling): Likewise.
8740 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
8741 * tree-vect-stmts.c (vectorizable_load): Likewise.
8742 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
8743 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
8744 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
8745 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
8746 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
8747 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
8748 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
8749 (evrp_dom_walker::before_dom_children): Likewise.
8750 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
8751 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
8752 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
8753 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
8754 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
8755 (get_range_pos_neg): Likewise.
8756 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
8757 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
8758 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
8759 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
8760 * config/avr/avr.c (avr_fold_builtin): Likewise.
8761 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
8762 * config/msp430/msp430.c (msp430_attr): Likewise.
8763 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
8764 * config/powerpcspe/powerpcspe-c.c
8765 (altivec_resolve_overloaded_builtin): Likewise.
8766 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
8767 (rs6000_expand_ternop_builtin): Likewise.
8768 * config/rs6000/rs6000-c.c
8769 (altivec_resolve_overloaded_builtin): Likewise.
8770 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
8771 (rs6000_expand_ternop_builtin): Likewise.
8772 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
8774 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8776 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
8777 when copying loop nest with only one inner loop.
8779 2017-10-10 Richard Biener <rguenther@suse.de>
8781 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
8782 blocks if SCEV is active.
8783 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
8785 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
8786 (scev_initialize): Assert we are not yet initialized.
8788 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8790 * tree-loop-distribution.c (generate_loops_for_partition): Remove
8791 inner loop's exit stmt by making it always exit the loop, otherwise
8792 we would generate an infinite empty loop.
8794 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8796 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
8797 renaming variables in new preheader if it's deleted.
8799 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8801 * tree-loop-distribution.c (struct partition): Remove unused field
8802 loops of the structure.
8803 (partition_alloc, partition_free): Ditto.
8804 (build_rdg_partition_for_vertex): Ditto.
8806 2017-10-09 Jeff Law <law@redhat.com>
8808 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
8809 return type to match prototype and documentation.
8811 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8813 * config/rs6000/rs6000.c (processor_costs): Move to ...
8814 * config/rs6000/rs6000.h: ... here.
8815 (rs6000_cost): Declare.
8817 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
8819 * except.c (setjmp_fn): New global variable.
8820 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
8821 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
8822 if DONT_USE_BUILTIN_SETJMP is defined.
8824 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8826 * target.def (insn_cost): New hook.
8827 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
8828 * doc/tm.texi: Regenerate.
8829 * rtlanal.c (insn_cost): Use the new hook.
8831 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8833 * combine.c (combine_validate_cost): Compute the new insn_cost,
8834 not just pattern_cost.
8835 (try_combine): Adjust comment.
8837 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8839 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
8841 * combine.c (uid_insn_cost): Adjust comment.
8842 (combine_validate_cost): Adjust comment. Use pattern_cost instead
8844 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
8845 * dse.c (find_shift_sequence): Ditto.
8846 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
8847 (bb_valid_for_noce_process_p): Use pattern_cost.
8848 * rtl.h (insn_rtx_cost): Delete.
8849 (pattern_cost): New prototype.
8850 (insn_cost): New prototype.
8851 * rtlanal.c (insn_rtx_cost): Rename to...
8852 (pattern_cost): ... this.
8855 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
8857 * config/i386/i386.md (*jcc_2): Remove insn pattern.
8858 (*jcc<mode>_0_r_i387): Ditto.
8859 (*jccxf_r_i387): Ditto.
8860 (*jcc<mode>_r_i387): Ditto.
8861 (*jccu<mode>_r_i387): Ditto.
8862 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
8863 (*jcc): Rename from *jcc_1.
8865 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8867 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
8868 deferred rescans after the lvx/stvx recombination pre-pass.
8870 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
8872 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
8873 memory operation instruction support.
8874 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
8875 (rs6000-ibm-aix[789]*): Likewise.
8876 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
8877 Document new functions.
8879 2017-10-09 Richard Biener <rguenther@suse.de>
8881 PR tree-optimization/82397
8882 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
8883 equality only for semantically equal trees.
8885 2017-10-09 Richard Biener <rguenther@suse.de>
8887 PR tree-optimization/82449
8888 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
8889 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
8890 allow constant addresses.
8891 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
8894 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8896 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
8899 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8902 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
8905 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8908 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
8910 2017-10-09 Jakub Jelinek <jakub@redhat.com>
8913 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
8914 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
8916 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
8918 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
8919 (WI_BINARY_PREDICATE_RESULT): Likewise.
8920 (wi::binary_traits::operator_result): New type.
8921 (wi::binary_traits::predicate_result): Likewise.
8922 (generic_wide_int::operator~, unary generic_wide_int::operator-)
8923 (generic_wide_int::operator==, generic_wide_int::operator!=)
8924 (generic_wide_int::operator&, generic_wide_int::and_not)
8925 (generic_wide_int::operator|, generic_wide_int::or_not)
8926 (generic_wide_int::operator^, generic_wide_int::operator+
8927 (binary generic_wide_int::operator-, generic_wide_int::operator*):
8929 (operator~, unary operator-, operator==, operator!=, operator&)
8930 (operator|, operator^, operator+, binary operator-, operator*): New
8932 * expr.c (get_inner_reference): Use wi::bit_and_not.
8933 * fold-const.c (fold_binary_loc): Likewise.
8934 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
8935 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
8936 (bit_value_binop): Likewise.
8937 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
8938 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
8939 (extract_range_from_binary_expr_1): Likewise.
8940 (masked_increment): Likewise.
8941 (simplify_bit_ops_using_ranges): Likewise.
8943 2017-10-09 Martin Jambor <mjambor@suse.cz>
8946 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
8947 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
8948 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
8950 (hsa_fixup_mov_insn_type): New function.
8951 (hsa_op_with_type::get_in_type): Use it.
8952 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
8953 immediates in an assert.
8954 (hsa_op_with_type::extend_int_to_32bit): New method.
8955 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
8956 types. Convert to dest type if necessary.
8957 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
8958 (reg_for_gimple_ssa): Pass false as min32int to
8959 hsa_type_for_scalar_tree_type.
8960 (gen_hsa_addr): Fixup type when creating addresable temporary.
8961 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
8962 (gen_hsa_unary_operation): Extend operands and convert to dest type if
8963 necessary. Call hsa_fixup_mov_insn_type.
8964 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
8965 extend operands and convert to dest type if necessary.
8966 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
8967 to dest type if necessary.
8968 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
8969 if conversion nt necessary and size matches.
8970 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
8971 to dest type if necessary.
8972 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
8973 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
8975 (gen_hsa_clrsb): Likewise.
8976 (gen_hsa_ffs): Likewise.
8977 (gen_hsa_divmod): Extend operands and convert to dest type if
8979 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
8981 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8983 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
8984 Remove empty default arguments. Use a brace block as output
8986 (conditional return): Ditto.
8988 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
8989 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
8991 (group_ending_nop): Ditto.
8992 (doloop_end): Ditto.
8993 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
8994 (splitters for those): Ditto.
8996 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8998 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
8999 a conditional jump (and the compare for it) so that pc_rtx is the
9001 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
9002 for the deleted and renamed ctr<mode>_internal[234] patterns.
9003 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
9004 Delete second conditional return pattern.
9005 (ctr<mode>_internal2): Delete this second bdnz pattern.
9006 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
9007 (ctr<mode>_internal4): Delete this second bdz pattern.
9009 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9011 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
9012 (always_initialized_rtx_for_ssa_name_p): New predicate.
9013 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
9014 (finish_out_of_ssa): Free new field of SA.
9015 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
9016 * tree-ssa-coalesce.c: Include tree-ssa.h.
9017 (get_parm_default_def_partitions): Remove extern keyword.
9018 (get_undefined_value_partitions): New function.
9019 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
9020 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
9022 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
9024 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
9026 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
9028 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9030 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
9031 for targets that preffer 128bit.
9033 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9035 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
9037 2017-10-08 Olivier Hainque <hainque@adacore.com>
9039 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
9040 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
9042 (thumb_set_return_address): Likewise.
9044 2017-10-08 Olivier Hainque <hainque@adacore.com>
9046 * common/config/arm/arm-common.c (arm_except_unwind_info):
9047 Handle DWARF2_UNWIND_INFO.
9049 2017-10-07 Michael Collison <michael.collison@arm.com>
9051 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9054 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9056 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9057 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9058 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9059 defined, force the creation of a new block for a dispatch label.
9061 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9063 * invoke.texi (Wsuggest-attribute=cold): Document.
9064 * common.opt (Wsuggest-attribute=cold): New
9065 * ipa-pure-const.c (warn_function_cold): New function.
9066 * predict.c (compute_function_frequency): Use it.
9067 * predict.h (warn_function_cold): Declare.
9069 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9071 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9074 2017-10-06 Martin Liska <mliska@suse.cz>
9076 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9077 keyword for member functions.
9078 (struct sanopt_tree_couple): New struct.
9079 (struct sanopt_tree_couple_hash): New function.
9080 (struct sanopt_ctx): Add new hash_map.
9081 (has_dominating_ubsan_ptr_check): New function.
9082 (record_ubsan_ptr_check_stmt): Likewise.
9083 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9084 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9085 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9087 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9090 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9091 aarch64_simd_valid_immediate on CONST_VECTORs.
9092 (aarch64_reg_or_bic_imm): Likewise.
9094 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9096 PR rtl-optimization/82396
9097 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9099 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9101 * graphite-dependences.c (scop_get_reads): Move code to...
9102 (scop_get_must_writes): Move code to...
9103 (scop_get_may_writes): Move code to...
9104 (scop_get_reads_and_writes): ... here.
9105 (scop_get_dependences): Call scop_get_reads_and_writes.
9107 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9109 PR tree-optimization/82434
9110 * fold-const.h (can_native_encode_type_p,
9111 can_native_encode_string_p): Remove.
9112 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9113 don't encode anything, just return what would be otherwise returned.
9114 (native_encode_fixed, native_encode_complex, native_encode_vector):
9116 (native_encode_string): Likewise. Inline by hand
9117 can_native_encode_string_p.
9118 (can_native_encode_type_p): Remove.
9119 (can_native_encode_string_p): Remove.
9120 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
9121 STRING_CSTs using can_native_encode_string_p, test all
9122 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
9123 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
9124 argument from native_encode_expr.
9125 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
9126 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
9129 2017-10-06 Richard Biener <rguenther@suse.de>
9131 PR tree-optimization/82397
9132 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
9133 operand_equal_p but rely on data_ref_compare_tree for detecting
9135 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
9136 to match up with dr_group_sort_cmp.
9138 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9141 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
9143 * config/s390/s390-builtin-types.def: Regenerate.
9145 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9148 * config/s390/s390-builtin-types.def: Regenerate.
9149 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
9150 Change flag from B_VXE to B_VX.
9151 (s390_vec_min_dbl): Remove B_VXE flag.
9153 2017-10-06 Richard Biener <rguenther@suse.de>
9155 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
9156 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
9157 translate_isl_ast_to_gimple::is_valid_rename,
9158 translate_isl_ast_to_gimple::get_rename,
9159 translate_isl_ast_to_gimple::get_def_bb_for_const,
9160 translate_isl_ast_to_gimple::get_new_name,
9161 translate_isl_ast_to_gimple::collect_all_ssa_names,
9162 translate_isl_ast_to_gimple::copy_loop_phi_args,
9163 translate_isl_ast_to_gimple::collect_all_ssa_names,
9164 translate_isl_ast_to_gimple::copy_loop_phi_args,
9165 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
9166 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
9167 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
9168 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
9169 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
9170 translate_isl_ast_to_gimple::copy_cond_phi_args,
9171 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
9172 translate_isl_ast_to_gimple::edge_for_new_close_phis,
9173 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
9174 translate_isl_ast_to_gimple::rename_uses,
9175 translate_isl_ast_to_gimple::rename_all_uses): Remove.
9176 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
9177 (set_rename_for_each_def): Likewise.
9178 (graphite_copy_stmts_from_block): Handle debug stmt resetting
9179 here. Handle rewriting SCEV analyzable uses here.
9180 (copy_bb_and_scalar_dependences): Generate code for PHI
9182 (graphite_regenerate_ast_isl): Adjust.
9183 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
9184 (add_write, add_read): New functions.
9185 (build_cross_bb_scalars_def): Use it and simplify.
9186 (build_cross_bb_scalars_use): Likewise.
9187 (graphite_find_cross_bb_scalar_vars): Inline into...
9188 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
9189 simulating out-of-SSA. Compute liveout and add dependencies.
9190 (build_scops): Force an empty entry block.
9191 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
9193 (sese_build_liveouts): Declare.
9194 (sese_trivially_empty_bb_p): Likewise.
9195 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
9196 compute liveout and debug_liveout.
9197 (sese_bad_liveouts_use): Remove.
9198 (sese_reset_debug_liveouts_bb): Likewise.
9199 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
9200 (sese_build_liveouts): Build liveout and debug_liveout and store
9202 (new_sese_info): Adjust.
9203 (free_sese_info): Likewise.
9204 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
9205 do not build liveout here.
9206 (move_sese_in_condition): Adjust region entry.
9207 (scev_analyzable_p): Match up with chrec_apply requirements.
9208 (sese_trivially_empty_bb_p): New.
9209 * tree-into-ssa.c (get_reaching_def): Properly support generating
9210 default-defs for incremental rewrite of anonymous names.
9212 2017-10-06 Richard Biener <rguenther@suse.de>
9214 * graphite-sese-to-poly.c (extract_affine): For casts increasing
9215 precision do not perform modulo reduction.
9217 2017-10-06 Richard Biener <rguenther@suse.de>
9219 PR tree-optimization/82436
9220 * tree-vect-slp.c (vect_supported_load_permutation_p): More
9221 conservatively choose the vectorization factor when checking
9222 whether we can perform the required load permutation.
9223 (vect_transform_slp_perm_load): Assert when we may not fail.
9225 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
9227 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
9228 message for incompatible -msdata=* and -mcall-* options.
9230 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9232 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
9233 rate for post-reload scheduling.
9235 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9237 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
9239 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9241 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
9242 to improve monte carlo in scimark.
9244 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9246 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
9247 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
9248 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
9249 pentium4_cost, nocona_cost): Set reassociation width to 1.
9250 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
9251 width to 2 for fp operations and 1 otherwise.
9252 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
9254 (atom_cost): Set reassociation width to 2.
9255 (slm_cost, generic_cost): Set fp reassociation width
9256 to 2 and 1 otherwise.
9257 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
9258 (core_cost): Set fp reassociation width to 4 and vector to 2.
9259 (ix86_reassociation_width): Rewrite using cost table; special case
9260 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
9261 and TARGET_AVX128_OPTIMAL.
9262 * config/i386/i386.h (processor_costs): Add
9263 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
9264 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
9265 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
9266 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
9267 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
9268 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
9270 2017-10-05 Nathan Sidwell <nathan@acm.org>
9272 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
9274 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9276 * config/arm/arm.c (arm_test_fpu_data): New.
9277 (arm_run_selftests): Call arm_test_fpu_data.
9279 2017-10-04 Nathan Sidwell <nathan@acm.org>
9281 * toplev.c (toplev::main): Remove excess parens on pretty_printer
9283 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
9285 2017-10-04 Sudakshina Das <sudi.das@arm.com>
9287 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
9288 check type for aarch64_simd_valid_immediate.
9289 (aarch64_output_simd_mov_immediate): Update prototype.
9290 (aarch64_simd_valid_immediate): Update prototype.
9291 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
9292 support for ORR-immediate.
9293 (and<mode>3): modified pattern to add support for BIC-immediate.
9294 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
9295 now checks for valid immediate for BIC and ORR based on new enum
9297 (aarch64_output_simd_mov_immediate): Function now used to output
9298 BIC/ORR imm as well based on new enum argument.
9299 * config/aarch64/constraints.md (Do): New vector immediate constraint.
9301 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
9302 (aarch64_reg_or_bic_imm): Likewise.
9304 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9306 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
9307 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
9309 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9313 PR rtl-optimization/82396
9314 * haifa-sched.c (autopref_multipass_init): Simplify
9316 (autopref_rank_data): Simplify sort order.
9317 * sched-int.h (autopref_multipass_data_): Remove
9318 multi_mem_insn_p, min_offset and max_offset.
9320 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9322 * doc/sourcebuild.texi: Document vect_peeling_profitable.
9324 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9326 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
9329 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9331 * doc/sourcebuild.texi: Document vect_long_mult.
9333 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
9335 PR tree-optimization/82413
9336 * fold-const.c (build_range_check): Use widest_int when comparing
9337 the maximum ETYPE value with HIGH.
9339 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9341 PR rtl-optimization/82396
9342 * haifa-sched.c (autopref_multipass_init): Simplify
9344 (autopref_rank_data): Simplify sort order.
9345 * sched-int.h (autopref_multipass_data_): Remove
9346 multi_mem_insn_p, min_offset and max_offset.
9348 2017-10-04 Jakub Jelinek <jakub@redhat.com>
9350 PR tree-optimization/82381
9351 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
9352 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
9355 PR tree-optimization/82374
9356 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
9357 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
9358 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
9359 current_function_decl to the new decl.
9361 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
9363 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
9364 helper macro for IEEE float128 hardware built-in functions.
9365 (SQRTF128_ODD): Add built-in functions with the round-to-odd
9367 (TRUNCF128_ODD): Likewise.
9368 (ADDF128_ODD): Likewise.
9369 (SUBF128_ODD): Likewise.
9370 (MULF128_ODD): Likewise.
9371 (DIVF128_ODD): Likewise.
9372 (FMAF128_ODD): Likewise.
9373 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
9374 UNSPEC_TRUNC_ROUND_TO_ODD.
9375 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
9376 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
9377 floating point round to odd instructions.
9378 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
9379 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
9380 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
9381 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
9382 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
9383 (trunc<mode>sf2_hw): Change the truncate with round to odd
9384 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
9385 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
9386 to odd hardware instructions.
9387 (sub<mode>3_odd): Likewise.
9388 (mul<mode>3_odd): Likewise.
9389 (div<mode>3_odd): Likewise.
9390 (sqrt<mode>2_odd): Likewise.
9391 (fma<mode>4_odd): Likewise.
9392 (fms<mode>4_odd): Likewise.
9393 (nfma<mode>4_odd): Likewise.
9394 (nfms<mode>4_odd): Likewise.
9395 (trunc<mode>df2_odd): Change the truncate with round to odd
9396 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
9398 * doc/extend.texi (PowerPC built-in functions): Update documentation
9399 for existing IEEE float128-bit built-in functions. Add built-in
9400 functions that generate the IEEE 128-bit floating point round to
9403 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
9405 PR rtl-optimization/77729
9406 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
9407 to (X&(C1&~C2))|C2 transformations.
9409 2017-10-03 Martin Jambor <mjambor@suse.cz>
9411 PR tree-optimization/82363
9412 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
9413 mismatch, mark lacc written regardless of racc.
9415 2017-10-03 Jakub Jelinek <jakub@redhat.com>
9417 PR tree-optimization/82381
9418 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
9419 stmt_to_insert nor wheather SSA_NAMEs are default defs.
9420 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
9421 fallthrough into reassoc_stmt_dominates_stmt_p.
9424 * combine.c (combine_instructions): Don't combine in unreachable
9427 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9430 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
9431 function to not use the have_cpu variable. Do not set cpu_index,
9432 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
9434 (rs6000_valid_attribute_p): Remove duplicate initializations of
9435 old_optimize and func_optimize.
9436 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
9437 (rs6000_activate_target_options): Make global.
9438 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
9441 2017-10-02 Jakub Jelinek <jakub@redhat.com>
9443 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
9444 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
9445 Set *poffset to 0 and *psize and *pmax_size to -1 if
9446 *poffset + *psize overflows in HOST_WIDE_INT.
9448 PR tree-optimization/82387
9449 PR tree-optimization/82388
9450 PR tree-optimization/82389
9451 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
9452 instead of live_bytes non-NULL.
9454 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
9457 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
9460 2017-10-02 Richard Biener <rguenther@suse.de>
9462 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
9463 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
9464 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
9466 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9468 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
9469 requested precision matches the type's.
9470 * calls.c (alloc_max_size): Calculate the new candidate size as
9471 a widest_int and use wi::to_widest when comparing it with the
9472 current candidate size.
9473 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
9474 zero rather than integer_zero_node.
9475 * match.pd: Check for a no-op conversion before using wi::add
9476 rather than after. Use tree_to_uhwi when summing small shift
9477 counts into an unsigned int.
9479 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9480 Alan Hayward <alan.hayward@arm.com>
9481 David Sherwood <david.sherwood@arm.com>
9484 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
9485 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
9486 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
9487 POINTER_AND_FP_REGS.
9489 2017-10-02 Richard Biener <rguenther@suse.de>
9491 PR tree-optimization/82355
9492 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9493 a mapping for the enclosing loop but avoid generating one for
9495 (copy_bb_and_scalar_dependences): Remove premature codegen
9496 error on PHIs in blocks duplicated into multiple places.
9497 * graphite-scop-detection.c
9498 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9499 in the region use it as loop and nest to analyze the DR in.
9500 (try_generate_gimple_bb): Likewise.
9501 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9502 (add_loop_constraints): For blocks in a loop not in the region
9503 create a dimension with a single iteration.
9504 * sese.h (gbb_loop_at_index): Remove assert.
9506 2017-10-01 Kevin Buettner <kevinb@redhat.com>
9508 * omp-expand.c (adjust_context_scope): New function.
9509 (expand_parallel_call): Call adjust_context_scope.
9511 2017-10-01 Jeff Law <law@redhat.com>
9513 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
9514 dom_opt_dom_walker class with direct access to private members.
9515 Add comments. Call test_for_singularity.
9516 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
9517 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
9518 m_dummy_cond anymore.
9519 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
9521 (pass_dominator:execute): Build the dummy_cond here and pass it
9522 to the dom_opt_dom_walker ctor.
9523 (test_for_singularity): New function.
9525 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
9526 Maya Rashish <coypu@sdf.org>
9528 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
9529 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
9530 (alpha*-*-netbsd*) Use nbsd_tm_file.
9531 (arm*-*-netbsdelf*) Likewise.
9532 (i[34567]86-*-netbsdelf*) Likewise.
9533 (x86_64-*-netbsd*) Likewise.
9534 (mips*-*-netbsd*) Likewise.
9535 (powerpc-*-netbsd*) Likewise.
9536 (sh*-*-netbsd*) Likewise.
9537 (sparc-*-netbsdelf*) Likewise.
9538 (sparc64-*-netbsd*) Likewise.
9539 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
9541 (vax-*-netbsdelf*) Likewise.
9542 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
9543 (UINT_FAST8_TYPE) Likewise.
9544 (INT_FAST16_TYPE) Check CHAR_FAST16.
9545 (UINT_FAST16_TYPE) Likewise.
9547 2017-09-30 Jakub Jelinek <jakub@redhat.com>
9550 * config/i386/i386.md
9551 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
9552 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
9553 *divmodsi4_zext_2): New define_insn_and_split.
9554 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
9555 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
9556 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
9557 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
9558 New define_insn_and_split.
9559 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
9560 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
9561 operands[1] having DImode when mode is SImode.
9563 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
9564 always SImode for DIV and MOD in REG_EQUAL notes.
9566 2017-09-29 Yury Gribov <tetra2005@gmail.com>
9569 * match.pd: Fix handling of NaNs in pattern.
9571 2017-09-29 Jeff Law <law@redhat.com>
9573 * sbitmap.c (bitmap_bit_in_range_p): New function.
9574 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
9575 * tree-ssa-dse.c (live_bytes_read): New function.
9576 (dse_classify_store): Ignore reads of dead bytes.
9578 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
9579 typos and whitespace errors.
9580 * config/i386/predicates.md (address_no_seg_operand): Likewise.
9581 * config/s390/s390.c (s390_emit_prologue): Likewise.
9583 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9586 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
9587 with a symbol for LRA.
9589 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9591 PR rtl-optimization/82338
9592 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
9594 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
9596 * genmodes.c (calc_wider_mode): Suppress qsort macro.
9597 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
9598 (qsort_chk): Declare.
9599 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
9600 (qsort_chk): New function.
9602 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9604 PR tree-optimization/82337
9605 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9606 phi definition if the PHI result appears in an abnormal PHI.
9607 (find_basis_for_base_expr): Don't record a basis if the LHS of the
9608 basis appears in an abnormal PHI.
9610 2017-09-29 Richard Biener <rguenther@suse.de>
9612 * graphite-isl-ast-to-gimple.c
9613 (translate_isl_ast_to_gimple::set_codegen_error): New function.
9614 (binary_op_to_tree): Use it.
9615 (get_rename_from_scev): Likewise.
9616 (copy_loop_phi_nodes): Likewise.
9617 (copy_bb_and_scalar_dependences): Likewise.
9618 (translate_pending_phi_nodes): Likewise.
9620 2017-09-29 Jakub Jelinek <jakub@redhat.com>
9623 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
9624 for movabsq $(i32 << shift), r64.
9626 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
9628 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
9629 index when encoding %esp as %rsp to avoid 0x67 prefix.
9631 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
9633 * config/i386/i386.md (*movsf_internal, *movdf_internal):
9634 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
9636 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
9638 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
9639 Extensions with more than 16 double VFP registers.
9640 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
9641 to_clear_mask and all code related to it. Replace the remaining
9642 entry by a sbitmap and adapt code accordingly.
9644 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
9646 * brig-builtins.def: Change pure attributes to const.
9648 2017-09-28 Joseph Myers <joseph@codesourcery.com>
9650 * config.gcc (default_gnu_indirect_function): Default to yes for
9651 sparc*-*-linux* with glibc.
9653 2017-09-28 Joseph Myers <joseph@codesourcery.com>
9655 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
9656 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
9657 when creating .init_array and .fini_array sections with priority
9660 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
9663 * config/aarch64/aarch64.c
9664 (aarch64_builtin_support_vector_misalignment): Always return false
9665 when misalignment is unknown.
9667 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
9669 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
9670 this function to return false if the definition used by the swap
9671 instruction is artificial, or if the memory address from which the
9672 constant value is loaded is not represented by a base address held
9673 in a register or if the base address register is a frame or stack
9674 pointer. Additionally, return false if the base address of the
9675 loaded constant is a SYMBOL_REF but is not considered to be a
9677 (replace_swapped_load_constant): New function.
9678 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
9679 loaded constant vector with a load of a swapped constant vector.
9681 2017-09-27 Carl Love <cel@us.ibm.com>
9683 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
9684 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
9685 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
9688 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
9690 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
9691 first, always call autopref_rank_data otherwise.
9693 2017-09-27 Richard Biener <rguenther@suse.de>
9695 * graphite-scop-detection.c (find_scop_parameters): Move
9696 loop bound handling ...
9697 (gather_bbs::before_dom_children): ... here, avoiding the need
9698 to build scop_info->loop_nest.
9699 (record_loop_in_sese): Remove.
9700 * sese.h (sese_info_t::loop_nest): Remove.
9701 * sese.c (new_sese_info): Do not allocate loop_nest.
9702 (free_sese_info): Do not free loop_nest.
9704 2017-09-27 Jakub Jelinek <jakub@redhat.com>
9707 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
9708 lhs from calls if the lhs has addressable type.
9710 2017-09-27 Richard Biener <rguenther@suse.de>
9712 * graphite.h (scop::max_alias_set): New member.
9713 * graphite-scop-detection.c: Remove references to non-existing
9714 --param in comments.
9715 (build_alias_sets): Record the maximum alias set used for drs.
9716 (build_scops): Support zero as unlimited for
9717 --param graphite-max-arrays-per-scop.
9718 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
9720 (build_poly_sr_1): ... here. Compute alias set based on the
9721 maximum alias set used for drs rather than
9722 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
9724 2017-09-27 Richard Biener <rguenther@suse.de>
9726 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
9727 --param loop-block-tile-size=0 to disable tiling.
9729 2017-09-27 Richard Biener <rguenther@suse.de>
9731 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
9732 (graphite-max-nb-scop-params): Document special value zero.
9733 * domwalk.h (dom_walker::STOP): New symbolical constant.
9734 (dom_walker::dom_walker): Add optional parameter for bb to
9736 (dom_walker::~dom_walker): Declare.
9737 (dom_walker::before_dom_children): Document STOP return value.
9738 (dom_walker::m_user_bb_to_rpo): New member.
9739 (dom_walker::m_bb_to_rpo): Likewise.
9740 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
9741 mapping here if not provided by the user.
9742 (dom_walker::~dom_walker): Free bb to RPO mapping if not
9743 provided by the user.
9744 (dom_walker::STOP): Define.
9745 (dom_walker::walk): Do not compute bb to RPO mapping here.
9746 Support STOP return value from before_dom_children to stop
9748 * graphite-optimize-isl.c (optimize_isl): If the schedule
9749 is the same still generate code if -fgraphite-identity
9750 or -floop-parallelize-all are given.
9751 * graphite-scop-detection.c: Include cfganal.h.
9752 (gather_bbs::gather_bbs): Get and pass through bb to RPO
9754 (gather_bbs::before_dom_children): Return STOP for BBs
9756 (build_scops): Compute bb to RPO mapping and pass it to
9757 the domwalk. Treat --param graphite-max-nb-scop-params=0
9758 as not limiting the number of params.
9759 * graphite.c (graphite_initialize): Remove limit on the
9760 number of basic-blocks in a function.
9761 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
9762 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
9763 default value of 10.
9765 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9767 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
9768 Adjust code to eliminate needing to do the shift right 32-bits
9769 operation after XSCVDPSPN.
9771 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9773 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
9774 ((X / Y) != 0 -> X >= Y): Likewise.
9776 2017-09-26 Carl Love <cel@us.ibm.com>
9778 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
9779 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
9780 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
9781 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
9782 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
9783 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
9784 definitions and overloading.
9785 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
9786 statement for P9V_BUILTIN_XST_LEN_R.
9787 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
9788 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
9789 define_expand and define_insn for the instructions and builtins.
9790 * doc/extend.texi: Update the built-in documentation file for the new
9792 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
9793 define_insn for the instructions
9795 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
9798 * gcc/config/netbsd-protos.h: New file.
9799 * gcc/config/netbsd.c: New file.
9800 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
9801 * gcc/config/t-netbsd: New file.
9802 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
9803 (tmake_file) Add t-netbsd.
9804 (extra_objs) Add netbsd.o.
9806 2017-09-26 Janus Weil <janus@gcc.gnu.org>
9810 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
9812 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9814 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
9815 sign extension from a vector register to a GPR by doing a 32-bit
9816 direct move and then an EXTSW.
9817 (extendsi<mode>2 splitter): Likewise.
9818 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
9819 right or vector extract after doing XSCVDPSPN. Use
9820 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
9822 (movdi_from_sf_zero_ext): Likewise.
9823 (reload_gpr_from_vsxsf): Likewise.
9824 (p8_mfvsrd_4_disf): Delete, no longer used.
9825 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
9826 then needing to move the SFmode to a GPR to use the XSCVDPSP
9827 instruction instead of FRSP and XSCVDPSPN.
9828 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
9829 it is adjacent to the other XSCVSPDP insns.
9830 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
9831 SFmode to be in traditional Altivec registers.
9832 (vsx_xscvdpspn): Eliminate useless alternative constraint.
9833 (vsx_xscvspdpn): Likewise.
9834 (vsx_xscvspdpn_scalar): Likewise.
9836 2017-09-26 Martin Jambor <mjambor@suse.cz>
9838 * tree-sra.c (compare_access_positions): Put integral types first,
9839 stabilize sorting of integral types, remove conditions putting
9840 non-full-precision integers last.
9841 (sort_and_splice_var_accesses): Disable scalarization if a
9842 non-integert would be represented by a non-full-precision integer.
9844 2017-09-26 Joseph Myers <joseph@codesourcery.com>
9846 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
9847 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
9848 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
9849 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
9850 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
9851 conditionals inside the function instead of around it. Call
9852 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
9853 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
9855 2017-09-26 Richard Biener <rguenther@suse.de>
9857 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
9859 (scop_detection::build_scop_breadth): ... this. Removed.
9860 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
9861 (scop_detection::harmful_stmt_in_bb): Likewise.
9862 (scop_detection::graphite_can_represent_stmt): Likewise.
9863 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
9864 (scop_detection::can_represent_loop): Remove recursion, fold in ...
9865 (scop_detection::can_represent_loop_1): ... this. Removed.
9866 (scop_detection::harmful_loop_in_region): Simplify after inlining
9867 the above and remove more quadraticness.
9868 (build_scops): Adjust.
9869 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
9872 2017-09-26 Jakub Jelinek <jakub@redhat.com>
9875 * config/i386/i386.c (ix86_print_operand_address_as): Only test
9876 REGNO (base) == SP_REG if base is a REG.
9879 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
9880 if it is different SSA_NAME.
9881 (optimize_range_tests_cmp_bitwise): New function.
9882 (optimize_range_tests): Call it.
9884 2017-09-26 Richard Biener <rguenther@suse.de>
9886 PR tree-optimization/82321
9887 * graphite.c (canonicalize_loop_closed_ssa): Properly check
9888 for the def being inside the loop.
9890 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9892 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
9894 * config/s390/s390-builtins.def: Fix constraint on op4.
9896 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9898 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
9899 independent expanders.
9900 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
9901 ("vec_ordered", "vec_unordered"): New expanders.
9903 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9905 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
9908 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9910 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
9911 vec_unpacks_lo_v16qi.
9912 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
9914 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9916 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
9917 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
9918 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
9920 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9922 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
9924 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
9926 ("*vec_slb<mode>"): New insn pattern.
9927 ("vec_shr_<mode>"): New expander.
9928 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
9929 and force the shift count operand to V16QImode.
9930 ("vec_srb<mode>"): Set shift count mode to V16QI.
9932 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9934 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
9935 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
9936 ("vec_widen_smult_hi_<mode>"): New expander definitions.
9938 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
9941 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
9943 2017-09-26 Richard Biener <rguenther@suse.de>
9945 PR tree-optimization/82320
9946 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
9949 2017-09-25 Jeff Law <law@redhat.com>
9951 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
9952 prototype for new argument.
9953 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
9954 mostly extracted from rs6000_emit_allocate_stack.
9955 (rs6000_emit_probe_stack_range_stack_clash): New function.
9956 (rs6000_emit_allocate_stack): Call
9957 rs6000_emit_probe_stack_range_stack_clash as needed.
9958 (rs6000_emit_probe_stack_range): Add additional argument
9959 to call to gen_probe_stack_range{si,di}.
9960 (output_probe_stack_range): New.
9961 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
9962 (output_probe_stack_range_stack_clash): New.
9963 (rs6000_emit_prologue): Emit notes into dump file as requested.
9964 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
9965 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
9966 Add additional operand and pass it to output_probe_stack_range.
9968 2017-09-25 Bin Cheng <bin.cheng@arm.com>
9970 PR tree-optimization/82163
9971 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
9972 (checking_verify_loop_closed_ssa): New parameter.
9973 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
9974 (check_loop_closed_ssa_stmt): Delete.
9975 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
9976 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
9977 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
9980 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
9982 * brig-builtins.def: Treat HSAIL barrier builtins as
9983 setjmp/longjump style functions.
9985 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9987 * target.def (constant_alignment): New hook.
9988 * defaults.h (CONSTANT_ALIGNMENT): Delete.
9989 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
9990 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
9991 * doc/tm.texi: Regenerate.
9992 * targhooks.h (default_constant_alignment): Declare.
9993 (constant_alignment_word_strings): Likewise.
9994 * targhooks.c (default_constant_alignment): New function.
9995 (constant_alignment_word_strings): Likewise.
9996 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
9997 instead of CONSTANT_ALIGNMENT.
9998 * varasm.c (align_variable, get_variable_align, build_constant_desc)
9999 (force_const_mem): Likewise.
10000 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
10001 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
10002 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
10003 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10004 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
10006 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
10007 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10008 constant_alignment_word_strings.
10009 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
10010 (CONSTANT_ALIGNMENT): Likewise.
10011 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10012 (arm_constant_alignment): New function.
10013 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
10014 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10015 constant_alignment_word_strings.
10016 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
10017 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10018 constant_alignment_word_strings.
10019 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
10020 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10021 (cris_constant_alignment): New function.
10022 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
10023 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10024 (epiphany_constant_alignment): New function.
10025 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
10026 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10027 constant_alignment_word_strings.
10028 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
10029 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10030 constant_alignment_word_strings.
10031 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
10032 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10033 constant_alignment_word_strings.
10034 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
10035 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
10036 * config/i386/i386.c (ix86_constant_alignment): Make static.
10037 Use the same interface as the target hook.
10038 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10039 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
10040 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10041 constant_alignment_word_strings.
10042 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
10043 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
10044 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10045 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
10046 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10047 constant_alignment_word_strings.
10048 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
10049 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10050 constant_alignment_word_strings.
10051 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
10052 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10053 constant_alignment_word_strings.
10054 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10055 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10057 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10058 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10059 * config/mips/mips.c (mips_constant_alignment): New function.
10060 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10061 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10062 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10063 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10064 (mmix_constant_alignment): Make static. Use the same interface
10065 as the target hook.
10066 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10067 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10068 constant_alignment_word_strings.
10069 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10070 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10071 constant_alignment_word_strings.
10072 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10073 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10074 constant_alignment_word_strings.
10075 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10076 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10077 (rs6000_constant_alignment): New function.
10078 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10079 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10080 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10081 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10082 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10083 (rs6000_constant_alignment): New function.
10084 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10085 * config/s390/s390.c (s390_constant_alignment): New function.
10086 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10087 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10088 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10089 constant_alignment_word_strings.
10090 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10091 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10092 (sparc_constant_alignment): New function.
10093 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10094 * config/spu/spu.c (spu_constant_alignment): New function.
10095 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10096 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10097 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10098 constant_alignment_word_strings.
10099 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10100 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10101 constant_alignment_word_strings.
10102 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10103 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10104 constant_alignment_word_strings.
10105 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10106 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10107 (visium_constant_alignment): New function.
10108 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10109 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10110 (xtensa_constant_alignment): New function.
10111 * system.h (CONSTANT_ALIGNMENT): Poison.
10113 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
10115 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10116 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
10117 (rs6000_builtin_valid_without_lhs): New helper function.
10118 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10119 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
10121 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10123 * target.h (vec_perm_indices): Use unsigned short rather than
10125 (auto_vec_perm_indices): Likewise.
10126 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
10127 Use unsigned int rather than unsigned char.
10128 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
10130 2017-09-25 Richard Biener <rguenther@suse.de>
10132 * cfgloop.h (sort_sibling_loops): Declare.
10133 * cfgloop.c (sort_sibling_loops_cmp): New helper.
10134 (sort_sibling_loops): New function sorting the sibling loop list
10136 * graphite.c (graphite_transform_loops): Sort sibling loops.
10138 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
10140 * target.def (vec_perm_const_ok): Change sel parameter to
10142 * optabs-query.c (can_vec_perm_p): Update accordingly.
10143 * doc/tm.texi: Regenerate.
10144 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
10145 auto_vec_perm_indices and remove separate nelt field.
10146 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
10147 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
10148 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
10149 (aarch64_expand_vec_perm_const): Update accordingly.
10150 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
10151 to vec_perm_indices.
10152 * config/arm/arm.c (expand_vec_perm_d): Change perm to
10153 auto_vec_perm_indices and remove separate nelt field.
10154 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
10155 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
10156 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
10158 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
10159 to vec_perm_indices.
10160 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
10161 sel to vec_perm_indices.
10162 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
10163 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
10164 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
10166 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
10169 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
10172 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
10173 on the FUNCTION_DECL function context if it has a DIE that is a
10176 2017-09-25 Richard Biener <rguenther@suse.de>
10178 PR tree-optimization/82285
10179 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
10182 2017-09-25 Tom de Vries <tom@codesourcery.com>
10186 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
10189 2017-09-25 Richard Biener <rguenther@suse.de>
10191 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
10192 ISL errors other than isl_error_quota happen. Dump if the
10193 schedule is the same.
10194 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
10195 errors instead of aborting inside ISL.
10197 2017-09-25 Iain Sandoe <iain@codesourcery.com>
10200 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
10201 of libgcc_eh for m64.
10202 * config/i386/darwin64.h: Likewise.
10204 2017-09-25 Richard Biener <rguenther@suse.de>
10206 PR middle-end/82144
10207 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
10208 attribute for incomplete types nor twice for complete ones.
10210 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
10213 * config/i386/i386.c (ix86_print_operand_address_as): Encode
10214 %esp as %rsp to avoid 0x67 prefix if there is no index or base
10217 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
10220 * config/i386/i386.opt (mprefer-avx256): Use
10221 ix86_target_flags variable.
10222 * config/i386/i386.c (ix86_target_string): Move
10223 -mprefer-avx256 to flag2_opts.
10225 2017-09-22 Jakub Jelinek <jakub@redhat.com>
10227 PR middle-end/35691
10228 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
10229 and x != -1 | y != -1 into (x & y) != -1.
10231 2017-09-22 Steve Ellcey <sellcey@cavium.com>
10233 * config.gcc: Add new case statement to set
10234 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
10235 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
10236 s390x-*-linux* case statements. Added aarch64 to the list of
10237 supported architectures.
10239 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10241 PR tree-optimization/82289
10242 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
10243 STMT_VINFO_RELEVANT_P.
10245 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10246 Alan Hayward <alan.hayward@arm.com>
10247 David Sherwood <david.sherwood@arm.com>
10249 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
10250 for VR_RANGE only; don't allow VR_ANTI_RANGE.
10251 (extract_range_from_binary_expr_1): Don't call
10252 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
10254 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10255 Alan Hayward <alan.hayward@arm.com>
10256 David Sherwood <david.sherwood@arm.com>
10258 * target.def (preferred_vector_alignment): New hook.
10259 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
10261 * doc/tm.texi: Regenerate.
10262 * targhooks.h (default_preferred_vector_alignment): Declare.
10263 * targhooks.c (default_preferred_vector_alignment): New function.
10264 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
10266 (DR_TARGET_ALIGNMENT): New macro.
10267 (aligned_access_p): Update commentary.
10268 (vect_known_alignment_in_bytes): New function.
10269 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
10271 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
10272 Calculate the misalignment based on the target alignment rather than
10274 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
10275 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
10276 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
10277 the target alignment, rather than masking the element misalignment
10278 with the number of elements in a vector. Also use the target
10279 alignment when calculating the maximum number of peels.
10280 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
10281 instead of TYPE_ALIGN_UNIT.
10282 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
10283 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
10284 (vect_create_addr_base_for_vector_ref): Update call accordingly.
10285 (vect_create_data_ref_ptr): Likewise.
10286 (vect_setup_realignment): Realign by ANDing with
10287 -DR_TARGET_MISALIGNMENT.
10288 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
10289 the number of peels based on DR_TARGET_ALIGNMENT.
10290 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
10291 with the guaranteed alignment boundary when deciding whether
10293 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
10294 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
10295 (ensure_base_align): Remove stmt_info parameter. Get the
10296 target base alignment from DR_TARGET_ALIGNMENT.
10297 (vectorizable_store): Update call accordingly. Interpret
10298 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
10300 (vectorizable_load): Likewise.
10302 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10303 Alan Hayward <alan.hayward@arm.com>
10304 David Sherwood <david.sherwood@arm.com>
10306 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
10307 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
10308 (vect_enhance_data_refs_alignment): Likewise.
10310 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10312 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10313 error. Only quit immediately if parsing is complete.
10314 (BEGIN): Initialize fatal_err and parse_done.
10315 (begin fpu, end fpu): Check number of arguments.
10316 (begin arch, end arch): Likewise.
10317 (begin cpu, end cpu): Likewise.
10318 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10319 (optalias): Likewise.
10321 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10323 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10324 * config/arm/arm-isa.h: Delete. Move definitions to ...
10325 * arm-cpus.in: ... here. Use new feature and fgroup values.
10326 * config/arm/arm.c (arm_option_override): Use lower case for feature
10328 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10329 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10330 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10332 (print_isa_bits_for): New function.
10333 (gen_isa): New function.
10334 (gen_comm_data): Use print_isa_bits_for.
10335 (define feature): New keyword.
10336 (define fgroup): New keyword.
10337 * config/arm/t-arm (TM_H): Remove.
10338 (GTM_H): Add arm-isa.h.
10339 (arm-isa.h): Add rule to generate file.
10340 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10341 case for feature bit names.
10343 2017-09-22 Richard Biener <rguenther@suse.de>
10345 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
10347 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
10348 print of no dependency loops ...
10349 * graphite.c (graphite_transform_loops): ... here.
10350 (canonicalize_loop_closed_ssa_form): Work from inner to outer
10352 (same_close_phi_node, remove_duplicate_close_phi,
10353 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
10354 (canonicalize_loop_closed_ssa): ... here and simplify.
10355 * graphite-optimize-isl.c: Include tree-vectorizer.h.
10356 (optimize_isl): Use dump_printf_loc to tell when we stopped
10357 optimizing because of an ISL timeout.
10359 2017-09-22 Richard Biener <rguenther@suse.de>
10361 PR tree-optimization/82291
10362 * tree-if-conv.c (predicate_mem_writes): Make sure to
10363 remove writes in blocks predicated with false.
10365 2017-09-22 Richard Biener <rguenther@suse.de>
10367 * sese.c: Include cfganal.h.
10368 (if_region_set_false_region): Remove.
10369 (create_if_region_on_edge): Likewise.
10370 (move_sese_in_condition): Re-implement without destroying
10373 2017-09-22 Richard Biener <rguenther@suse.de>
10375 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
10376 Verify both BBs contain loop PHI nodes before dispatching to
10377 copy_loop_phi_args.
10378 (graphite_regenerate_ast_isl): Do not recompute dominators,
10379 do not verify three times. Restructure for clarity.
10380 * graphite-scop-detection.c (same_close_phi_node,
10381 remove_duplicate_close_phi, make_close_phi_nodes_unique,
10382 defined_in_loop_p, canonicalize_loop_closed_ssa,
10383 canonicalize_loop_closed_ssa_form): Simplify, remove excess
10384 checking and SSA rewrite, move to ...
10385 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
10386 (graphite_initialize): Do not pass in ctx, do not reset the
10387 SCEV cache, compute only dominators.
10388 (graphite_transform_loops): Allocate ISL ctx after
10389 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
10390 Maintain post-dominators only around build_scops.
10391 * sese.c (if_region_set_false_region): Make static. Free
10392 and recompute dominators.
10393 (move_sese_in_condition): Assert we don't get called with
10394 post-dominators computed.
10395 * sese.h (if_region_set_false_region): Remove.
10397 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
10399 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
10400 mode attribute for TARGET_AVX512VL.
10402 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
10404 * config/i386/i386.opt (mprefer-avx256): New option.
10405 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
10407 (ix86_preferred_simd_mode): Return 256-bit AVX modes
10408 for TARGET_PREFER_AVX256.
10409 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
10411 2017-09-21 Jeff Law <law@redhat.com>
10413 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
10414 Fix dump output if the only stack space is for pushed registers.
10416 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10418 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
10421 2017-09-21 Martin Sebor <msebor@redhat.com>
10424 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
10425 code (in C++) or code that triggers warnings.
10427 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
10429 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
10431 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
10433 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
10434 * sched-rgn.c: Ditto.
10435 * sel-sched-ir.c: Ditto.
10437 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
10439 * toplev.h (set_random_seed): Adjust return type.
10440 * toplev.c (init_local_tick): Move eager initialization of random_seed
10441 to get_random_seed. Adjust comment.
10442 (init_random_seed): Inline to get_random_seed, delete.
10443 (get_random_seed): Initialize random_seed lazily.
10444 (set_random_seed): Do not return previous value.
10445 (print_switch_value): Do not call get_random_seed.
10447 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
10449 * cgraph.c (delete_function_version): New, broken out from...
10450 (cgraph_node::delete_function_version): ...here. Rename to
10451 cgraph_node::delete_function_version_by_decl. Update all uses.
10452 (cgraph_node::remove): Call delete_function_version.
10454 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10457 * tree-inline.c (expand_call_inline): Emit clobber stmts for
10458 VAR_DECLs to which addressable non-volatile parameters are mapped
10459 and for id->retvar after the return value assignment. Clear
10460 id->retval and id->retbnd after inlining.
10462 2017-09-21 Richard Biener <rguenther@suse.de>
10464 PR tree-optimization/82276
10465 PR tree-optimization/82244
10466 * tree-vrp.c (build_assert_expr_for): Set
10467 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
10469 (remove_range_assertions): Revert earlier change.
10471 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
10474 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
10476 2017-09-21 Richard Biener <rguenther@suse.de>
10478 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10479 Restore valid IL after code generation errors.
10480 * graphite.c (graphite_transform_loops): Diagnose code
10481 generation issues as MSG_MISSED_OPTIMIZATION and continue
10482 with processing SCOPs.
10484 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10485 Alan Hayward <alan.hayward@arm.com>
10486 David Sherwood <david.sherwood@arm.com>
10488 * calls.c (compute_argument_addresses): Use simplify_gen_binary
10489 rather than choosing between plus_constant and gen_rtx_<CODE>.
10490 * expr.c (emit_push_insn): Likewise.
10491 (expand_expr_real_2): Likewise.
10493 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10494 Alan Hayward <alan.hayward@arm.com>
10495 David Sherwood <david.sherwood@arm.com>
10497 * loop-unroll.c (split_iv): Call copy_rtx on the step.
10499 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10500 Alan Hayward <alan.hayward@arm.com>
10501 David Sherwood <david.sherwood@arm.com>
10503 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
10504 calling tree_to_uhwi.
10506 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10507 Alan Hayward <alan.hayward@arm.com>
10508 David Sherwood <david.sherwood@arm.com>
10510 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
10511 INTEGER_CST rather than a negative test for ADDR_EXPR.
10513 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10514 Alan Hayward <alan.hayward@arm.com>
10515 David Sherwood <david.sherwood@arm.com>
10517 * tree-vrp.c (extract_range_from_binary_expr_1): Check
10518 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
10520 2017-09-21 Richard Biener <rguenther@suse.de>
10522 PR tree-optimization/71351
10523 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
10524 graphite_create_new_loop_guard): Remove, fold remaining parts
10526 (translate_isl_ast_node_for): ... here and simplify.
10528 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10531 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
10532 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
10533 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
10534 alternative always use QI mode, for -Os imov (=R,R) alternative
10535 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
10538 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10539 Jeff Law <law@redhat.com>
10541 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
10542 (allocate_stack_space): New function, partially extracted from
10543 s390_emit_prologue.
10544 (s390_emit_prologue): Track offset to most recent stack probe.
10545 Code to allocate space moved into allocate_stack_space.
10546 Dump actions when no stack is allocated.
10547 (s390_prologue_plus_offset): New function.
10548 (s390_emit_stack_probe): Likewise.
10550 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
10552 * common.opt (Wa, Wl, Wp, g, gz=): Add
10554 (gno-column-info): Remove.
10555 (gcolumn-info): Drop RejectNegative.
10556 (gno-): New prefix.
10557 (gno-record-gcc-switches): Remove.
10558 (grecord-gcc-switches): Drop RejectNegative.
10559 (gno-split-dwarf): Remove.
10560 (gsplit-dwarf): Drop RejectNegative.
10561 (gno-strict-dwarf): Remove.
10562 (gstrict-dwarf): Drop RejectNegative.
10563 * config/darwin.opt (gfull, gused): Add RejectNegative.
10564 * dwarf2out.c (gen_producer_string): Drop
10565 gno-record-gcc-switches handler.
10566 * optc-gen.awk: Add g to prefixes with negative forms.
10567 * opts-common.c (remapping_prefix_p): New.
10568 (find_opt): Check it.
10569 (generate_canonical_option): Test g prefix.
10570 (option_map): Add -gno- mapping.
10571 (add_misspelling_candidates): Check remapping_prefix_p.
10573 2017-09-20 Jeff Law <law@redhat.com>
10575 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
10576 thinko in stack clash protection support.
10578 * explow.c (compute_stack_clash_protection_loop_data): Use
10579 CONST_INT_P instead of explicit test. Verify object is a
10580 CONST_INT_P before looking at INTVAL.
10581 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
10582 instead of explicit test.
10584 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
10587 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
10588 address instead of to r1 and r11.
10590 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
10592 * config.gcc: Support "knm".
10593 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
10594 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
10596 * config/i386/i386.c (m_KNM): Define.
10597 (processor_target_table): Add "knm".
10599 (ix86_option_override_internal): Add "knm".
10600 (ix86_issue_rate): Add PROCESSOR_KNM.
10601 (ix86_adjust_cost): Ditto.
10602 (ia32_multipass_dfa_lookahead): Ditto.
10603 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
10604 (fold_builtin_cpu): Add M_INTEL_KNM.
10605 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
10606 (processor_type): Add PROCESSOR_KNM.
10607 * config/i386/x86-tune.def: Add m_KNM.
10608 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
10610 2017-09-20 Richard Biener <rguenther@suse.de>
10612 PR tree-optimization/80213
10613 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
10614 are allowed in empty BBs as well.
10615 (canonicalize_loop_closed_ssa): Also look for other complex
10617 (scop_detection::get_sese): Include the loop-closed PHI block
10619 (scop_detection::merge_sese): Remove code adding extra blocks.
10620 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
10621 (build_scops): Assert the final returned scop is invalid.
10623 2017-09-20 Richard Biener <rguenther@suse.de>
10625 PR tree-optimization/82264
10626 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
10628 (vn_phi_lookup): Likewise.
10629 (vn_phi_insert): Likewise.
10631 2017-09-20 Jakub Jelinek <jakub@redhat.com>
10633 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
10634 that fits into uhwi or shwi, add DW_AT_const_value regardless
10635 of early_dwarf without going through RTL, using add_AT_unsigned
10638 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
10639 (DEBUG_LTO_ABBREV_SECTION): Likewise.
10640 (DEBUG_LTO_MACINFO_SECTION): Likewise.
10641 (DEBUG_MACRO_SECTION): Likewise.
10642 (DEBUG_LTO_MACRO_SECTION): Likewise.
10643 (DEBUG_STR_DWO_SECTION): Likewise.
10644 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
10645 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
10646 (DEBUG_LTO_DWO_LINE_SECTION): Define.
10647 (DEBUG_LTO_LINE_STR_SECTION): Define.
10648 (init_sections_and_labels): Initialize debug_line_str_section
10649 variable. Initialize debug_loc_section for -gdwarf-5 to
10650 DEBUG_LOCLISTS_SECTION. Formatting fixes.
10652 2017-09-20 Richard Biener <rguenther@suse.de>
10654 * graphite-sese-to-poly.c (extract_affine): Properly handle
10655 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
10657 2017-09-20 Richard Biener <rguenther@suse.de>
10659 PR tree-optimization/81373
10660 * graphite-scop-detection.c (build_cross_bb_scalars_def):
10661 Force SESE live-out defs to be handled even if they are
10664 2017-09-19 Jeff Law <law@redhat.com>
10666 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
10667 nothing for stack adjustments with REG_STACK_CHECK.
10668 * sched-deps.c (parse_add_or_inc): Reject insns with
10669 REG_STACK_CHECK from dependency breaking.
10670 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
10671 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
10672 * reg-notes.def (STACK_CHECK): New note.
10674 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
10675 (ix86_expand_prologue): Dump stack clash info as needed.
10676 Call ix86_adjust_stack_and_probe_stack_clash as needed.
10678 * function.c (dump_stack_clash_frame_info): New function.
10679 * function.h (dump_stack_clash_frame_info): Prototype.
10680 (enum stack_clash_probes): New enum.
10682 * config/alpha/alpha.c (alpha_expand_prologue): Also check
10683 flag_stack_clash_protection.
10684 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
10685 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
10686 (arm_frame_pointer_required): Likewise.
10687 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
10688 (ia64_expand_prologue): Likewise.
10689 * config/mips/mips.c (mips_expand_prologue): Likewise.
10690 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
10691 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10692 (sparc_flat_expand_prologue): Likewise.
10693 * config/spu/spu.c (spu_expand_prologue): Likewise.
10695 * explow.c: Include "params.h".
10696 (anti_adjust_stack_and_probe_stack_clash): New function.
10697 (get_stack_check_protect): Likewise.
10698 (compute_stack_clash_protection_loop_data): Likewise.
10699 (emit_stack_clash_protection_loop_start): Likewise.
10700 (emit_stack_clash_protection_loop_end): Likewise.
10701 (allocate_dynamic_stack_space): Use get_stack_check_protect.
10702 Use anti_adjust_stack_and_probe_stack_clash.
10703 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
10704 (emit_stack_clash_protection_loop_start): Likewise.
10705 (emit_stack_clash_protection_loop_end): Likewise.
10706 * rtl.h (get_stack_check_protect): Prototype.
10707 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
10708 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
10709 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
10711 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
10713 * doc/tm.texi: Rebuilt.
10714 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
10715 get_stack_check_protect.
10716 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
10717 * config/arm/arm.c (arm_expand_prologue): Likewise.
10718 (arm_frame_pointer_required): Likewise.
10719 * config/i386/i386.c (ix86_expand_prologue): Likewise.
10720 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
10721 * config/mips/mips.c (mips_expand_prologue): Likewise.
10722 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
10723 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
10724 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10725 (sparc_flat_expand_prologue): Likewise.
10727 * common.opt (-fstack-clash-protection): New option.
10728 * flag-types.h (enum stack_check_type): Note difference between
10729 -fstack-check= and -fstack-clash-protection.
10730 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
10731 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
10732 * toplev.c (process_options): Issue warnings/errors for cases
10733 not handled with -fstack-clash-protection.
10734 * doc/invoke.texi (-fstack-clash-protection): Document new option.
10735 (-fstack-check): Note additional problem with -fstack-check=generic.
10736 Note that -fstack-check is primarily for Ada and refer users
10737 to -fstack-clash-protection for stack-clash-protection.
10738 Document new params for stack clash protection.
10740 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
10742 * config/i386/i386.c (ix86_split_long_move): Do not handle
10743 address used for LEA in a special way.
10745 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
10747 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
10749 2017-09-19 Martin Sebor <msebor@redhat.com>
10752 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
10753 of incompatible types.
10755 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
10757 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10758 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
10759 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10760 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
10762 2017-09-19 Richard Biener <rguenther@suse.de>
10764 PR tree-optimization/82244
10765 * tree-vrp.c (remove_range_assertions): Do not propagate
10766 a constant to abnormals but replace the assert with a copy.
10768 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
10770 PR rtl-optimization/57878
10771 PR rtl-optimization/68988
10772 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
10773 avoidance test involving non_reload_pseudos. Move frequency test
10774 below the general fragmentation avoidance test.
10776 2017-09-19 Richard Biener <rguenther@suse.de>
10778 PR tree-optimization/69728
10779 * graphite-sese-to-poly.c (schedule_error): New global.
10780 (add_loop_schedule): Handle empty domain by failing the
10782 (build_original_schedule): Handle schedule_error.
10784 2017-09-19 Richard Biener <rguenther@suse.de>
10786 * graphite-scop-detection.c (scop_detection::can_represent_loop):
10787 Do not iterate to sibling loops but only to siblings of inner
10790 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
10793 * config/m68k/m68k.md (moveq feeding equality comparison): Check
10794 that the registers are different.
10796 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
10798 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
10799 to processor_model and "amdfam17h" to arch_names_table.
10800 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
10802 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10805 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
10807 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10808 Alan Hayward <alan.hayward@arm.com>
10809 David Sherwood <david.sherwood@arm.com>
10811 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
10813 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
10815 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
10816 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
10817 vect_schedule_slp_instance.
10818 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
10819 Update call to vect_slp_analyze_node_operations. Simplify return
10821 (vect_slp_analyze_bb_1): Update call accordingly.
10822 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
10823 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
10824 (vect_schedule_slp): Update call accordingly.
10826 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10827 Alan Hayward <alan.hayward@arm.com>
10828 David Sherwood <david.sherwood@arm.com>
10830 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
10831 with types that aren't in fact scalar.
10833 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10835 * tree-vect-slp.c (vect_record_max_nunits): New function,
10837 (vect_build_slp_tree_1): ...here.
10838 (vect_build_slp_tree_2): Call it for phis too.
10840 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10842 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
10843 to vect_get_vec_def_for_operand when getting the mask operand.
10845 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10846 Alan Hayward <alan.hayward@arm.com>
10847 David Sherwood <david.sherwood@arm.com>
10849 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
10852 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10853 Alan Hayward <alan.hayward@arm.com>
10854 David Sherwood <david.sherwood@arm.com>
10856 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
10857 calculation for vector booleans.
10859 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10860 Alan Hayward <alan.hayward@arm.com>
10861 David Sherwood <david.sherwood@arm.com>
10863 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
10865 (vect_transform_stmt): ...here.
10866 (vect_analyze_stmt): Use it instead of calling
10867 vectorizable_live_operation directly.
10869 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
10871 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
10872 non-SIMT targets in acc vector loops.
10874 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
10876 * configure.ac: Add arc and check if assembler supports gdwarf2.
10877 * configure: Regenerate.
10879 2017-09-18 Richard Biener <rguenther@suse.de>
10881 PR tree-optimization/82220
10882 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
10883 epilogue niters from the min_profitable_iters compute.
10885 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10888 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
10889 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
10890 (ix86_init_pic_reg): Revert 2017-09-01 changes.
10892 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
10895 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
10896 switching to a new text section.
10898 2017-09-18 Richard Biener <rguenther@suse.de>
10900 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
10902 (build_alias_set): Reject aliases with no access function.
10904 2017-09-18 Richard Biener <rguenther@suse.de>
10906 PR tree-optimization/79622
10907 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
10909 (build_cross_bb_scalars_use): Likewise.
10911 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
10913 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
10915 2017-09-18 Alan Modra <amodra@gmail.com>
10918 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
10919 stack_pointer_rtx for count 0. Update comments. Break up
10920 large rtl expression.
10922 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
10924 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
10925 Increase to 20 bytes.
10926 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
10927 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
10928 or avx version of the stub.
10930 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
10933 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
10934 compute the minimum stack alignment. Also update preferred stack
10935 boundary for leaf functions.
10937 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
10939 PR tree-optimization/82228
10940 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
10943 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10945 * common/config/nds32/nds32-common.c
10946 (nds32_option_optimization_table): Refine formatting.
10947 (nds32_option_optimization_table): Use -fsched-pressure and
10948 -fomit-frame-pointer for specific optimization level.
10950 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10952 * config/nds32/nds32.c: Refine formatting and comments.
10953 * config/nds32/nds32.h: Likewise.
10954 * config/nds32/nds32.md: Likewise.
10955 * config/nds32/nds32-cost.c: Likewise.
10956 * config/nds32/nds32-isr.c: Likewise.
10957 * config/nds32/nds32-md-auxiliary.c: Likewise.
10958 * config/nds32/nds32-multiple.md: Likewise.
10959 * config/nds32/nds32-predicates.c: Likewise.
10961 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
10962 Jakub Jelinek <jakub@redhat.com>
10964 Add support for -std=c++2a.
10965 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
10967 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
10969 2017-09-15 Steve Ellcey <sellcey@cavium.com>
10972 * doc/extend.texi (Common Function Attributes): Add
10973 references to ARM, AArch64, and S/390 specific attributes.
10974 (Function Specific Option Pragmas): Add AArch64 and S/390
10975 to list of back ends that support the target pragma.
10977 2017-09-15 Nathan Sidwell <nathan@acm.org>
10979 * doc/standards.texi: Fix C++17 description. Update URLs for
10982 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
10984 * common.opt (Wcast-align=strict): New warning option.
10985 * doc/invoke.texi: Document -Wcast-align=strict.
10987 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
10989 * cgraph.h (cgraph_thunk_info): Add comments.
10990 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
10991 assert for VIRTUAL_* arguments stricter.
10993 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
10995 PR tree-optimization/71026
10996 * match.pd: Move RDIV patterns from fold-const.c
10997 * fold-const.c (distribute_real_division): Removed.
10998 (fold_binary_loc): Remove calls to distribute_real_divison.
11000 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11002 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
11003 c++1z and gnu++1z as deprecated. Change other references to
11004 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
11005 Change -Wc++1z-compat to -Wc++17-compat.
11006 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
11007 * dwarf2out.c (highest_c_language): Handle C++17.
11008 (gen_compile_unit_die): Likewise.
11010 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11012 PR rtl-optimization/82192
11013 * combine.c (make_extraction): Don't look through non-paradoxical
11014 SUBREGs or TRUNCATE if pos + len is or might be bigger than
11017 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11018 Alan Hayward <alan.hayward@arm.com>
11019 David Sherwood <david.sherwood@arm.com>
11021 * target.def (function_arg_offset): New hook.
11022 * targhooks.h (default_function_arg_offset): Declare.
11023 * targhooks.c (default_function_arg_offset): New function.
11024 * function.c (locate_and_pad_parm): Use
11025 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
11026 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
11027 (TARGET_FUNCTION_ARG_OFFSET): ...this.
11028 * doc/tm.texi: Regenerate.
11029 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
11030 * config/spu/spu.c (spu_function_arg_offset): New function.
11031 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
11032 * system.h (FUNCTION_ARG_OFFSET): Poison.
11034 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11035 Alan Hayard <alan.hayward@arm.com>
11036 David Sherwood <david.sherwood@arm.com>
11038 * target.def (truly_noop_truncation): New hook.
11039 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
11040 than TRULY_NOOP_TRUNCATION.
11041 * hooks.h (hook_bool_uint_uint_true): Declare.
11042 * hooks.c (hook_bool_uint_uint_true): New function.
11043 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
11044 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
11045 * doc/tm.texi: Regenerate.
11046 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
11047 rather than TRULY_NOOP_TRUNCATION in comments.
11048 (simplify_comparison): Likewise.
11049 (record_truncated_value): Likewise.
11050 * expmed.c (extract_bit_field_1): Likewise.
11051 (extract_split_bit_field): Likewise.
11052 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
11053 instead of TRULY_NOOP_TRUNCATION.
11054 * function.c (assign_parm_setup_block): Likewise.
11055 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11056 * rtlhooks.c: Include target.h.
11057 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11058 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11059 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11060 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11061 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11062 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11063 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11064 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11065 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11066 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11067 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11068 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11069 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11070 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11071 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11072 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11073 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11074 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11075 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11076 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11077 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11078 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11079 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11080 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11081 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11082 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11083 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11084 rather than TRULY_NOOP_TRUNCATION in comments.
11085 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11086 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11087 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11088 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11089 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11090 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11091 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11092 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11093 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11094 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11095 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11096 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11097 rather than TRULY_NOOP_TRUNCATION in comments.
11098 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11099 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11100 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11101 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11102 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11103 TRULY_NOOP_TRUNCATION condition.
11104 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11105 (TRULY_NOOP_TRUNCATION): Delete.
11106 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11107 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11108 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11109 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11110 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11111 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11112 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11113 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11114 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11115 rather than TRULY_NOOP_TRUNCATION in comments.
11116 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
11117 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
11118 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
11119 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
11120 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
11121 * system.h (TRULY_NOOP_TRUNCATION): Poison.
11123 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
11126 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
11127 (*cmp_ior): Likewise.
11128 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
11129 (*ior_scc_scc_cmp): Likewise.
11130 (*and_scc_scc): Likewise.
11131 (*and_scc_scc_cmp): Likewise.
11133 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11134 Alan Hayard <alan.hayward@arm.com>
11135 David Sherwood <david.sherwood@arm.com>
11137 * target.def (can_change_mode_class): New hook.
11138 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
11139 (hard_regno_nregs): Likewise.
11140 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
11141 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
11142 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
11143 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
11144 (LOAD_EXTEND_OP): Update accordingly.
11145 * doc/tm.texi: Regenerate.
11146 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
11147 CANNOT_CHANGE_MODE_CLASS.
11148 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
11149 (REG_CAN_CHANGE_MODE_P): ...this new macro.
11150 * combine.c (simplify_set): Update accordingly.
11151 * emit-rtl.c (validate_subreg): Likewise.
11152 * recog.c (general_operand): Likewise.
11153 * regcprop.c (mode_change_ok): Likewise.
11154 * reload1.c (choose_reload_regs): Likewise.
11155 (inherit_piecemeal_p): Likewise.
11156 * rtlanal.c (simplify_subreg_regno): Likewise.
11157 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
11158 instead of CANNOT_CHANGE_MODE_CLASS.
11159 (reload_cse_simplify_operands): Likewise.
11160 * reload.c (push_reload): Use targetm.can_change_mode_class
11161 instead of CANNOT_CHANGE_MODE_CLASS.
11162 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
11163 REG_CANNOT_CHANGE_MODE_P.
11164 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11165 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
11166 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11167 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11168 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11169 (arm_can_change_mode_class): New function.
11170 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11171 than CANNOT_CHANGE_MODE_CLASS in comments.
11172 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11173 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
11174 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
11175 (ix86_can_change_mode_class): ...this new function, inverting the
11176 sense of the return value.
11177 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11178 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11179 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11180 (ia64_can_change_mode_class): New function.
11181 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11182 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
11183 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
11184 (m32c_can_change_mode_class): ...this new function, inverting the
11185 sense of the return value.
11186 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11187 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11188 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
11189 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
11190 (mips_can_change_mode_class): ...this new function, inverting the
11191 sense of the return value.
11192 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11193 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11194 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11195 (msp430_can_change_mode_class): New function.
11196 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11197 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
11198 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11199 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11200 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11201 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
11202 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11203 (pa_cannot_change_mode_class): Replace with...
11204 (pa_can_change_mode_class): ...this new function, inverting the
11205 sense of the return value.
11206 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11207 than CANNOT_CHANGE_MODE_CLASS in comments.
11208 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11209 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
11210 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11211 (pdp11_cannot_change_mode_class): Replace with...
11212 (pdp11_can_change_mode_class): ...this new function, inverting the
11213 sense of the return value.
11214 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11215 * config/powerpcspe/powerpcspe-protos.h
11216 (rs6000_cannot_change_mode_class_ptr): Delete.
11217 * config/powerpcspe/powerpcspe.c
11218 (rs6000_cannot_change_mode_class_ptr): Delete.
11219 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11220 (rs6000_option_override_internal): Assign to
11221 targetm.can_change_mode_class instead of
11222 rs6000_cannot_change_mode_class_ptr.
11223 (rs6000_cannot_change_mode_class): Replace with...
11224 (rs6000_can_change_mode_class): ...this new function, inverting the
11225 sense of the return value.
11226 (rs6000_debug_cannot_change_mode_class): Replace with...
11227 (rs6000_debug_can_change_mode_class): ...this new function.
11228 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11229 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
11230 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11231 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11232 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
11234 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
11235 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11236 (rs6000_option_override_internal): Assign to
11237 targetm.can_change_mode_class instead of
11238 rs6000_cannot_change_mode_class_ptr.
11239 (rs6000_cannot_change_mode_class): Replace with...
11240 (rs6000_can_change_mode_class): ...this new function, inverting the
11241 sense of the return value.
11242 (rs6000_debug_cannot_change_mode_class): Replace with...
11243 (rs6000_debug_can_change_mode_class): ...this new function.
11244 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11245 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
11246 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
11247 (s390_can_change_mode_class): ...this new function, inverting the
11248 sense of the return value.
11249 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11250 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11251 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
11252 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11253 (sh_cannot_change_mode_class): Replace with...
11254 (sh_can_change_mode_class): ...this new function, inverting the
11255 sense of the return value.
11256 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11257 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11258 (sparc_can_change_mode_class): New function.
11259 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11260 * config/spu/spu.c (spu_can_change_mode_class): New function.
11261 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11262 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11263 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11264 (visium_can_change_mode_class): New function.
11265 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
11267 2017-09-15 Richard Biener <rguenther@suse.de>
11269 PR tree-optimization/82217
11270 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
11271 but not undefined case.
11273 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11276 * postreload.c (reload_cse_simplify_operands): Skip
11277 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
11279 2017-09-15 Richard Biener <rguenther@suse.de>
11281 PR tree-optimization/68823
11282 * graphite-scop-detection.c (build_alias_set): If we have a
11283 possible dependence check whether we can handle them by just
11284 looking at the DRs DR_ACCESS_FNs.
11285 (build_scops): If build_alias_set fails, fail the SCOP.
11287 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11289 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
11290 to support float128 built-in functions that require the ISA 3.0
11292 (BU_FLOAT128_3_HW): Likewise.
11293 (SQRTF128): Add support for the IEEE 128-bit square root and fma
11294 built-in functions.
11295 (FMAF128): Likewise.
11297 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
11298 support for built-in functions that need the ISA 3.0 IEEE 128-bit
11299 floating point instructions.
11300 (rs6000_invalid_builtin): Likewise.
11301 (rs6000_builtin_mask_names): Likewise.
11302 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
11303 (RS6000_BTM_FLOAT128_HW): Likewise.
11304 (RS6000_BTM_COMMON): Likewise.
11305 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
11307 * doc/extend.texi (RS/6000 built-in functions): Document the
11308 IEEE 128-bit floating point square root and fused multiply-add
11309 built-in functions.
11311 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
11313 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
11314 reg (r2) isn't in the set of registers defined in the prologue.
11316 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11317 Alan Hayward <alan.hayward@arm.com>
11318 David Sherwood <david.sherwood@arm.com>
11320 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
11321 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
11322 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
11323 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
11324 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
11326 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
11327 max_vectorization_factor.
11328 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
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_min_worthwhile_factor): Delete.
11335 (vect_worthwhile_without_simd_p): Declare.
11336 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
11337 (vectorizable_reduction): Use it.
11338 * tree-vect-stmts.c (vectorizable_shift): Likewise.
11339 (vectorizable_operation): Likewise.
11341 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11342 Alan Hayward <alan.hayward@arm.com>
11343 David Sherwood <david.sherwood@arm.com>
11345 * tree-vectorizer.h (vect_get_num_copies): New function.
11346 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
11347 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11348 (vectorizable_induction): Likewise.
11349 (vectorizable_live_operation): Likewise.
11350 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11351 (vectorizable_bswap): Likewise.
11352 (vectorizable_call): Likewise.
11353 (vectorizable_conversion): Likewise.
11354 (vectorizable_assignment): Likewise.
11355 (vectorizable_shift): Likewise.
11356 (vectorizable_operation): Likewise.
11357 (vectorizable_store): Likewise.
11358 (vectorizable_load): Likewise.
11359 (vectorizable_condition): Likewise.
11360 (vectorizable_comparison): Likewise.
11361 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
11363 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11364 Alan Hayward <alan.hayward@arm.com>
11365 David Sherwood <david.sherwood@arm.com>
11367 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
11368 of vect_init_vector.
11370 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11371 Alan Hayward <alan.hayward@arm.com>
11372 David Sherwood <david.sherwood@arm.com>
11374 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
11375 an inline wrapper that provides a location.
11376 (gimple_build_vector): Likewise.
11377 * gimple-fold.c (gimple_build_vector_from_val): New function.
11378 (gimple_build_vector): Likewise.
11379 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
11380 functions to build the initial value. Always return a gimple value.
11381 (get_initial_defs_for_reduction): Likewise. Only compute
11383 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
11384 vect_init_vector on the results from get_initial_def(s)_for_reduction.
11385 (vectorizable_induction): Use gimple_build_vector rather than
11388 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11389 Alan Hayward <alan.hayward@arm.com>
11390 David Sherwood <david.sherwood@arm.com>
11392 * target.h (vec_perm_indices): New typedef.
11393 (auto_vec_perm_indices): Likewise.
11394 * optabs-query.h: Include target.h
11395 (can_vec_perm_p): Take a vec_perm_indices *.
11396 * optabs-query.c (can_vec_perm_p): Likewise.
11397 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
11398 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11399 * tree-vect-generic.c (lower_vec_perm): Likewise.
11400 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
11401 (vect_grouped_load_supported): Likewise.
11402 (vect_shift_permute_load_chain): Likewise.
11403 (vect_permute_store_chain): Use auto_vec_perm_indices.
11404 (vect_permute_load_chain): Likewise.
11405 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
11406 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
11407 Update uses of can_vec_perm_p.
11408 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
11409 mode with a number of elements. Take a vec_perm_indices *.
11410 (vect_create_epilog_for_reduction): Update accordingly.
11411 Use auto_vec_perm_indices.
11412 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
11413 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
11414 (vect_transform_slp_perm_load): Likewise.
11415 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
11416 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
11417 (vect_gen_perm_mask_checked): Likewise.
11418 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
11419 (vect_gen_perm_mask_checked): Likewise.
11420 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
11421 (vectorizable_store): Likewise.
11422 (vectorizable_load): Likewise.
11423 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
11424 (vectorizable_bswap): Likewise.
11426 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11427 Alan Hayward <alan.hayward@arm.com>
11428 David Sherwood <david.sherwood@arm.com>
11430 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
11431 * tree.c (build_vector): Likewise.
11432 (build_vector_from_ctor): Update accordingly.
11433 (build_vector_from_val): Likewise.
11434 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
11435 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11436 * tree-vect-generic.c (add_rshift): Likewise.
11437 (expand_vector_divmod): Likewise.
11438 (optimize_vector_constructor): Likewise.
11439 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
11440 (vect_transform_slp_perm_load): Likewise.
11441 (vect_schedule_slp_instance): Likewise.
11442 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
11443 (vectorizable_call): Likewise.
11444 (vect_gen_perm_mask_any): Likewise. Add elements in order.
11445 * expmed.c (make_tree): Likewise.
11446 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
11447 a vector passed to build_vector.
11448 (fold_convert_const): Likewise.
11449 (exact_inverse): Likewise.
11450 (fold_ternary_loc): Likewise.
11451 (fold_relational_const): Likewise.
11452 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
11453 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
11454 (const_unop): Likewise. Store the reduction accumulator in a
11455 variable rather than an array.
11456 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
11457 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
11458 the new vector, rather than constructing it after the input arrays.
11459 (native_interpret_vector): Use auto_vec<tree> when building
11460 a vector passed to build_vector. Add elements in order.
11461 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
11462 auto_vec<tree> when building a vector passed to build_vector.
11463 (vect_create_epilog_for_reduction): Likewise.
11464 (vectorizable_induction): Likewise.
11465 (get_initial_def_for_reduction): Likewise. Fix indentation of
11467 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
11469 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
11470 passed to build_vector.
11472 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11473 Alan Hayward <alan.hayward@arm.com>
11474 David Sherwood <david.sherwood@arm.com>
11476 * tree-core.h (tree_base::u): Add an "nelts" field.
11477 (tree_vector): Use VECTOR_CST_NELTS as the length.
11478 * tree.c (tree_size): Likewise.
11479 (make_vector): Initialize VECTOR_CST_NELTS.
11480 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
11481 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
11482 TYPE_VECTOR_SUBPARTS.
11483 * expr.c (const_vector_mask_from_tree): Consistently use "units"
11484 as the number of units, setting it from VECTOR_CST_NELTS.
11485 (const_vector_from_tree): Likewise.
11486 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
11487 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
11488 (fold_negate_expr_1): Likewise.
11489 (fold_convert_const): Likewise.
11490 (const_binop): Likewise. Differentiate the number of output and
11492 (const_unop): Likewise.
11493 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
11494 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
11495 in cases that did the opposite.
11497 2017-09-14 Richard Biener <rguenther@suse.de>
11499 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
11502 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
11504 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
11506 2017-09-14 Jakub Jelinek <jakub@redhat.com>
11509 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
11510 if and where to split a bb, except for splitting before debug insn
11511 sequences followed by non-label real insn. Delete debug insns
11512 in between basic blocks.
11514 * combine.c (make_compound_operation_int): Formatting fixes.
11516 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
11517 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11518 * config/netbsd.h (LINK_EH_SPEC): Likewise.
11519 * config/sol2.h (LINK_EH_SPEC): Likewise.
11520 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11521 * config/s390/linux.h (LINK_SPEC): Likewise.
11522 * config/freebsd.h (LINK_EH_SPEC): Likewise.
11523 * config/openbsd.h (LINK_EH_SPEC): Likewise.
11524 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11525 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
11526 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
11527 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11528 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
11529 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
11531 2017-09-13 Jakub Jelinek <jakub@redhat.com>
11533 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
11535 (ENDFILE_LINUX_SPEC): Likewise.
11536 (LINK_EH_SPEC): Likewise.
11537 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
11538 (LINK_OS_LINUX_SPEC32): Likewise.
11539 (LINK_OS_LINUX_SPEC64): Likewise.
11540 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
11541 (LINK_OS_LINUX_SPEC): Likewise.
11543 2017-09-13 Martin Liska <mliska@suse.cz>
11545 PR middle-end/82154
11546 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
11547 CASE_HIGH is NULL_TREE.
11549 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11550 Alan Hayward <alan.hayward@arm.com>
11551 David Sherwood <david.sherwood@arm.com>
11553 * target.def (secondary_memory_needed): New hook.
11554 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
11555 instead of SECONDARY_MEMORY_NEEDED.
11556 (secondary_memory_needed_mode): Likewise.
11557 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
11558 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
11559 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
11560 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
11561 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
11562 * doc/tm.texi: Regenerate.
11563 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
11564 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
11565 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11566 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
11567 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
11568 * config/i386/i386.c (inline_secondary_memory_needed): Put the
11569 mode argument first and change the reg_class arguments to reg_class_t.
11570 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
11571 Make static. Update the call to inline_secondary_memory_needed.
11572 (ix86_register_move_cost): Update the call to
11573 inline_secondary_memory_needed.
11574 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11575 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
11577 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
11578 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11580 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
11581 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
11582 * config/mips/mips.c (mips_secondary_memory_needed): Make static
11583 and match hook interface. Add comment from mips.h.
11584 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11585 * config/mmix/mmix.md (truncdfsf2): Refer to
11586 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11588 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
11589 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
11590 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11591 (pa_secondary_memory_needed): New function.
11592 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
11593 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
11594 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11595 (pdp11_secondary_memory_needed): Make static and match hook interface.
11596 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
11597 * config/powerpcspe/powerpcspe-protos.h
11598 (rs6000_secondary_memory_needed_ptr): Delete.
11599 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
11601 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11602 (rs6000_option_override_internal): Assign to
11603 targetm.secondary_memory_needed rather than
11604 rs6000_secondary_memory_needed_ptr.
11605 (rs6000_secondary_memory_needed): Match hook interface.
11606 (rs6000_debug_secondary_memory_needed): Likewise.
11607 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
11608 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
11609 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
11610 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11611 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
11612 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
11614 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
11615 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11616 (rs6000_option_override_internal): Assign to
11617 targetm.secondary_memory_needed rather than
11618 rs6000_secondary_memory_needed_ptr.
11619 (rs6000_secondary_memory_needed): Match hook interface.
11620 (rs6000_debug_secondary_memory_needed): Likewise.
11621 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
11622 * config/s390/s390.c (s390_secondary_memory_needed): New function.
11623 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11624 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
11625 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11626 (sparc_secondary_memory_needed): New function.
11627 * lra-constraints.c (check_and_process_move): Refer to
11628 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11630 (curr_insn_transform): Likewise.
11631 (process_alt_operands): Use targetm.secondary_memory_needed
11632 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11633 (check_secondary_memory_needed_p): Likewise.
11634 (choose_split_class): Likewise.
11635 * reload.c: Unconditionally include code that was previously
11636 conditional on SECONDARY_MEMORY_NEEDED.
11637 (push_secondary_reload): Use targetm.secondary_memory_needed
11638 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11639 (push_reload): Likewise.
11640 * reload1.c: Unconditionally include code that was previously
11641 conditional on SECONDARY_MEMORY_NEEDED.
11642 (choose_reload_regs): Use targetm.secondary_memory_needed
11643 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11644 (gen_reload): Likewise.
11645 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
11647 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11648 Alan Hayward <alan.hayward@arm.com>
11649 David Sherwood <david.sherwood@arm.com>
11651 * target.def (secondary_memory_needed_mode): New hook:
11652 * targhooks.c (default_secondary_memory_needed_mode): Declare.
11653 * targhooks.h (default_secondary_memory_needed_mode): New function.
11654 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
11655 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
11656 * doc/tm.texi: Regenerate.
11657 * lra-constraints.c (check_and_process_move): Use
11658 targetm.secondary_memory_needed_mode instead of
11659 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
11660 (curr_insn_transform): Likewise.
11661 * reload.c (get_secondary_mem): Likewise.
11662 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11663 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
11665 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11666 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11667 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
11668 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11669 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
11671 * config/powerpcspe/powerpcspe-protos.h
11672 (rs6000_secondary_memory_needed_mode): Delete.
11673 * config/powerpcspe/powerpcspe.c
11674 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11675 (rs6000_secondary_memory_needed_mode): Make static.
11676 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11677 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
11679 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11681 (rs6000_secondary_memory_needed_mode): Make static.
11682 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11683 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
11684 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11685 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11686 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11688 (sparc_secondary_memory_needed_mode): New function.
11689 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
11691 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
11693 * config/aarch64/constraints.md (Umq): New constraint.
11694 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
11696 (mov<mode>): Update condition.
11698 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11700 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
11701 when bitposition is the same.
11703 2017-09-13 Richard Biener <rguenther@suse.de>
11705 * dwarf2out.c (output_die_symbol): Remove.
11706 (output_die): Do not output a DIEs symbol.
11708 2017-09-13 Richard Biener <rguenther@suse.de>
11710 PR middle-end/82128
11711 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
11712 default-def to avoid breaking iterator update with the weird
11713 interaction with cgraph_update_edges_for_call_stmt_node.
11715 2017-09-13 Richard Biener <rguenther@suse.de>
11717 * tree-cfg.c (verify_gimple_assign_binary): Add verification
11718 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
11719 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
11720 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
11722 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
11724 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
11725 Disable pc relative literal load irrespective of
11726 TARGET_FIX_ERR_A53_84341 for default.
11728 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
11730 * config/sparc/sparc.c (output_return): Output the source location of
11731 the insn in the delay slot, if any.
11732 (output_sibcall): Likewise.
11734 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
11737 * common.opt (-static-pie): New alias.
11738 (shared): Negate static-pie.
11739 (-no-pie): Update help text.
11741 (static-pie): New option.
11742 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
11743 -static-pie support.
11744 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
11745 (LINK_EH_SPEC): Likewise.
11746 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11747 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11748 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11749 * gcc.c (LINK_COMMAND_SPEC): Likewise.
11750 (init_gcc_specs): Likewise.
11751 (init_spec): Likewise.
11752 (display_help): Update help message for -pie.
11753 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
11756 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
11758 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
11759 (movdi_aarch64): Likewise.
11760 (movti_aarch64): Likewise.
11762 2017-09-12 Simon Wright <simon@pushface.org>
11765 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
11766 calculation of the minor version, always output as 0.
11768 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11771 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11772 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
11773 on it early, rather than manual conversion late. For
11774 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
11775 instead of performing manual conversion.
11777 2017-09-12 Carl Love <cel@us.ibm.com>
11779 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
11780 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
11782 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11783 VMULOSW): Add definitions.
11784 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11785 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11786 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11787 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11788 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11790 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11792 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
11794 (movti_aarch64): Likewise.
11795 (movdf_aarch64): Likewise.
11796 (movtf_aarch64): Likewise.
11797 (load_pairdi): Likewise.
11798 (store_pairdi): Likewise.
11799 (load_pairdf): Likewise.
11800 (store_pairdf): Likewise.
11801 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
11802 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
11803 (ldr_got_small_<mode>): Likewise.
11804 (ldr_got_small_28k_<mode>): Likewise.
11805 (ldr_got_tiny): Likewise.
11806 * config/aarch64/iterators.md (ldst_sz): New.
11807 (ldpstp_sz): Likewise.
11808 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
11810 (thunderx_load): Split load_8 to load_16.
11811 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
11813 (thunderx2t99_storepair_basic): Split store_8 to store_16.
11814 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
11815 (xgene1_store_pair): Split store_8 to store_16.
11816 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
11817 (falkor_st_0_st_sd): Split store_8 to store_16.
11819 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11821 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
11822 and store1/2/3/4 to store_4/8/12/16.
11823 * config/aarch64/aarch64.md: Update for rename.
11824 * config/arm/arm.md: Likewise.: Likewise.
11825 * config/arm/arm.c: Likewise.
11826 * config/arm/thumb1.md: Likewise.
11827 * config/arm/thumb2.md: Likewise.
11828 * config/arm/vfp.md: Likewise.
11829 * config/arm/arm-generic.md: Likewise.
11830 * config/arm/arm1020e.md: Likewise.
11831 * config/arm/arm1026ejs.md: Likewise.
11832 * config/arm/arm1136jfs.md: Likewise.
11833 * config/arm/arm926ejs.md: Likewise.
11834 * config/arm/cortex-a15.md: Likewise.
11835 * config/arm/cortex-a17.md: Likewise.
11836 * config/arm/cortex-a5.md: Likewise.
11837 * config/arm/cortex-a53.md: Likewise.
11838 * config/arm/cortex-a57.md: Likewise.
11839 * config/arm/cortex-a7.md: Likewise.
11840 * config/arm/cortex-a8.md: Likewise.
11841 * config/arm/cortex-a9.md: Likewise.
11842 * config/arm/cortex-m4.md: Likewise.
11843 * config/arm/cortex-m7.md: Likewise.
11844 * config/arm/cortex-r4.md: Likewise.
11845 * config/arm/exynos-m1.md: Likewise.
11846 * config/arm/fa526.md: Likewise.
11847 * config/arm/fa606te.md: Likewise.
11848 * config/arm/fa626te.md: Likewise.
11849 * config/arm/fa726te.md: Likewise.
11850 * config/arm/fmp626.md: Likewise.
11851 * config/arm/iwmmxt.md: Likewise.
11852 * config/arm/ldmstm.md: Likewise.
11853 * config/arm/marvell-pj4.md: Likewise.
11854 * config/arm/xgene1.md: Likewise.
11855 * config/aarch64/thunderx.md: Likewise.
11856 * config/aarch64/thunderx2t99.md: Likewise.
11857 * config/aarch64/falkor.md: Likewise.
11859 2017-09-12 Martin Liska <mliska@suse.cz>
11861 * attribs.c (private_lookup_attribute): New function.
11862 * attribs.h (private_lookup_attribute): Declared here.
11863 (lookup_attribute): Called from this place.
11865 2017-09-12 Richard Biener <rguenther@suse.de>
11867 PR tree-optimization/82157
11868 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
11869 stmts with side-effects.
11871 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11872 Alan Hayward <alan.hayward@arm.com>
11873 David Sherwood <david.sherwood@arm.com>
11875 * target.def (hard_regno_nregs): New hook.
11876 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
11877 * targhooks.h (default_hard_regno_nregs): Declare.
11878 * targhooks.c (default_hard_regno_nregs): New function.
11879 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
11880 (TARGET_HARD_REGNO_NREGS): ...this hook.
11881 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
11882 (CLASS_MAX_NREGS): Likewise.
11883 * doc/tm.texi: Regenerate.
11884 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
11885 instead of HARD_REGNO_NREGS.
11886 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
11887 HARD_REGNO_NREGS in the comment.
11888 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
11889 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
11890 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
11891 Return an unsigned int.
11892 (TARGET_HARD_REGNO_NREGS): Redefine.
11893 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
11894 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
11895 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11896 (arc_hard_regno_nregs): New function.
11897 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
11898 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
11899 (arm_hard_regno_nregs): New function.
11900 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
11901 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
11902 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
11903 (TARGET_HARD_REGNO_NREGS): Redefine.
11904 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
11905 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
11906 (HARD_REGNO_NREGS): Delete.
11907 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
11908 (cr16_hard_regno_nregs): New function.
11909 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
11910 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
11911 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
11912 (cris_hard_regno_nregs): New function.
11913 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
11914 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
11915 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11916 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
11917 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
11918 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
11919 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
11920 (frv_hard_regno_nregs): Make static. Take and return an
11922 (frv_class_max_nregs): Remove outdated copy of documentation.
11923 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
11924 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
11925 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
11926 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
11927 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
11928 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
11929 (TARGET_HARD_REGNO_NREGS): Redefine.
11930 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
11931 (CLASS_MAX_NREGS): Update comment.
11932 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
11933 (ia64_hard_regno_nregs): New function.
11934 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
11935 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
11936 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
11937 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
11938 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
11940 (m32c_hard_regno_nregs): Likewise. Make static.
11941 (TARGET_HARD_REGNO_NREGS): Redefine.
11942 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
11943 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
11944 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
11945 (m68k_hard_regno_nregs): New function.
11946 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
11947 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
11948 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
11949 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
11950 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
11951 Take and return an unsigned int.
11952 (TARGET_HARD_REGNO_NREGS): Redefine.
11953 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
11954 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11955 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
11956 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
11957 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
11958 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
11959 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
11960 (msp430_hard_regno_nregs): Make static. Take and return an
11962 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
11963 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
11964 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
11965 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
11966 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
11967 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
11968 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
11969 (TARGET_HARD_REGNO_NREGS): Redefine.
11970 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
11971 (PA_HARD_REGNO_NREGS): ...this.
11972 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
11973 (PA_HARD_REGNO_NREGS): ...this.
11974 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11975 (pa_hard_regno_nregs): New function.
11976 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
11977 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
11978 (pdp11_hard_regno_nregs): New function.
11979 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
11980 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
11981 (rs6000_hard_regno_nregs_hook): New function.
11982 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
11983 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
11984 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
11985 Take and return an unsigned int. Move earlier in file.
11986 (TARGET_HARD_REGNO_NREGS): Redefine.
11987 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
11988 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
11989 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
11990 (rl78_hard_regno_nregs): Make static. Take and return an
11992 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
11993 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
11994 (rs6000_hard_regno_nregs_hook): New function.
11995 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
11996 * config/rx/rx.c (rx_hard_regno_nregs): New function.
11997 (TARGET_HARD_REGNO_NREGS): Redefine.
11998 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
11999 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
12000 instead of HARD_REGNO_NREGS.
12001 (s390_hard_regno_nregs): New function.
12002 (s390_hard_regno_mode_ok): Add comment from s390.h.
12003 (TARGET_HARD_REGNO_NREGS): Redefine.
12004 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
12005 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
12006 (sh_hard_regno_nregs): New function.
12007 (sh_pass_in_reg_p): Use it.
12008 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
12009 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
12010 (sparc_hard_regno_nregs): New function.
12011 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
12012 * config/spu/spu.c (spu_hard_regno_nregs): New function.
12013 (spu_function_arg_advance): Use it, supplying a valid register number.
12014 (TARGET_HARD_REGNO_NREGS): Redefine.
12015 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
12016 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
12017 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
12018 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
12019 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
12020 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
12021 (CLASS_MAX_NREGS): Remove copy of old documentation.
12022 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
12023 (visium_hard_regno_nregs): New function.
12024 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
12025 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
12026 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
12027 xtensa_hard_regno_nregs): New function.
12028 * system.h (HARD_REGNO_NREGS): Poison.
12030 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12032 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
12033 hard_regno_nregs instead of HARD_REGNO_NREGS.
12034 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
12035 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
12036 (c6x_expand_epilogue): Likewise.
12037 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
12038 (frv_read_iacc_argument): Likewise.
12039 * config/sh/sh.c: Include regs.h.
12040 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12041 (regs_used): Likewise.
12042 (output_stack_adjust): Likewise.
12043 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
12044 * expmed.c: Include regs.h.
12045 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12046 * ree.c: Include regs.h.
12047 (combine_reaching_defs): Use hard_regno_nregs instead of
12049 (add_removable_extension): Likewise.
12051 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12053 * regs.h (hard_regno_nregs): Turn into a function.
12054 (end_hard_regno): Update accordingly.
12055 * caller-save.c (setup_save_areas): Likewise.
12056 (save_call_clobbered_regs): Likewise.
12057 (replace_reg_with_saved_mem): Likewise.
12058 (insert_restore): Likewise.
12059 (insert_save): Likewise.
12060 * combine.c (can_change_dest_mode): Likewise.
12061 (move_deaths): Likewise.
12062 (distribute_notes): Likewise.
12063 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12064 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12065 (rs6000_split_multireg_move): Likewise.
12066 (rs6000_register_move_cost): Likewise.
12067 (rs6000_memory_move_cost): Likewise.
12068 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12069 (rs6000_split_multireg_move): Likewise.
12070 (rs6000_register_move_cost): Likewise.
12071 (rs6000_memory_move_cost): Likewise.
12072 * cselib.c (cselib_reset_table): Likewise.
12073 (cselib_lookup_1): Likewise.
12074 * emit-rtl.c (set_mode_and_regno): Likewise.
12075 * function.c (aggregate_value_p): Likewise.
12076 * ira-color.c (setup_profitable_hard_regs): Likewise.
12077 (check_hard_reg_p): Likewise.
12078 (calculate_saved_nregs): Likewise.
12079 (assign_hard_reg): Likewise.
12080 (improve_allocation): Likewise.
12081 (calculate_spill_cost): Likewise.
12082 * ira-emit.c (modify_move_list): Likewise.
12083 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12084 (ira_hard_reg_in_set_p): Likewise.
12085 * ira.c (setup_reg_mode_hard_regset): Likewise.
12086 (clarify_prohibited_class_mode_regs): Likewise.
12087 (check_allocation): Likewise.
12088 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12089 (lra_setup_reg_renumber): Likewise.
12090 (setup_try_hard_regno_pseudos): Likewise.
12091 (spill_for): Likewise.
12092 (assign_hard_regno): Likewise.
12093 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12094 * lra-constraints.c (in_class_p): Likewise.
12095 (lra_constraint_offset): Likewise.
12096 (simplify_operand_subreg): Likewise.
12097 (lra_constraints): Likewise.
12098 (split_reg): Likewise.
12099 (split_if_necessary): Likewise.
12100 (invariant_p): Likewise.
12101 (inherit_in_ebb): Likewise.
12102 * lra-lives.c (process_bb_lives): Likewise.
12103 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12104 (get_hard_regs): Likewise.
12105 (do_remat): Likewise.
12106 * lra-spills.c (assign_spill_hard_regs): Likewise.
12107 * mode-switching.c (create_pre_exit): Likewise.
12108 * postreload.c (reload_combine_recognize_pattern): Likewise.
12109 * recog.c (peep2_find_free_register): Likewise.
12110 * regcprop.c (kill_value_regno): Likewise.
12111 (set_value_regno): Likewise.
12112 (copy_value): Likewise.
12113 (maybe_mode_change): Likewise.
12114 (find_oldest_value_reg): Likewise.
12115 (copyprop_hardreg_forward_1): Likewise.
12116 * regrename.c (check_new_reg_p): Likewise.
12117 (regrename_do_replace): Likewise.
12118 * reload.c (push_reload): Likewise.
12119 (combine_reloads): Likewise.
12120 (find_dummy_reload): Likewise.
12121 (operands_match_p): Likewise.
12122 (find_reloads): Likewise.
12123 (find_equiv_reg): Likewise.
12124 (reload_adjust_reg_for_mode): Likewise.
12125 * reload1.c (count_pseudo): Likewise.
12126 (count_spilled_pseudo): Likewise.
12127 (find_reg): Likewise.
12128 (clear_reload_reg_in_use): Likewise.
12129 (free_for_value_p): Likewise.
12130 (allocate_reload_reg): Likewise.
12131 (choose_reload_regs): Likewise.
12132 (reload_adjust_reg_for_temp): Likewise.
12133 (emit_reload_insns): Likewise.
12134 (delete_output_reload): Likewise.
12135 * rtlanal.c (subreg_get_info): Likewise.
12136 * sched-deps.c (sched_analyze_reg): Likewise.
12137 * sel-sched.c (init_regs_for_mode): Likewise.
12138 (mark_unavailable_hard_regs): Likewise.
12139 (choose_best_reg_1): Likewise.
12140 (verify_target_availability): Likewise.
12141 * valtrack.c (dead_debug_insert_temp): Likewise.
12142 * var-tracking.c (track_loc_p): Likewise.
12143 (emit_note_insn_var_location): Likewise.
12144 * varasm.c (make_decl_rtl): Likewise.
12145 * reginfo.c (choose_hard_reg_mode): Likewise.
12146 (init_reg_modes_target): Refer directly to
12147 this_target_regs->x_hard_regno_nregs.
12149 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12151 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
12152 instead of hard_regno_nregs.
12154 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12156 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
12157 end_hard_regno instead of hard_regno_nregs.
12158 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
12159 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
12160 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
12161 * ira-color.c (improve_allocation): Likewise.
12162 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12163 * lra-lives.c (mark_regno_live): Likewise.
12164 (mark_regno_dead): Likewise.
12165 * lra-remat.c (operand_to_remat): Likewise.
12166 * lra.c (collect_non_operand_hard_regs): Likewise.
12167 * postreload.c (reload_combine_note_store): Likewise.
12168 (move2add_valid_value_p): Likewise.
12169 * reload.c (regno_clobbered_p): Likewise.
12171 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12173 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
12175 * config/v850/v850.c (v850_reorg): Likewise.
12176 * reload.c (refers_to_regno_for_reload_p): Likewise.
12177 (find_equiv_reg): Likewise.
12178 * reload1.c (reload_reg_reaches_end_p): Likewise.
12180 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12182 * caller-save.c (add_used_regs): Use REG_NREGS instead of
12184 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
12185 * config/arm/arm.c (output_move_neon): Likewise.
12186 (arm_attr_length_move_neon): Likewise.
12187 (neon_split_vcombine): Likewise.
12188 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
12189 (c6x_mark_reg_written): Likewise.
12190 (c6x_dwarf_register_span): Likewise.
12191 * config/i386/i386.c (ix86_save_reg): Likewise.
12192 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
12193 (rws_access_reg): Likewise.
12194 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
12195 * mode-switching.c (create_pre_exit): Likewise.
12196 * ree.c (combine_reaching_defs): Likewise.
12197 (add_removable_extension): Likewise.
12198 * regcprop.c (find_oldest_value_reg): Likewise.
12199 (copyprop_hardreg_forward_1): Likewise.
12200 * reload.c (reload_inner_reg_of_subreg): Likewise.
12201 (push_reload): Likewise.
12202 (combine_reloads): Likewise.
12203 (find_dummy_reload): Likewise.
12204 (reload_adjust_reg_for_mode): Likewise.
12205 * reload1.c (find_reload_regs): Likewise.
12206 (forget_old_reloads_1): Likewise.
12207 (reload_reg_free_for_value_p): Likewise.
12208 (reload_adjust_reg_for_temp): Likewise.
12209 (emit_reload_insns): Likewise.
12210 (delete_output_reload): Likewise.
12211 * sel-sched.c (choose_best_reg_1): Likewise.
12212 (choose_best_pseudo_reg): Likewise.
12214 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12215 Alan Hayward <alan.hayward@arm.com>
12216 David Sherwood <david.sherwood@arm.com>
12218 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
12219 * target.def (slow_unaligned_access): New hook.
12220 * targhooks.h (default_slow_unaligned_access): Declare.
12221 * targhooks.c (default_slow_unaligned_access): New function.
12222 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
12223 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
12224 * doc/tm.texi: Regenerate.
12225 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
12226 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
12227 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
12229 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
12230 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
12232 (rs6000_slow_unaligned_access): New function.
12233 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12234 (expand_block_compare): Likewise.
12235 (expand_strn_compare): Likewise.
12236 (rs6000_rtx_costs): Likewise.
12237 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
12238 (riscv_slow_unaligned_access): Likewise.
12239 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
12240 (riscv_slow_unaligned_access_p): ...this and make static.
12241 (riscv_option_override): Update accordingly.
12242 (riscv_slow_unaligned_access): New function.
12243 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12244 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
12245 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12246 (rs6000_slow_unaligned_access): New function.
12247 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12248 (rs6000_rtx_costs): Likewise.
12249 * config/rs6000/rs6000-string.c (expand_block_compare)
12250 (expand_strn_compare): Use targetm.slow_unaligned_access instead
12251 of SLOW_UNALIGNED_ACCESS.
12252 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
12253 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
12254 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
12255 of SLOW_UNALIGNED_ACCESS.
12256 * expmed.c (simple_mem_bitfield_p): Likewise.
12257 * expr.c (alignment_for_piecewise_move): Likewise.
12258 (emit_group_load_1): Likewise.
12259 (emit_group_store): Likewise.
12260 (copy_blkmode_from_reg): Likewise.
12261 (emit_push_insn): Likewise.
12262 (expand_assignment): Likewise.
12263 (store_field): Likewise.
12264 (expand_expr_real_1): Likewise.
12265 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
12266 * lra-constraints.c (simplify_operand_subreg): Likewise.
12267 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12268 * gimple-ssa-store-merging.c: Likewise in block comment at start
12270 * tree-ssa-strlen.c: Include target.h.
12271 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
12272 of SLOW_UNALIGNED_ACCESS.
12273 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
12275 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12277 PR rtl-optimization/82185
12278 * expmed.c (emit_store_flag_int): Only test tem if it has been
12281 2017-09-12 Richard Biener <rguenther@suse.de>
12283 PR middle-end/82149
12284 * match.pd ((FTYPE) N CMP CST): Fix typo.
12286 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
12288 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
12290 (mips_near_type_p): Add 'short_call' attribute as a synonym
12292 * doc/extend.texi (short_call): Document new function attribute.
12294 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12297 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
12298 assertion check that in the condition.
12299 (get_atomic_generic_size): Likewise. Before testing if parameter
12300 has pointer type, if it has array type, call for C++
12301 default_conversion to perform array-to-pointer conversion.
12303 2017-09-12 Richard Biener <rguenther@suse.de>
12305 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
12306 for operations we cannot scalarize.
12308 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
12310 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
12311 vectors heap vectors. Clean up comments.
12312 Make visited_bbs a reference.
12313 (profitable_jump_thread_path): Make GC
12314 vectors heap vectors. Clean up comments.
12316 (convert_and_register_jump_thread_path): Make GC vectors heap
12318 (check_subpath_and_update_thread_path): Same. Clean up comments.
12319 Make visited_bbs a reference.
12320 (handle_phi): Abstract common code to to
12321 register_jump_thread_path_if_profitable.
12322 Rename VAR_BB to DEF_BB.
12324 Make GC vectors heap vectors.
12325 Make visited_bbs a reference.
12326 (handle_assignment): Same.
12327 (register_jump_thread_path_if_profitable): New.
12328 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
12330 Make GC vectors heap vectors. Clean up comments.
12331 Make visited_bbs a reference.
12332 (find_jump_threads_backwards): Make visited_bbs live in the stack.
12333 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
12336 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
12339 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
12340 words of E_DImode object are reachable by xtensa_uimm8x4 access.
12342 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
12344 Revert r251800 and r251799.
12346 2017-09-11 Martin Jambor <mjambor@suse.cz>
12349 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
12350 arguments in advance.
12351 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
12352 use it to find predecessor edges.
12353 (naive_outof_ssa): Collect vector of predecessors.
12355 2017-09-08 Jason Merrill <jason@redhat.com>
12357 PR c++/70029 - ICE with ref-qualifier and -flto
12358 * langhooks.h (struct lang_hooks_for_types): Add
12359 copy_lang_qualifiers.
12360 * attribs.c (build_type_attribute_qual_variant): Use it.
12361 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
12363 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
12364 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
12366 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
12369 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
12370 (*mulsi3_sp64): New instruction.
12371 (mulsi3): New expander.
12373 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
12375 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
12377 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12379 * sancov.c: Include memmodel.h.
12381 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
12384 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
12387 2017-09-07 Carl Love <cel@us.ibm.com>
12389 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
12390 the sldi instruction.
12392 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
12394 * sancov.c: Include tm_p.h.
12396 2017-09-07 Jakub Jelinek <jakub@redhat.com>
12399 * output.h (switch_to_other_text_partition): New declaration.
12400 * varasm.c (switch_to_other_text_partition): New function.
12401 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
12402 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
12403 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
12404 to the other text partition before emitting LCL label and switch back
12405 after emitting the word after it.
12407 2017-09-07 Richard Biener <rguenther@suse.de>
12409 * passes.def (pass_split_crit_edges): Remove instance before PRE.
12410 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
12411 critical edges here, after loop init.
12412 (pass_data_pre): Remove PROP_no_crit_edges flags.
12413 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
12414 for valueization of call args to avoid leaking VN_TOP.
12415 (visit_use): Assert we do not visit default defs.
12416 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
12417 Use error_mark_node to more easily detect leaking VN_TOP.
12418 All default-defs are varying, not VN_TOP. Mark them visited.
12419 (run_scc_vn): Make code match comment.
12421 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12423 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
12424 OPTION_MASK_FLOAT128_KEYWORD.
12425 (POWERPC_MASKS): Likewise.
12426 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
12427 support for the -mfloat128-type option, and make -mfloat128
12428 default on PowerPC Linux systems. Define or undefine
12429 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
12430 Define __float128 to be __ieee128 if IEEE 128-bit support is
12431 enabled, or undefine it.
12432 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
12433 Delete defining __FLOAT128_TYPE__.
12434 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
12435 -mfloat128-type option and make -mfloat128 default on PowerPC
12437 (TARGET_FLOAT128_TYPE): Likewise.
12438 (-mfloat128-type): Likewise.
12439 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12440 Delete the -mfloat128-type option and make -mfloat128 default on
12441 PowerPC Linux systems. Always use __ieee128 to be the keyword for
12442 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
12443 128-bit floating point is enabled. Change tests from using
12444 -mfloat128-type to -mfloat128.
12445 (rs6000_mangle_type): Use the correct mangling for the __float128
12446 type even if normal long double is restricted to 64-bits.
12447 (floatn_mode): Enable the _Float128 type by default on VSX Linux
12449 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
12450 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
12451 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
12453 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
12454 documentation for -mfloat128.
12456 2017-09-06 Olivier Hainque <hainque@adacore.com>
12458 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
12460 2017-09-06 Wish Wu <wishwu007@gmail.com>
12461 Jakub Jelinek <jakub@redhat.com>
12463 * asan.c (initialize_sanitizer_builtins): Add
12464 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
12465 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
12466 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
12467 BT_FN_VOID_UINT64_PTR variables.
12468 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
12469 (BT_FN_VOID_UINT16_UINT16): Likewise.
12470 (BT_FN_VOID_UINT32_UINT32): Likewise.
12471 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
12472 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
12473 (BT_FN_VOID_UINT64_PTR): Likewise.
12474 * common.opt (flag_sanitize_coverage): New variable.
12475 (fsanitize-coverage=trace-pc): Remove.
12476 (fsanitize-coverage=): Add.
12477 * flag-types.h (enum sanitize_coverage_code): New enum.
12478 * fold-const.c (fold_range_test): Disable non-short-circuit
12479 optimization if flag_sanitize_coverage.
12480 (fold_truth_andor): Likewise.
12481 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
12482 * opts.c (COVERAGE_SANITIZER_OPT): Define.
12483 (coverage_sanitizer_opts): New array.
12484 (get_closest_sanitizer_option): Add OPTS argument, handle also
12485 OPT_fsanitize_coverage_.
12486 (parse_sanitizer_options): Adjusted to also handle
12487 OPT_fsanitize_coverage_.
12488 (common_handle_option): Add OPT_fsanitize_coverage_.
12489 * sancov.c (instrument_comparison, instrument_switch): New function.
12490 (sancov_pass): Add trace-cmp support.
12491 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
12492 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
12493 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
12494 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
12495 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
12496 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
12497 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
12498 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
12499 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
12500 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
12502 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12504 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12505 error. Only quit immediately if parsing is complete.
12506 (BEGIN): Initialize fatal_err and parse_done.
12507 (begin fpu, end fpu): Check number of arguments.
12508 (begin arch, end arch): Likewise.
12509 (begin cpu, end cpu): Likewise.
12510 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12511 (optalias): Likewise.
12513 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12515 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12516 * config/arm/arm-isa.h: Delete. Move definitions to ...
12517 * arm-cpus.in: ... here. Use new feature and fgroup values.
12518 * config/arm/arm.c (arm_option_override): Use lower case for feature
12520 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12521 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12522 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12524 (print_isa_bits_for): New function.
12525 (gen_isa): New function.
12526 (gen_comm_data): Use print_isa_bits_for.
12527 (define feature): New keyword.
12528 (define fgroup): New keyword.
12529 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
12530 (arm-isa.h): Add rule to generate file.
12531 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12532 case for feature bit names.
12534 2017-09-06 Richard Biener <rguenther@suse.de>
12536 * tree-ssa-pre.c (NECESSARY): Remove.
12537 (create_expression_by_pieces): Do not touch pass-local flags.
12538 (insert_into_preds_of_block): Likewise.
12539 (do_pre_regular_insertion): Likewise.
12540 (eliminate_insert): Likewise.
12541 (eliminate_dom_walker::before_dom_children): Likewise.
12542 (fini_eliminate): Do not look at inserted_exprs.
12543 (mark_operand_necessary): Remove.
12544 (remove_dead_inserted_code): Replace with simple work-list
12545 algorithm based on inserted_exprs and SSA uses.
12546 (pass_pre::execute): Re-order fini_eliminate and
12547 remove_dead_inserted_code.
12549 2017-09-06 Olivier Hainque <hainque@adacore.com>
12551 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
12552 for VxWorks 7. Adjust surrounding comments.
12554 2017-09-06 Richard Biener <rguenther@suse.de>
12556 * gimple-ssa-strength-reduction.c
12557 (find_candidates_dom_walker::before_dom_children): Also allow
12560 2017-09-06 Richard Biener <rguenther@suse.de>
12562 PR tree-optimization/82108
12563 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
12564 for gap in the non-permutation SLP case.
12566 2017-09-06 Martin Jambor <mjambor@suse.cz>
12568 PR tree-optimization/82078
12569 * tree-sra.c (sort_and_splice_var_accesses): Move call to
12570 add_access_to_work_queue...
12571 (build_accesses_from_assign): ...here.
12572 (propagate_all_subaccesses): Make sure racc is the group
12573 representative, if there is one.
12575 2017-09-06 Jakub Jelinek <jakub@redhat.com>
12577 PR middle-end/82095
12578 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
12581 2017-09-06 Richard Biener <rguenther@suse.de>
12583 * gimple-ssa-strength-reduction.c
12584 (find_candidates_dom_walker::before_doom_children): Use a
12585 type and not a mode check.
12587 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
12590 * config/arm/predicates.md (arm_general_adddi_operand): Create new
12592 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
12594 2017-09-05 Jeff Law <law@redhat.com>
12596 PR tree-optimization/64910
12597 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
12598 cases where we have 3 or more operands.
12600 2017-09-05 Jakub Jelinek <jakub@redhat.com>
12602 PR middle-end/81768
12603 * omp-low.c (lower_omp_for): Recompute tree invariant if
12604 gimple_omp_for_initial/final is ADDR_EXPR.
12606 PR middle-end/81768
12607 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
12608 into gimple val before gimplification fo the COND_EXPR.
12610 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
12612 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
12613 REGION_COPY argument.
12614 (thread_through_all_blocks): Remove unused argument to
12615 duplicate_thread_path.
12617 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12618 Alan Hayward <alan.hayward@arm.com>
12619 David Sherwood <david.sherwood@arm.com>
12621 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
12622 Take a scalar_mode rather than a machine_mode.
12623 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12624 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
12625 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12626 (aarch64_gen_adjusted_ldpstp): Likewise.
12627 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
12629 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12630 Alan Hayward <alan.hayward@arm.com>
12631 David Sherwood <david.sherwood@arm.com>
12633 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
12634 Take a scalar_int_mode instead of a machine_mode.
12635 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
12636 (aarch64_output_scalar_simd_mov_immediate): Likewise.
12637 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
12638 (aarch64_simd_attr_length_rglist): Delete.
12639 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
12640 a scalar_int_mode instead of a machine_mode.
12641 (aarch64_add_offset): Likewise.
12642 (aarch64_internal_mov_immediate): Likewise
12643 (aarch64_add_constant_internal): Likewise.
12644 (aarch64_add_constant): Likewise.
12645 (aarch64_movw_imm): Likewise.
12646 (aarch64_rtx_arith_op_extract_p): Likewise.
12647 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
12648 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
12649 Remove assert that the mode isn't a vector.
12650 (aarch64_output_scalar_simd_mov_immediate): Likewise.
12651 (aarch64_expand_mov_immediate): Update calls after above changes.
12652 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
12653 (aarch64_and_bitmask_imm): Check for scalar integer modes.
12654 (aarch64_move_imm): Likewise.
12655 (aarch64_can_const_movi_rtx_p): Likewise.
12656 (aarch64_strip_extend): Likewise.
12657 (aarch64_extr_rtx_p): Likewise.
12658 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
12659 a CONST_INT when the mode parameter is VOIDmode.
12660 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
12662 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12664 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
12665 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12666 (bitwise_type_for_mode): Update accordingly.
12668 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12670 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
12671 * stor-layout.c (mode_for_size_tree): Likewise.
12672 (mode_for_array): Update accordingly.
12673 (layout_decl): Likewise.
12674 (compute_record_mode): Likewise. Only set the mode once.
12676 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12678 * target.def (get_mask_mode): Change return type to opt_mode.
12680 * doc/tm.texi: Regenerate.
12681 * targhooks.h (default_get_mask_mode): Return an opt_mode.
12682 * targhooks.c (default_get_mask_mode): Likewise.
12683 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12684 * optabs-query.c (can_vec_mask_load_store_p): Update use of
12685 targetm.get_mask_mode.
12686 * tree.c (build_truth_vector_type): Likewise.
12688 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12690 * machmode.h (mode_for_vector): Return an opt_mode.
12691 * stor-layout.c (mode_for_vector): Likewise.
12692 (mode_for_int_vector): Update accordingly.
12693 (layout_type): Likewise.
12694 * config/i386/i386.c (emit_memmov): Likewise.
12695 (ix86_expand_set_or_movmem): Likewise.
12696 (ix86_expand_vector_init): Likewise.
12697 (ix86_get_mask_mode): Likewise.
12698 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
12700 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
12701 * expmed.c (extract_bit_field_1): Likewise.
12702 * expr.c (expand_expr_real_2): Likewise.
12703 * optabs-query.c (can_vec_perm_p): Likewise.
12704 (can_vec_mask_load_store_p): Likewise.
12705 * optabs.c (expand_vec_perm): Likewise.
12706 * targhooks.c (default_get_mask_mode): Likewise.
12707 * tree-vect-stmts.c (vectorizable_store): Likewise.
12708 (vectorizable_load): Likewise.
12709 (get_vectype_for_scalar_type_and_size): Likewise.
12711 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12713 * machmode.h (mode_for_int_vector): New function.
12714 * stor-layout.c (mode_for_int_vector): Likewise.
12715 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
12716 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
12717 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12718 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12719 (s390_expand_vcond): Likewise.
12721 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12723 * machmode.h (opt_machine_mode): New type.
12724 (opt_mode<T>): Allow construction from anything that can be
12726 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
12727 (mode_for_size): Return an opt_machine_mode.
12728 * stor-layout.c (mode_for_size): Likewise.
12729 (mode_for_size_tree): Update call accordingly.
12730 (bitwise_mode_for_mode): Likewise.
12731 (make_fract_type): Likewise.
12732 (make_accum_type): Likewise.
12733 * caller-save.c (replace_reg_with_saved_mem): Update call
12735 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12736 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12737 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12738 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12739 * expmed.c (extract_bit_field_1): Likewise.
12740 * reload.c (get_secondary_mem): Likewise.
12741 * varasm.c (assemble_integer): Likewise.
12742 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
12745 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12747 * machmode.h (decimal_float_mode_for_size): New function.
12748 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
12749 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
12750 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
12751 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
12752 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
12753 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
12755 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12757 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
12758 (get_builtin_sync_mode): Likewise.
12759 (expand_ifn_atomic_compare_exchange): Likewise.
12760 (expand_builtin_atomic_clear): Likewise.
12761 (expand_builtin_atomic_test_and_set): Likewise.
12762 (fold_builtin_atomic_always_lock_free): Likewise.
12763 * calls.c (compute_argument_addresses): Likewise.
12764 (emit_library_call_value_1): Likewise.
12765 (store_one_arg): Likewise.
12766 * combine.c (combine_instructions): Likewise.
12767 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
12768 * config/arm/arm.c (arm_function_value): Likewise.
12769 (aapcs_allocate_return_reg): Likewise.
12770 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
12771 * config/i386/i386.c (construct_container): Likewise.
12772 (ix86_gimplify_va_arg): Likewise.
12773 (ix86_expand_sse_cmp): Likewise.
12774 (emit_memmov): Likewise.
12775 (emit_memset): Likewise.
12776 (expand_small_movmem_or_setmem): Likewise.
12777 (ix86_expand_pextr): Likewise.
12778 (ix86_expand_pinsr): Likewise.
12779 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
12780 * config/microblaze/microblaze.c (microblaze_block_move_straight):
12782 * config/mips/mips.c (mips_function_value_1) Likewise.
12783 (mips_block_move_straight): Likewise.
12784 (mips_expand_ins_as_unaligned_store): Likewise.
12785 * config/powerpcspe/powerpcspe.c
12786 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12787 (rs6000_darwin64_record_arg_flush): Likewise.
12788 * config/rs6000/rs6000.c
12789 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12790 (rs6000_darwin64_record_arg_flush): Likewise.
12791 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
12792 (sparc_function_value_1): Likewise.
12793 * config/spu/spu.c (adjust_operand): Likewise.
12794 (spu_emit_branch_or_set): Likewise.
12795 (arith_immediate_p): Likewise.
12796 * emit-rtl.c (gen_lowpart_common): Likewise.
12797 * expr.c (expand_expr_real_1): Likewise.
12798 * function.c (assign_parm_setup_block): Likewise.
12799 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
12800 * reload1.c (alter_reg): Likewise.
12801 * stor-layout.c (mode_for_vector): Likewise.
12802 (layout_type): Likewise.
12804 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12806 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
12807 (spu_convert_move): Likewise.
12808 * lower-subreg.c (resolve_simple_move): Likewise.
12810 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12813 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
12814 define_insn to a define_expand.
12815 (altivec_vsum2sws_direct): New define_insn.
12816 (altivec_vsumsws): Convert from a define_insn to a define_expand.
12818 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12820 * config/arm/arm.c (arm_option_params_internal): Improve setting of
12823 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
12828 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
12829 realign stack if stack alignment needed is less than incoming
12832 2017-09-05 Marek Polacek <polacek@redhat.com>
12835 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
12838 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12840 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
12842 2017-09-05 Richard Biener <rguenther@suse.de>
12844 PR tree-optimization/82084
12845 * fold-const.c (can_native_encode_string_p): Handle wide characters.
12847 2017-09-05 Richard Biener <rguenther@suse.de>
12849 PR tree-optimization/82102
12850 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
12852 2017-09-05 Martin Liska <mliska@suse.cz>
12854 PR tree-optimization/82032
12855 * tree-cfg.c (generate_range_test): New function.
12856 * tree-cfg.h (generate_range_test): Declared here.
12857 * tree-cfgcleanup.c (convert_single_case_switch): New function.
12858 (cleanup_control_expr_graph): Use it.
12859 * tree-switch-conversion.c (try_switch_expansion): Remove
12861 (emit_case_nodes): Use generate_range_test.
12863 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12866 * config/i386/i386.md (*<btsc><mode>_mask): Add
12867 TARGET_USE_BT to insn constraint.
12868 (*btr<mode>_mask): Ditto.
12870 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
12872 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
12873 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
12875 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
12878 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
12879 TARGET_NEON and TARGET_IWMMXT.
12880 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
12881 TARGET_NEON and TARGET_IWMMXT.
12882 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
12884 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12886 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
12887 (ix86_rewrite_tls_address): Ditto.
12888 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
12889 (ix86_rewrite_tls_address_1): Ditto.
12890 (ix86_rewrite_tls_address): Ditto.
12891 * config/i386/predicates.md (tls_address_pattern): New predicate.
12892 * config/i386/i386.md (TLS address splitter): New splitter.
12894 2017-09-04 Richard Biener <rguenther@suse.de>
12896 PR tree-optimization/82084
12897 * fold-const.h (can_native_encode_string_p): Declare.
12898 * fold-const.c (can_native_encode_string_p): Factor out from ...
12899 (native_encode_string): ... here.
12900 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
12901 vectorizing stores from constants we later cannot handle.
12903 2017-09-04 Marek Polacek <polacek@redhat.com>
12906 * doc/invoke.texi: Update -Wtautological-compare documentation.
12908 2017-09-04 Jeff Law <law@redhat.com>
12910 PR tree-optimization/64910
12911 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
12912 swap the first and last operand if the last is a constant.
12914 2017-09-04 Marek Polacek <polacek@redhat.com>
12917 * convert.c (do_narrow): When sanitizing signed integer overflows,
12918 bail out for signed types.
12919 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
12921 2017-09-04 Richard Biener <rguenther@suse.de>
12923 PR tree-optimization/82060
12924 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12925 Move devirtualization after stmt folding and before EH/AB/noreturn
12926 cleanup to get the stmt refs canonicalized. Use a bool instead
12927 of gimple_modified_p since that doesn't work for NOPs. Schedule
12928 NOPs generated by folding for removal.
12930 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12931 Alan Hayward <alan.hayward@arm.com>
12932 David Sherwood <david.sherwood@arm.com>
12934 * coretypes.h (pad_direction): New enum.
12935 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
12936 (FUNCTION_ARG_PADDING): Likewise.
12937 * target.def (function_arg_padding): New hook.
12938 * targhooks.h (default_function_arg_padding): Declare.
12939 * targhooks.c (default_function_arg_padding): New function.
12940 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
12941 (TARGET_FUNCTION_ARG_PADDING): ...this.
12942 * doc/tm.texi: Regenerate.
12943 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
12944 instead of direction.
12945 (compute_argument_addresses): Likewise.
12946 (load_register_parameters): Likewise.
12947 (emit_library_call_value_1): Likewise.
12948 (store_one_arg): Use targetm.calls.function_arg_padding instead
12949 of FUNCTION_ARG_PADDING.
12950 (must_pass_in_stack_var_size_or_pad): Likewise.
12951 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
12952 (emit_group_store): Likewise.
12953 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
12954 instead of FUNCTION_ARG_PADDING.
12955 (emit_push_insn): Likewise, and propagate enum change throughout
12957 * function.h (direction): Delete.
12958 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
12960 * function.c (assign_parm_find_stack_rtl): Likewise.
12961 (assign_parm_setup_block_p): Likewise.
12962 (assign_parm_setup_block): Likewise.
12963 (gimplify_parameters): Likewise.
12964 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
12965 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
12967 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
12968 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12969 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
12970 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
12971 (aarch64_function_arg_padding): ...this new function.
12972 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
12973 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12974 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
12975 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12976 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
12977 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12978 (arm_pad_arg_upward): Replace with...
12979 (arm_function_arg_padding): ...this new function.
12980 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
12982 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
12983 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
12984 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12985 (ia64_hpux_function_arg_padding): Replace with...
12986 (ia64_function_arg_padding): ...this new function. Use pad_direction
12987 instead of direction. Check for TARGET_HPUX.
12988 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
12989 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12990 (iq2000_function_arg_padding): New function.
12991 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
12992 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
12993 (mips_function_arg_padding): ...this new function.
12994 (mips_pad_reg_upward): Update accordingly.
12995 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12996 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
12997 targetm.calls.function_arg_padding.
12998 (FUNCTION_ARG_PADDING): Delete.
12999 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
13000 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
13001 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13002 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
13003 (nios2_block_reg_padding): Return pad_direction instead of direction.
13004 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
13005 instead of direction.
13006 (nios2_function_arg_padding): Likewise. Make static.
13007 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13008 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
13009 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
13010 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
13011 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13012 (pa_function_arg_padding): Make static. Return pad_direction instead
13014 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
13015 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13016 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
13017 instead of direction. Use targetm.calls.function_arg_padding.
13018 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
13019 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13020 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
13021 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
13022 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
13024 (function_arg_padding): Rename to...
13025 (rs6000_function_arg_padding): ...this. Make static. Return
13026 pad_direction instead of direction.
13027 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13028 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
13029 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
13030 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
13031 instead of direction. Use targetm.calls.function_arg_padding.
13032 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
13033 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13034 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
13035 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
13036 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13037 (function_arg_padding): Rename to...
13038 (rs6000_function_arg_padding): ...this. Make static. Return
13039 pad_direction instead of direction.
13040 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13041 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
13042 * config/s390/s390.c (s390_function_arg_padding): New function.
13043 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13044 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
13045 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
13046 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13047 (function_arg_padding): Rename to...
13048 (sparc_function_arg_padding): ...this. Make static. Return
13049 pad_direction instead of direction.
13050 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
13051 * config/spu/spu.c (spu_function_arg_padding): New function.
13052 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13053 * system.h (FUNCTION_ARG_PADDING): Poison.
13055 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13056 Alan Hayward <alan.hayward@arm.com>
13057 David Sherwood <david.sherwood@arm.com>
13059 * target.def (modes_tieable_p): New hook.
13060 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13061 (TARGET_MODES_TIEABLE_P): ...this.
13062 * doc/tm.texi.in: Regenerate.
13063 * hooks.h (hook_bool_mode_mode_true): Declare.
13064 * hooks.c (hook_bool_mode_mode_true): New function.
13065 * combine.c (subst): Use targetm.modes_tieable_p instead of
13067 * dse.c (find_shift_sequence): Likewise.
13068 * expmed.c (extract_low_bits): Likewise.
13069 * lower-subreg.c: Include target.h.
13070 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13072 * rtlanal.c (rtx_cost): Likewise.
13073 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13074 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13075 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13076 (TARGET_MODES_TIEABLE_P): Redefine.
13077 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13078 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13079 (TARGET_MODES_TIEABLE_P): Redefine.
13080 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13081 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13082 (arc_modes_tieable_p): New function.
13083 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13084 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13085 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13086 (arm_modes_tieable_p): Make static.
13087 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13088 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13089 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13090 (TARGET_MODES_TIEABLE_P): Redefine.
13091 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13092 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13093 (TARGET_MODES_TIEABLE_P): Redefine.
13094 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13095 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13096 (cr16_modes_tieable_p): New function.
13097 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13098 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13099 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13100 (TRULY_NOOP_TRUNCATION): Update comment.
13101 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13102 (TRULY_NOOP_TRUNCATION): Update comment.
13103 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13104 (frv_modes_tieable_p): New function.
13105 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13106 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13107 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13108 (TARGET_MODES_TIEABLE_P): Redefine.
13109 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13110 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13111 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13112 (TARGET_MODES_TIEABLE_P): Redefine.
13113 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
13114 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
13115 (ia64_modes_tieable_p): New function.
13116 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
13117 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
13118 (iq2000_modes_tieable_p): New function.
13119 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
13120 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
13121 (lm32_modes_tieable_p): New function.
13122 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
13123 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
13124 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
13125 (TARGET_MODES_TIEABLE_P): Redefine.
13126 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
13127 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
13128 (m32r_modes_tieable_p): New function.
13129 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
13130 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
13131 (m68k_modes_tieable_p): New function.
13132 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
13133 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
13134 (mcore_modes_tieable_p): New function.
13135 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
13136 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
13138 (TARGET_MODES_TIEABLE_P): Redefine.
13139 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
13140 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
13141 * config/mips/mips.c (mips_modes_tieable_p): Make static.
13142 (TARGET_MODES_TIEABLE_P): Redefine.
13143 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
13144 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
13145 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
13146 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
13147 (mn10300_modes_tieable_p): ...this and make static.
13148 (TARGET_MODES_TIEABLE_P): Redefine.
13149 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
13150 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
13151 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
13152 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
13153 (msp430_modes_tieable_p): Make static.
13154 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
13155 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
13156 (TARGET_MODES_TIEABLE_P): Redefine.
13157 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
13158 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
13159 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
13160 (TARGET_MODES_TIEABLE_P): Redefine.
13161 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
13162 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
13163 * config/pa/pa.c (pa_modes_tieable_p): Make static.
13164 (TARGET_MODES_TIEABLE_P): Redefine.
13165 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
13166 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
13167 (pdp11_modes_tieable_p): New function.
13168 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
13169 * config/powerpcspe/powerpcspe.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/powerpcspe/powerpcspe.md: Update comment.
13173 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
13174 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
13175 (TARGET_MODES_TIEABLE_P): Redefine.
13176 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
13177 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
13178 (rl78_modes_tieable_p): New function.
13179 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
13180 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
13181 (rs6000_modes_tieable_p): New function.
13182 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13183 * config/rs6000/rs6000.md: Update comment.
13184 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
13185 * config/rx/rx.c (rx_modes_tieable_p): New function.
13186 (TARGET_MODES_TIEABLE_P): Redefine.
13187 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
13188 * config/s390/s390.c (s390_modes_tieable_p): New function.
13189 (TARGET_MODES_TIEABLE_P): Redefine.
13190 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
13191 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
13192 (sh_modes_tieable_p): New function.
13193 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
13194 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
13195 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
13196 (sparc_modes_tieable_p): Make static.
13197 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
13198 * config/spu/spu.c (spu_modes_tieable_p): New function.
13199 (TARGET_MODES_TIEABLE_P): Redefine.
13200 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
13201 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
13202 (TARGET_MODES_TIEABLE_P): Redefine.
13203 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
13204 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
13205 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
13206 * config/v850/v850.c (v850_modes_tieable_p): New function.
13207 (TARGET_MODES_TIEABLE_P): Redefine.
13208 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
13209 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
13210 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
13211 (visium_modes_tieable_p): New function.
13212 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
13213 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
13214 (xtensa_modes_tieable_p): New function.
13215 * system.h (MODES_TIEABLE_P): Poison.
13217 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13218 Alan Hayward <alan.hayward@arm.com>
13219 David Sherwood <david.sherwood@arm.com>
13221 * target.def (hard_regno_mode_ok): New hook.
13222 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
13223 (TARGET_HARD_REGNO_MODE_OK): ...this.
13224 * doc/tm.texi.in: Regenerate.
13225 * hooks.h (hook_bool_uint_mode_true): Declare.
13226 * hooks.c (hook_bool_uint_mode_true): New function.
13227 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
13228 HARD_REGNO_MODE_OK.
13229 * genpreds.c (write_insn_preds_c): Add an include of target.h.
13230 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
13231 instead of HARD_REGNO_MODE_OK.
13232 * caller-save.c: Include target.h.
13233 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
13234 HARD_REGNO_MODE_OK.
13235 * combine.c (can_combine_p): Likewise.
13236 (combinable_i3pat): Likewise.
13237 (can_change_dest_mode): Likewise.
13238 * expr.c (init_expr_target): Likewise.
13239 (convert_move): Likewise.
13240 (convert_modes): Likewise.
13241 * ira.c (setup_prohibited_class_mode_regs): Likewise.
13242 (setup_prohibited_mode_move_regs): Likewise.
13243 * ira.h (target_ira): Likewise.
13244 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13245 * lra-constraints.c (process_alt_operands): Likewise.
13246 (split_reg): Likewise.
13247 * recog.c (peep2_find_free_register): Likewise.
13248 * ree.c (combine_reaching_defs): Likewise.
13249 * regcprop.c (maybe_mode_change): Likewise.
13250 * reginfo.c (init_reg_sets_1): Likewise.
13251 (choose_hard_reg_mode): Likewise.
13252 (simplifiable_subregs): Likewise.
13253 * regrename.c (check_new_reg_p): Likewise.
13254 * reload.c (find_valid_class): Likewise.
13255 (find_valid_class_1): Likewise.
13256 (reload_inner_reg_of_subreg): Likewise.
13257 (push_reload): Likewise.
13258 (combine_reloads): Likewise.
13259 (find_dummy_reload): Likewise.
13260 (find_reloads): Likewise.
13261 * reload1.c (find_reg): Likewise.
13262 (set_reload_reg): Likewise.
13263 (allocate_reload_reg): Likewise.
13264 (choose_reload_regs): Likewise.
13265 (reload_adjust_reg_for_temp): Likewise.
13266 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
13267 (simplify_subreg_regno): Likewise.
13268 * sel-sched.c (init_regs_for_mode): Likewise.
13269 * varasm.c (make_decl_rtl): Likewise.
13270 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
13271 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
13272 HARD_REGNO_MODE_OK.
13273 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
13274 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
13275 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13276 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
13277 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
13278 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13279 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
13280 (arc_mode_class): Delete.
13281 (HARD_REGNO_MODE_OK): Delete.
13282 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13283 (arc_hard_regno_mode_ok): Rename old array to...
13284 (arc_hard_regno_mode_ok_modes): ...this.
13285 (arc_conditional_register_usage): Update accordingly.
13286 (arc_mode_class): Make static.
13287 (arc_hard_regno_mode_ok): New function.
13288 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
13289 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
13290 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13291 (arm_hard_regno_mode_ok): Make static.
13292 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
13293 HARD_REGNO_MODE_OK.
13294 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
13295 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
13296 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
13298 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13299 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
13300 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
13301 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
13302 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
13303 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13304 * config/bfin/predicates.md (valid_reg_operand): Use
13305 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
13306 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
13307 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
13308 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13309 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
13310 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
13311 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13312 (cr16_hard_regno_mode_ok): Make static and return a bool.
13313 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
13314 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13315 (cris_hard_regno_mode_ok): New function.
13316 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
13317 (epiphany_mode_class): Delete.
13318 (HARD_REGNO_MODE_OK): Delete.
13319 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
13320 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13321 (hard_regno_mode_ok): Rename to...
13322 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
13323 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
13324 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
13325 HARD_REGNO_MODE_OK.
13326 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
13327 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
13328 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13329 (frv_hard_regno_mode_ok): Make static and return a bool.
13330 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
13331 HARD_REGNO_MODE_OK.
13332 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
13333 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
13334 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
13335 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
13337 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13338 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
13339 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
13340 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
13342 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13343 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
13344 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13345 (ia64_hard_regno_mode_ok): New function.
13346 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
13347 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13348 (iq2000_hard_regno_mode_ok): New function.
13349 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
13350 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13351 (lm32_hard_regno_mode_ok): New function.
13352 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
13353 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
13354 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
13355 instead of HARD_REGNO_MODE_OK.
13356 (m32c_hard_regno_ok): Rename to...
13357 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
13358 (m32c_cannot_change_mode_class): Update accordingly.
13359 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13360 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
13361 (m32r_mode_class): Delete.
13362 (HARD_REGNO_MODE_OK): Delete.
13363 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13364 (m32r_hard_regno_mode_ok): Rename to...
13365 (m32r_hard_regno_modes): ...this.
13366 (m32r_mode_class): Make static.
13367 (m32r_hard_regno_mode_ok): New function.
13368 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
13369 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
13370 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13371 (m68k_hard_regno_mode_ok): Make static.
13372 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
13373 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13374 (mcore_hard_regno_mode_ok): New function.
13375 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
13376 (HARD_REGNO_MODE_OK): Delete.
13377 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
13379 (microblaze_hard_regno_mode_ok_p): ...this and make static.
13380 (microblaze_hard_regno_mode_ok): New function.
13381 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13382 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
13383 (mips_hard_regno_mode_ok): Delete.
13384 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
13385 (mips_hard_regno_mode_ok_p): ...this and make static.
13386 (mips_hard_regno_mode_ok_p): Rename to...
13387 (mips_hard_regno_mode_ok_uncached): ...this.
13388 (mips_hard_regno_mode_ok): New function.
13389 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
13390 of HARD_REGNO_MODE_OK.
13391 (mips_option_override): Update after above name changes.
13392 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13393 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
13394 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
13395 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
13396 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
13397 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13398 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
13399 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
13400 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
13401 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13402 (msp430_hard_regno_mode_ok): Make static and return a bool.
13403 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
13404 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
13405 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
13407 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13408 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
13409 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
13410 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
13411 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
13412 (PA_HARD_REGNO_MODE_OK): ...this
13413 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
13414 (PA_HARD_REGNO_MODE_OK): ...this.
13415 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13416 (pa_hard_regno_mode_ok): New function.
13417 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
13418 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13419 (pdp11_hard_regno_mode_ok): New function.
13420 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
13421 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
13423 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
13425 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13426 (rs6000_hard_regno_mode_ok): Rename to...
13427 (rs6000_hard_regno_mode_ok_uncached): ...this.
13428 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13429 (rs6000_hard_regno_mode_ok): New function.
13430 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
13431 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
13432 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
13433 (riscv_hard_regno_mode_ok): ...this and make static.
13434 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13435 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
13436 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
13437 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13438 (rl78_hard_regno_mode_ok): Make static and return bool.
13439 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
13440 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
13442 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
13443 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13444 (rs6000_hard_regno_mode_ok): Rename to...
13445 (rs6000_hard_regno_mode_ok_uncached): ...this.
13446 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13447 (rs6000_hard_regno_mode_ok): New function.
13448 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
13449 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
13450 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13451 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
13452 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
13453 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
13454 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13455 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
13456 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
13457 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13458 (sh_hard_regno_mode_ok): Make static.
13459 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
13460 instead of HARD_REGNO_MODE_OK.
13461 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
13462 (sparc_mode_class): Delete.
13463 (HARD_REGNO_MODE_OK): Delete.
13464 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13465 (hard_regno_mode_classes): Make static.
13466 (sparc_mode_class): Likewise.
13467 (sparc_hard_regno_mode_ok): New function.
13468 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
13469 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
13470 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
13472 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13473 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
13474 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
13475 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
13476 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
13477 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13478 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
13479 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
13480 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13481 (visium_hard_regno_mode_ok): New function.
13482 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
13483 instead of HARD_REGNO_MODE_OK.
13484 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
13485 (HARD_REGNO_MODE_OK): Delete.
13486 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
13487 (xtensa_hard_regno_mode_ok_p): ...this and make static.
13488 (xtensa_option_override): Update accordingly.
13489 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13490 (xtensa_hard_regno_mode_ok): New function.
13491 * system.h (HARD_REGNO_MODE_OK): Poison.
13493 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13494 Alan Hayward <alan.hayward@arm.com>
13495 David Sherwood <david.sherwood@arm.com>
13497 * target.def (hard_regno_call_part_clobbered): New hook.
13498 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
13499 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
13500 * doc/tm.texi: Regenerate.
13501 * hooks.h (hook_bool_uint_mode_false): Declare.
13502 * hooks.c (hook_bool_uint_mode_false): New function.
13503 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13504 * cselib.c (cselib_process_insn): Use
13505 targetm.hard_regno_call_part_clobbered instead of
13506 HARD_REGNO_CALL_PART_CLOBBERED.
13507 * ira-conflicts.c (ira_build_conflicts): Likewise.
13508 * ira-costs.c (ira_tune_allocno_costs): Likewise.
13509 * lra-constraints.c (need_for_call_save_p): Likewise.
13510 * lra-lives.c: Include target.h.
13511 (check_pseudos_live_through_calls): Use
13512 targetm.hard_regno_call_part_clobbered instead of
13513 HARD_REGNO_CALL_PART_CLOBBERED.
13514 * regcprop.c: Include target.h.
13515 (copyprop_hardreg_forward_1): Use
13516 targetm.hard_regno_call_part_clobbered instead of
13517 HARD_REGNO_CALL_PART_CLOBBERED.
13518 * reginfo.c (choose_hard_reg_mode): Likewise.
13519 * regrename.c (check_new_reg_p): Likewise.
13520 * reload.c (find_equiv_reg): Likewise.
13521 * reload1.c (emit_reload_insns): Likewise.
13522 * sched-deps.c (deps_analyze_insn): Likewise.
13523 * sel-sched.c (init_regs_for_mode): Likewise.
13524 (mark_unavailable_hard_regs): Likewise.
13525 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
13526 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13527 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
13529 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13530 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13531 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
13533 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
13535 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13536 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13537 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
13539 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13540 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13541 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
13543 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13544 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
13546 * config/powerpcspe/powerpcspe.c
13547 (rs6000_hard_regno_call_part_clobbered): New function.
13548 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13549 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13550 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
13552 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13553 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13554 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
13556 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13557 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13558 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
13560 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13561 Alan Hayward <alan.hayward@arm.com>
13562 David Sherwood <david.sherwood@arm.com>
13564 * rtl.h (subreg_memory_offset): Declare.
13565 * emit-rtl.c (subreg_memory_offset): New function.
13566 * expmed.c (store_bit_field_1): Use it.
13567 * expr.c (undefined_operand_subword_p): Likewise.
13568 * simplify-rtx.c (simplify_subreg): Likewise.
13570 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
13572 PR rtl-optimization/57448
13575 * optabs.c (expand_atomic_load): Place compiler memory barriers if
13576 using atomic_load pattern.
13577 (expand_atomic_store): Likewise.
13579 2017-09-04 Jakub Jelinek <jakub@redhat.com>
13582 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
13583 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
13584 handling. Use replace_call_with_value with NULL instead of
13585 gsi_replace, unlink_stmt_vdef and release_defs.
13587 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
13590 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
13592 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13595 * rtl.h (emit_library_call_value_1): Declare.
13596 (emit_library_call): Replace declaration with a series of overloads.
13597 Remove the parameter count argument.
13598 (emit_library_call_value): Likewise.
13599 * calls.c (emit_library_call_value_1): Make global. Replace varargs
13600 with an "rtx_mode_t *".
13601 (emit_library_call_value): Delete.
13602 (emit_library_call): Likewise.
13603 * asan.c (asan_emit_stack_protection): Update calls accordingly.
13604 (asan_emit_allocas_unpoison): Likewise.
13605 * builtins.c (expand_builtin_powi): Likewise.
13606 (expand_asan_emit_allocas_unpoison): Likewise.
13607 * cfgexpand.c (expand_main_function): Likewise.
13608 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
13609 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
13610 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
13611 * config/arm/arm.c (arm_trampoline_init): Likewise.
13612 (arm_call_tls_get_addr): Likewise.
13613 (arm_expand_divmod_libfunc): Likewise.
13614 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
13615 (smulsi3_highpart): Likewise.
13616 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
13617 (c6x_expand_compare): Likewise.
13618 (c6x_expand_movmem): Likewise.
13619 * config/frv/frv.c (frv_trampoline_init): Likewise.
13620 * config/i386/i386.c (ix86_trampoline_init): Likewise.
13621 (ix86_expand_divmod_libfunc): Likewise.
13622 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
13623 (ia64_expand_compare): Likewise.
13624 (ia64_profile_hook): Likewise.
13625 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
13626 (nonlocal_goto): Likewise.
13627 (restore_stack_nonlocal): Likewise.
13628 * config/m32r/m32r.c (block_move_call): Likewise.
13629 (m32r_trampoline_init): Likewise.
13630 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
13631 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
13632 (m68k_call_m68k_read_tp): Likewise.
13633 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
13634 (microblaze_expand_divide): Likewise.
13635 * config/mips/mips.h (mips_args): Likewise.
13636 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
13637 (MIPS_ICACHE_SYNC): Likewise.
13638 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
13639 (nios2_trampoline_init): Likewise.
13640 * config/pa/pa.c (hppa_tls_call): Likewise.
13641 (pa_trampoline_init): Likewise.
13642 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
13643 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
13644 (expand_strn_compare): Likewise.
13645 (rs6000_generate_compare): Likewise.
13646 (rs6000_expand_float128_convert): Likewise.
13647 (output_profile_hook): Likewise.
13648 (rs6000_trampoline_init): Likewise.
13649 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
13650 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
13651 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
13652 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
13653 (rs6000_generate_compare): Likewise.
13654 (rs6000_expand_float128_convert): Likewise.
13655 (output_profile_hook): Likewise.
13656 (rs6000_trampoline_init): Likewise.
13657 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
13658 * config/sh/sh.c (sh_trampoline_init): Likewise.
13659 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
13660 (sparc_emit_float_lib_cmp): Likewise.
13661 (sparc32_initialize_trampoline): Likewise.
13662 (sparc64_initialize_trampoline): Likewise.
13663 (sparc_profile_hook): Likewise.
13664 * config/spu/spu.c (ea_load_store): Likewise.
13665 * config/spu/spu.md (floatunssidf2): Likewise.
13666 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
13667 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
13668 * config/visium/visium.c (expand_block_move_4): Likewise.
13669 (expand_block_move_2): Likewise.
13670 (expand_block_move_1): Likewise.
13671 (expand_block_set_4): Likewise.
13672 (expand_block_set_2): Likewise.
13673 (expand_block_set_1): Likewise.
13674 (visium_trampoline_init): Likewise.
13675 (visium_profile_hook): Likewise.
13676 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
13677 (xtensa_setup_frame_addresses): Likewise.
13678 (xtensa_trampoline_init): Likewise.
13679 * except.c (sjlj_emit_function_enter): Likewise.
13680 (sjlj_emit_function_exit): Likewise.
13681 * explow.c (allocate_dynamic_stack_space): Likewise.
13682 (probe_stack_range): Likewise.
13683 * expr.c (convert_mode_scalar): Likewise.
13684 * optabs.c (expand_binop): Likewise.
13685 (expand_twoval_binop_libfunc): Likewise.
13686 (expand_unop): Likewise.
13687 (prepare_cmp_insn): Likewise.
13688 (prepare_float_lib_cmp): Likewise.
13689 (expand_float): Likewise.
13690 (expand_fix): Likewise.
13691 (expand_fixed_convert): Likewise.
13692 (maybe_emit_sync_lock_test_and_set): Likewise.
13693 (expand_atomic_compare_and_swap): Likewise.
13694 (expand_mem_thread_fence): Likewise.
13695 (expand_atomic_fetch_op): Likewise.
13697 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
13699 * doc/generic.texi (OpenACC): Adjust URL.
13700 * doc/invoke.texi (C Dialect Options): Ditto.
13702 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
13704 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
13705 predicate for operand 1. Add (m,<S>) constraint.
13706 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
13707 Prevent memory operand 1 with register operand 2.
13709 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
13711 PR rtl-optimization/82024
13712 * combine.c (try_combine): If the combination result is a PARALLEL,
13713 and we only need to retain the SET in there that would be placed
13714 at I2, check that we can place that at I3 instead, before doing so.
13716 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13719 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
13721 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
13722 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
13725 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
13726 Jeff Law <law@redhat.com>
13728 * varasm.c (bss_initializer_p): Do not put constants into .bss
13729 (categorize_decl_for_section): Handle bss_initializer_p returning
13730 false when DECL_INITIAL is NULL.
13732 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13735 * config/s390/s390.c (s390_can_inline_p): New function.
13737 2017-09-01 Jeff Law <law@redhat.com>
13739 PR tree-optimization/82052
13740 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13741 Always initialize the returned slot after a hash table miss
13742 when INSERT is true.
13744 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
13746 * config/s390/s390.md (mem_signal_fence): Remove.
13747 * doc/md.texi (mem_signal_fence): Remove.
13748 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
13750 * target-insns.def (mem_signal_fence): Remove.
13752 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13755 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
13758 * asan.c (create_odr_indicator): Strip name encoding from assembler
13759 name before appending it after __odr_asan_.
13761 2017-09-01 Martin Liska <mliska@suse.cz>
13763 PR tree-optimization/82059
13764 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
13765 frequency only when an edge is redirected.
13767 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13769 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
13770 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
13771 (arc_conditional_register_usage): Remove ARC600 lp_count
13773 (arc_file_start): Emit Tag_ARC_CPU_variation.
13774 (arc_can_use_doloop_p): New conditions to use ZOLs.
13775 (hwloop_fail): New function.
13776 (hwloop_optimize): Likewise.
13777 (hwloop_pattern_reg): Likewise.
13778 (arc_doloop_hooks): New struct, to be used with reorg_loops.
13779 (arc_reorg_loops): New function, calls reorg_loops.
13780 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
13781 (arc600_corereg_hazard): Remove ZOL checking, case handled by
13783 (arc_loop_hazard): Remove function, functionality moved into
13785 (arc_hazard): Remove arc_loop_hazard call.
13786 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
13787 into hwloop_optimize.
13788 (arc_label_align): Remove ZOL handling.
13789 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
13790 * config/arc/arc.md (doloop_begin): Remove pattern.
13791 (doloop_begin_i): Likewise.
13792 (doloop_end_i): Likewise.
13793 (doloop_fallback): Likewise.
13794 (doloop_fallback_m): Likewise.
13795 (doloop_end): Reimplement expand.
13796 (arc_lp): New pattern for LP instruction.
13797 (loop_end): New pattern.
13798 (loop_fail): Likewise.
13799 (decrement_and_branch_until_zero): Likewise.
13800 * config/arc/arc.opt (mlpc-width): New option.
13801 * doc/invoke.texi (mlpc-width): Document option.
13803 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13805 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
13806 (arc_ccfsm_advance): Fix checking for delay slots.
13807 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
13809 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13811 * config/arc/arc.md (movqi_insn): Add stores to save constant long
13813 (movhi_insn): Update store instruction constraint which are saving
13814 6-bit short immediates.
13815 (movsi_insn): Consider also short scaled load operations.
13816 (zero_extendhisi2_i): Use Usd constraint instead of T.
13817 (extendhisi2_i): Add q constraint.
13818 (arc_clzsi2): Add type and length attributes.
13819 (arc_ctzsi2): Likewise.
13820 * config/arc/constraints.md (Usc): Update constraint, the
13821 assembler can parse two relocations for a single instruction.
13823 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13825 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
13826 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
13828 2017-08-31 Olivier Hainque <hainque@adacore.com>
13830 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
13831 match as powerpc-wrs-vxworks*.
13833 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
13835 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
13836 register constraint for by-element operand.
13837 (aarch64_mls_elt_merge<mode>): Likewise.
13839 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13841 * config/arc/arc.c (arc_can_follow_jump): Check for short
13844 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13846 * config.gcc: Use g.opt for arc.
13847 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
13848 functionality moved to ...
13849 (legitimate_scaled_address_p): New function, ...here.
13850 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
13851 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
13852 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
13854 (arc_override_options): Handle G option.
13855 (arc_output_pic_addr_const): Correct function definition.
13856 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
13857 (arc_decl_anon_ns_mem_p): Delete.
13858 (arc_in_small_data_p): Overhaul this function to take into
13859 consideration the value given via G option.
13860 (arc_rewrite_small_data_1): Renamed and corrected old
13861 arc_rewrite_small_data function.
13862 (arc_rewrite_small_data): New function.
13863 (small_data_pattern): Don't use pic_offset_table_rtx.
13864 * config/arc/arc.h (CC1_SPEC): Recognize G option.
13865 * config/arc/simdext.md (movmisalignv2hi): Use
13866 prepare_move_operands function.
13868 (movmisalign*): Likewise.
13869 * doc/invoke.texi (ARC options): Document -G option.
13871 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13873 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
13875 * config/arc/arc.c (arc_print_operand): Output scalled address for
13876 sdata whenever is possible.
13877 (arc_in_small_data_p): Allow sdata for 64bit datum when double
13878 load/stores are available.
13879 (compact_sda_memory_operand): Check for the alignment required by
13880 code density instructions.
13881 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
13883 * config/arc/constraints.md (Usd): Update constraint.
13884 (Us0): New constraint.
13885 (Usc): Update constraint.
13887 2017-08-31 Richard Biener <rguenther@suse.de>
13889 PR middle-end/82054
13890 * dwarf2out.c (dwarf2out_early_global_decl): Process each
13891 function only once.
13893 2017-08-31 Tamar Christina <tamar.christina@arm.com>
13895 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
13896 Resize type_signature.
13898 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13899 Alan Hayward <alan.hayward@arm.com>
13900 David Sherwood <david.sherwood@arm.com>
13902 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
13903 subregs whose inner modes can be stored in GPRs.
13904 (aarch64_classify_index): Likewise.
13906 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13907 Alan Hayward <alan.hayward@arm.com>
13908 David Sherwood <david.sherwood@arm.com>
13910 * config/aarch64/iterators.md (V_cmp_result): Rename to...
13911 (V_INT_EQUIV): ...this.
13912 (v_cmp_result): Rename to...
13913 (v_int_equiv): ...this.
13914 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
13915 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
13916 (copysign<mode>3): Likewise.
13917 (aarch64_simd_bsl<mode>_internal): Likewise.
13918 (aarch64_simd_bsl<mode>): Likewise.
13919 (vec_cmp<mode><mode>): Likewise.
13920 (vcond<mode><mode>): Likewise.
13921 (vcond<v_cmp_mixed><mode>): Likewise.
13922 (vcondu<mode><v_cmp_mixed>): Likewise.
13923 (aarch64_cm<optab><mode>): Likewise.
13924 (aarch64_cmtst<mode>): Likewise.
13925 (aarch64_fac<optab><mode>): Likewise.
13926 (vec_perm_const<mode>): Likewise.
13927 (vcond_mask_<mode><v_cmp_result>): Rename to...
13928 (vcond_mask_<mode><v_int_equiv>): ...this.
13929 (vec_cmp<mode><v_cmp_result>): Rename to...
13930 (vec_cmp<mode><v_int_equiv>): ...this.
13932 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13933 Alan Hayward <alan.hayward@arm.com>
13934 David Sherwood <david.sherwood@arm.com>
13936 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
13938 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
13939 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
13940 (UNSPEC_LD4_DREG): New unspecs.
13941 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
13942 (aarch64_ld2<mode>_dreg_be): Replace with...
13943 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
13945 (aarch64_ld3<mode>_dreg_le)
13946 (aarch64_ld3<mode>_dreg_be): Replace with...
13947 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
13949 (aarch64_ld4<mode>_dreg_le)
13950 (aarch64_ld4<mode>_dreg_be): Replace with...
13951 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
13954 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13956 PR tree-optimization/81987
13957 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
13958 insert an initializer in a location not dominated by the stride
13961 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13963 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
13964 on the entire header of the finally block in the fallthru case.
13966 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13968 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
13970 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
13972 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
13973 rs6000_emit_move_from_cr and call renamed function.
13974 (rs6000_emit_prologue): Call renamed functions.
13975 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
13976 movesi_from_cr, remove volatile CRs.
13978 2017-08-30 Jon Beniston <jon@beniston.com>
13979 Richard Biener <rguenther@suse.de>
13981 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
13982 of VECTOR_MODE_P check.
13983 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
13984 element vector types.
13986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13988 * df.h (df_read_modify_subreg_p): Remove in favor of...
13989 * rtl.h (read_modify_subreg_p): ...this new function. Take a
13990 const_rtx instead of an rtx.
13991 * cprop.c (local_cprop_find_used_regs): Update accordingly.
13992 * df-problems.c (df_word_lr_mark_ref): Likewise.
13993 * ira-lives.c (mark_pseudo_reg_live): Likewise.
13994 (mark_pseudo_reg_dead): Likewise.
13995 (mark_ref_dead): Likewise.
13996 * reginfo.c (init_subregs_of_mode): Likewise.
13997 * sched-deps.c (sched_analyze_1): Likewise.
13998 * df-scan.c (df_def_record_1): Likewise.
13999 (df_uses_record): Likewise.
14000 (df_read_modify_subreg_p): Remove in favor of...
14001 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
14002 const_rtx instead of an rtx.
14004 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14005 Alan Hayward <alan.hayward@arm.com>
14006 David Sherwood <david.sherwood@arm.com>
14008 * rtl.h (partial_subreg_p): New function.
14009 * caller-save.c (save_call_clobbered_regs): Use it.
14010 * calls.c (expand_call): Likewise.
14011 * combine.c (combinable_i3pat): Likewise.
14012 (simplify_set): Likewise.
14013 (make_extraction): Likewise.
14014 (make_compound_operation_int): Likewise.
14015 (gen_lowpart_or_truncate): Likewise.
14016 (force_to_mode): Likewise.
14017 (make_field_assignment): Likewise.
14018 (reg_truncated_to_mode): Likewise.
14019 (record_truncated_value): Likewise.
14020 (move_deaths): Likewise.
14021 * cse.c (record_jump_cond): Likewise.
14022 (cse_insn): Likewise.
14023 * cselib.c (cselib_lookup_1): Likewise.
14024 * expmed.c (extract_bit_field_using_extv): Likewise.
14025 * function.c (assign_parm_setup_reg): Likewise.
14026 * ifcvt.c (noce_convert_multiple_sets): Likewise.
14027 * ira-build.c (create_insn_allocnos): Likewise.
14028 * lra-coalesce.c (merge_pseudos): Likewise.
14029 * lra-constraints.c (match_reload): Likewise.
14030 (simplify_operand_subreg): Likewise.
14031 (curr_insn_transform): Likewise.
14032 * lra-lives.c (process_bb_lives): Likewise.
14033 * lra.c (new_insn_reg): Likewise.
14034 (lra_substitute_pseudo): Likewise.
14035 * regcprop.c (mode_change_ok): Likewise.
14036 (maybe_mode_change): Likewise.
14037 (copyprop_hardreg_forward_1): Likewise.
14038 * reload.c (push_reload): Likewise.
14039 (find_reloads): Likewise.
14040 (find_reloads_subreg_address): Likewise.
14041 * reload1.c (alter_reg): Likewise.
14042 (eliminate_regs_1): Likewise.
14043 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14045 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
14047 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
14048 back to if statements, including unpack.
14050 2017-08-30 Martin Liska <mliska@suse.cz>
14052 PR inline-asm/82001
14053 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14055 (func_checker::compare_asm_inputs_outputs): ... this function.
14056 (func_checker::compare_gimple_asm): Use the function to compare
14057 also ASM constrains.
14058 * ipa-icf-gimple.h: Rename the function.
14060 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14061 Alan Hayward <alan.hayward@arm.com>
14062 David Sherwood <david.sherwood@arm.com>
14064 * coretypes.h (complex_mode): New type.
14065 * gdbhooks.py (build_pretty_printer): Handle it.
14066 * machmode.h (complex_mode): New class.
14067 (complex_mode::includes_p): New function.
14068 (is_complex_int_mode): Likewise.
14069 (is_complex_float_mode): Likewise.
14070 * genmodes.c (get_mode_class): Handle complex mode classes.
14071 * function.c (expand_function_end): Use is_complex_int_mode.
14073 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14074 Alan Hayward <alan.hayward@arm.com>
14075 David Sherwood <david.sherwood@arm.com>
14077 * coretypes.h (scalar_mode_pod): New typedef.
14078 * gdbhooks.py (build_pretty_printer): Handle it.
14079 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14080 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14081 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14082 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14083 as_a <scalar_mode>.
14085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14086 Alan Hayward <alan.hayward@arm.com>
14087 David Sherwood <david.sherwood@arm.com>
14089 * machmode.h (mode_for_vector): Take a scalar_mode instead
14091 * stor-layout.c (mode_for_vector): Likewise.
14092 * explow.c (promote_mode): Use as_a <scalar_mode>.
14093 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14096 Alan Hayward <alan.hayward@arm.com>
14097 David Sherwood <david.sherwood@arm.com>
14099 * target.def (preferred_simd_mode): Take a scalar_mode
14100 instead of a machine_mode.
14101 * targhooks.h (default_preferred_simd_mode): Likewise.
14102 * targhooks.c (default_preferred_simd_mode): Likewise.
14103 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14104 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14105 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14106 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14107 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14108 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14109 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14110 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14112 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14113 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
14114 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
14115 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
14116 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
14117 * doc/tm.texi: Regenerate.
14118 * optabs-query.c (can_vec_mask_load_store_p): Return false for
14121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14122 Alan Hayward <alan.hayward@arm.com>
14123 David Sherwood <david.sherwood@arm.com>
14125 * target.def (scalar_mode_supported_p): Take a scalar_mode
14126 instead of a machine_mode.
14127 * targhooks.h (default_scalar_mode_supported_p): Likewise.
14128 * targhooks.c (default_scalar_mode_supported_p): Likewise.
14129 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
14130 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14131 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
14132 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
14133 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
14134 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
14135 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
14136 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
14137 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
14138 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
14139 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
14140 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
14142 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
14143 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
14144 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14145 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14146 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
14148 * doc/tm.texi: Regenerate.
14150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14151 Alan Hayward <alan.hayward@arm.com>
14152 David Sherwood <david.sherwood@arm.com>
14154 * coretypes.h (opt_scalar_mode): New typedef.
14155 * gdbhooks.py (build_pretty_printers): Handle it.
14156 * machmode.h (mode_iterator::get_2xwider): Add overload for
14158 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
14160 * expr.c (convert_mode_scalar): Likewise.
14161 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14162 * optabs.c (expand_float): Likewise.
14163 (expand_fix): Likewise.
14164 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14166 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14167 Alan Hayward <alan.hayward@arm.com>
14168 David Sherwood <david.sherwood@arm.com>
14170 * optabs.c (expand_float): Explicitly check for scalars before
14171 using a branching expansion.
14172 (expand_fix): Likewise.
14174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14175 Alan Hayward <alan.hayward@arm.com>
14176 David Sherwood <david.sherwood@arm.com>
14178 * expr.c (convert_mode): Split scalar handling out into...
14179 (convert_mode_scalar): ...this new function. Treat the modes
14182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14183 Alan Hayward <alan.hayward@arm.com>
14184 David Sherwood <david.sherwood@arm.com>
14186 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
14188 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
14190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14191 Alan Hayward <alan.hayward@arm.com>
14192 David Sherwood <david.sherwood@arm.com>
14194 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
14195 rather than a machine_mode.
14196 (fixed_from_string): Likewise.
14197 (fixed_convert): Likewise.
14198 (fixed_convert_from_int): Likewise.
14199 (fixed_convert_from_real): Likewise.
14200 (real_convert_from_fixed): Likewise.
14201 * fixed-value.c (fixed_from_double_int): Likewise.
14202 (fixed_from_string): Likewise.
14203 (fixed_convert): Likewise.
14204 (fixed_convert_from_int): Likewise.
14205 (fixed_convert_from_real): Likewise.
14206 (real_convert_from_fixed): Likewise.
14207 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
14209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14210 Alan Hayward <alan.hayward@arm.com>
14211 David Sherwood <david.sherwood@arm.com>
14213 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
14214 of separate mode class checks. Do not allow vector modes here.
14215 (immed_wide_int_const): Use as_a <scalar_mode>.
14216 * explow.c (trunc_int_for_mode): Likewise.
14217 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
14218 (wi::shwi): Likewise.
14219 (wi::min_value): Likewise.
14220 (wi::max_value): Likewise.
14221 * dwarf2out.c (loc_descriptor): Likewise.
14222 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
14223 for CONST_WIDE_INT.
14225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14226 Alan Hayward <alan.hayward@arm.com>
14227 David Sherwood <david.sherwood@arm.com>
14229 * tree.h (SCALAR_TYPE_MODE): New macro.
14230 * expr.c (expand_expr_addr_expr_1): Use it.
14231 (expand_expr_real_2): Likewise.
14232 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
14233 (fold_convert_const_fixed_from_int): Likewise.
14234 (fold_convert_const_fixed_from_real): Likewise.
14235 (native_encode_fixed): Likewise
14236 (native_encode_complex): Likewise
14237 (native_encode_vector): Likewise.
14238 (native_interpret_fixed): Likewise.
14239 (native_interpret_real): Likewise.
14240 (native_interpret_complex): Likewise.
14241 (native_interpret_vector): Likewise.
14242 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
14243 (simd_clone_adjust_argument_types): Likewise.
14244 (simd_clone_init_simd_arrays): Likewise.
14245 (simd_clone_adjust): Likewise.
14246 * stor-layout.c (layout_type): Likewise.
14247 * tree.c (build_minus_one_cst): Likewise.
14248 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14249 * tree-inline.c (estimate_move_cost): Likewise.
14250 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
14251 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
14252 (vectorizable_reduction): Likewise.
14253 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
14254 (vect_recog_mixed_size_cond_pattern): Likewise.
14255 (check_bool_pattern): Likewise.
14256 (adjust_bool_pattern): Likewise.
14257 (search_type_for_mask_1): Likewise.
14258 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
14259 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14260 (vectorizable_load): Likewise.
14261 (vectorizable_store): Likewise.
14262 * ubsan.c (ubsan_encode_value): Likewise.
14263 * varasm.c (output_constant): Likewise.
14265 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14266 Alan Hayward <alan.hayward@arm.com>
14267 David Sherwood <david.sherwood@arm.com>
14269 * coretypes.h (scalar_mode): New class.
14270 * machmode.h (scalar_mode): Likewise.
14271 (scalar_mode::includes_p): New function.
14272 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
14273 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
14274 * genmodes.c (get_mode_class): Handle remaining scalar modes.
14275 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
14276 * expmed.c (store_bit_field_1): Likewise.
14277 (extract_bit_field_1): Likewise.
14278 * expr.c (write_complex_part): Likewise.
14279 (read_complex_part): Likewise.
14280 (emit_move_complex_push): Likewise.
14281 (expand_expr_real_2): Likewise.
14282 * function.c (assign_parm_setup_reg): Likewise.
14283 (assign_parms_unsplit_complex): Likewise.
14284 * optabs.c (expand_binop): Likewise.
14285 * rtlanal.c (subreg_get_info): Likewise.
14286 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14287 * varasm.c (output_constant_pool_2): Likewise.
14289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14290 Alan Hayward <alan.hayward@arm.com>
14291 David Sherwood <david.sherwood@arm.com>
14293 * expmed.c (extract_high_half): Use scalar_int_mode and remove
14295 (expmed_mult_highpart_optab): Likewise.
14296 (expmed_mult_highpart): Likewise.
14298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14299 Alan Hayward <alan.hayward@arm.com>
14300 David Sherwood <david.sherwood@arm.com>
14302 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
14303 instead of a machine_mode.
14304 (builtin_memset_read_str): Likewise.
14305 * builtins.c (c_readstr): Likewise.
14306 (builtin_memcpy_read_str): Likewise.
14307 (builtin_strncpy_read_str): Likewise.
14308 (builtin_memset_read_str): Likewise.
14309 (builtin_memset_gen_str): Likewise.
14310 (expand_builtin_signbit): Use scalar_int_mode for local variables.
14311 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
14312 instead of a machine_mode.
14313 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
14315 (make_extraction): Likewise.
14316 (try_widen_shift_mode): Take and return scalar_int_modes instead
14318 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
14319 a scalar_int_mode instead of a machine_mode.
14320 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
14321 (avr_addr_space_pointer_mode): Likewise.
14322 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
14323 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
14324 (msp430_unwind_word_mode): Likewise.
14325 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
14326 (spu_addr_space_pointer_mode): Likewise.
14327 (spu_addr_space_address_mode): Likewise.
14328 (spu_libgcc_cmp_return_mode): Likewise.
14329 (spu_libgcc_shift_count_mode): Likewise.
14330 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
14331 (rl78_addr_space_pointer_mode): Likewise.
14332 (fl78_unwind_word_mode): Likewise.
14333 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
14335 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
14336 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
14337 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
14338 (mips_valid_pointer_mode): Likewise.
14339 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
14340 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
14341 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
14343 (ft32_addr_space_address_mode): Likewise.
14344 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
14345 scalar_int_mode instead of a machine_mode.
14346 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
14348 (m32c_addr_space_address_mode): Likewise.
14349 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
14350 (rs6000_eh_return_filter_mode): Likewise.
14351 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
14352 (rs6000_eh_return_filter_mode): Likewise.
14353 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
14354 (s390_libgcc_shift_count_mode): Likewise.
14355 (s390_unwind_word_mode): Likewise.
14356 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
14358 * target.def (mode_rep_extended): Likewise.
14359 (valid_pointer_mode): Likewise.
14360 (addr_space.valid_pointer_mode): Likewise.
14361 (eh_return_filter_mode): Return a scalar_int_mode rather than
14363 (libgcc_cmp_return_mode): Likewise.
14364 (libgcc_shift_count_mode): Likewise.
14365 (unwind_word_mode): Likewise.
14366 (addr_space.pointer_mode): Likewise.
14367 (addr_space.address_mode): Likewise.
14368 * doc/tm.texi: Regenerate.
14369 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
14371 (do_jump): Use scalar_int_mode for local variables.
14372 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
14373 rather than a machine_mode.
14374 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
14375 (scompare_loc_descriptor_wide): Likewise.
14376 (scompare_loc_descriptor_narrow): Likewise.
14377 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
14379 * except.c (sjlj_emit_dispatch_table): Likewise.
14380 (expand_builtin_eh_copy_values): Likewise.
14381 * explow.c (convert_memory_address_addr_space_1): Likewise.
14382 Take a scalar_int_mode rather than a machine_mode.
14383 (convert_memory_address_addr_space): Take a scalar_int_mode rather
14384 than a machine_mode.
14385 (memory_address_addr_space): Use scalar_int_mode for local variables.
14386 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
14387 rather than a machine_mode.
14388 * expmed.c (mask_rtx): Likewise.
14389 (init_expmed_one_conv): Likewise.
14390 (expand_mult_highpart_adjust): Likewise.
14391 (extract_high_half): Likewise.
14392 (expmed_mult_highpart_optab): Likewise.
14393 (expmed_mult_highpart): Likewise.
14394 (expand_smod_pow2): Likewise.
14395 (expand_sdiv_pow2): Likewise.
14396 (emit_store_flag_int): Likewise.
14397 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
14399 (extract_low_bits): Likewise.
14400 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
14402 * expr.c (pieces_addr::adjust): Likewise.
14403 (can_store_by_pieces): Likewise.
14404 (store_by_pieces): Likewise.
14405 (clear_by_pieces_1): Likewise.
14406 (expand_expr_addr_expr_1): Likewise.
14407 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
14408 (expand_expr_real_1): Likewise.
14409 (try_casesi): Likewise.
14410 * final.c (shorten_branches): Likewise.
14411 * fold-const.c (fold_convert_const_int_from_fixed): Change the
14412 type of "mode" to machine_mode.
14413 * internal-fn.c (expand_arith_overflow_result_store): Take a
14414 scalar_int_mode rather than a machine_mode.
14415 (expand_mul_overflow): Use scalar_int_mode for local variables.
14416 * loop-doloop.c (doloop_modify): Likewise.
14417 (doloop_optimize): Likewise.
14418 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
14419 than a machine_mode.
14420 (expand_doubleword_shift_condmove): Likewise.
14421 (expand_doubleword_shift): Likewise.
14422 (expand_doubleword_clz): Likewise.
14423 (expand_doubleword_popcount): Likewise.
14424 (expand_doubleword_parity): Likewise.
14425 (expand_absneg_bit): Use scalar_int_mode for local variables.
14426 (prepare_float_lib_cmp): Likewise.
14427 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
14428 rather than a machine_mode.
14429 (convert_memory_address_addr_space): Likewise.
14430 (get_mode_bounds): Likewise.
14431 (get_address_mode): Return a scalar_int_mode rather than a
14433 * rtlanal.c (get_address_mode): Likewise.
14434 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
14435 than a machine_mode.
14436 * targhooks.c (default_mode_rep_extended): Likewise.
14437 (default_valid_pointer_mode): Likewise.
14438 (default_addr_space_valid_pointer_mode): Likewise.
14439 (default_eh_return_filter_mode): Return a scalar_int_mode rather
14440 than a machine_mode.
14441 (default_libgcc_cmp_return_mode): Likewise.
14442 (default_libgcc_shift_count_mode): Likewise.
14443 (default_unwind_word_mode): Likewise.
14444 (default_addr_space_pointer_mode): Likewise.
14445 (default_addr_space_address_mode): Likewise.
14446 * targhooks.h (default_eh_return_filter_mode): Likewise.
14447 (default_libgcc_cmp_return_mode): Likewise.
14448 (default_libgcc_shift_count_mode): Likewise.
14449 (default_unwind_word_mode): Likewise.
14450 (default_addr_space_pointer_mode): Likewise.
14451 (default_addr_space_address_mode): Likewise.
14452 (default_mode_rep_extended): Take a scalar_int_mode rather than
14454 (default_valid_pointer_mode): Likewise.
14455 (default_addr_space_valid_pointer_mode): Likewise.
14456 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
14458 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
14459 rather than a machine_mode.
14460 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
14461 for local variables.
14462 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14463 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
14464 than a machine_mode.
14466 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14467 Alan Hayward <alan.hayward@arm.com>
14468 David Sherwood <david.sherwood@arm.com>
14470 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
14471 the mode argument to scalar_int_mode.
14472 (do_jump_by_parts_zero_rtx): Likewise.
14473 (do_jump_by_parts_equality_rtx): Likewise.
14474 (do_jump_by_parts_greater): Take a mode argument.
14475 (do_jump_by_parts_equality): Likewise.
14476 (do_jump_1): Update calls accordingly.
14478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14479 Alan Hayward <alan.hayward@arm.com>
14480 David Sherwood <david.sherwood@arm.com>
14482 * is-a.h (safe_dyn_cast): New function.
14483 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
14484 (jump_table_for_label): Likewise.
14485 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
14486 instead of an rtx_insn *.
14487 (shorten_branches): Use dyn_cast instead of LABEL_P and
14488 JUMP_TABLE_DATA_P. Use jump_table_for_label and
14489 rtx_jump_table_data::get_data_mode.
14490 (final_scan_insn): Likewise.
14492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14493 Alan Hayward <alan.hayward@arm.com>
14494 David Sherwood <david.sherwood@arm.com>
14496 * combine.c (try_combine): Use is_a <scalar_int_mode> when
14497 trying to combine a full-register integer set with a subreg
14500 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14501 Alan Hayward <alan.hayward@arm.com>
14502 David Sherwood <david.sherwood@arm.com>
14504 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
14505 that is always either address_mode or pointer_mode.
14507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14508 Alan Hayward <alan.hayward@arm.com>
14509 David Sherwood <david.sherwood@arm.com>
14511 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
14512 when the two are known to be equal.
14514 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14515 Alan Hayward <alan.hayward@arm.com>
14516 David Sherwood <david.sherwood@arm.com>
14518 * simplify-rtx.c (simplify_const_unary_operation): Use
14519 is_a <scalar_int_mode> instead of checking for a nonzero
14520 precision. Forcibly convert op_mode to a scalar_int_mode
14521 in that case. More clearly differentiate the operand and
14522 result modes and use the former when deciding what the value
14523 of a count-bits operation should be. Use is_int_mode instead
14524 of checking for a MODE_INT. Remove redundant check for whether
14525 this mode has a zero precision.
14527 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14528 Alan Hayward <alan.hayward@arm.com>
14529 David Sherwood <david.sherwood@arm.com>
14531 * optabs.c (widen_leading): Change the type of the mode argument
14532 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
14533 (widen_bswap): Likewise.
14534 (expand_parity): Likewise.
14535 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
14536 (expand_ffs): Likewise.
14537 (epand_unop): Check for scalar integer modes before calling the
14540 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14541 Alan Hayward <alan.hayward@arm.com>
14542 David Sherwood <david.sherwood@arm.com>
14544 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
14546 (expand_expr_real_1): Update call accordingly.
14548 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14549 Alan Hayward <alan.hayward@arm.com>
14550 David Sherwood <david.sherwood@arm.com>
14552 * expmed.c (store_bit_field_using_insv): Add op0_mode and
14553 value_mode arguments. Use scalar_int_mode internally.
14554 (store_bit_field_1): Rename the new integer mode from imode
14555 to op0_mode and use it instead of GET_MODE (op0). Update calls
14556 to store_split_bit_field, store_bit_field_using_insv and
14557 store_fixed_bit_field.
14558 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
14559 Use scalar_int_mode internally. Use a bit count rather than a mode
14560 when calculating the largest bit size for get_best_mode.
14561 Update calls to store_split_bit_field and store_fixed_bit_field_1.
14562 (store_fixed_bit_field_1): Add mode and value_mode arguments.
14563 Remove assertion that OP0 has a scalar integer mode.
14564 (store_split_bit_field): Add op0_mode and value_mode arguments.
14565 Update calls to extract_fixed_bit_field.
14566 (extract_bit_field_using_extv): Add an op0_mode argument.
14567 Use scalar_int_mode internally.
14568 (extract_bit_field_1): Rename the new integer mode from imode to
14569 op0_mode and use it instead of GET_MODE (op0). Update calls to
14570 extract_split_bit_field, extract_bit_field_using_extv and
14571 extract_fixed_bit_field.
14572 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
14573 to extract_split_bit_field and extract_fixed_bit_field_1.
14574 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
14575 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
14576 on the target mode.
14577 (extract_split_bit_field): Add an op0_mode argument. Update call
14578 to extract_fixed_bit_field.
14580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14581 Alan Hayward <alan.hayward@arm.com>
14582 David Sherwood <david.sherwood@arm.com>
14584 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
14585 * explow.c (hard_function_value): Likewise.
14586 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
14587 convert_to_mode call outside the loop.
14588 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
14589 for the mode iterator. Require the mode specified by max_pieces
14591 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
14593 (copy_blkmode_to_reg): Likewise.
14594 (set_storage_via_setmem): Likewise.
14595 * optabs.c (prepare_cmp_insn): Likewise.
14596 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14597 * stor-layout.c (finish_bitfield_representative): Likewise.
14599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14600 Alan Hayward <alan.hayward@arm.com>
14601 David Sherwood <david.sherwood@arm.com>
14603 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
14604 * expr.c (convert_move): Use them.
14605 (convert_modes): Likewise.
14606 (store_expr_with_bounds): Likewise.
14608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14609 Alan Hayward <alan.hayward@arm.com>
14610 David Sherwood <david.sherwood@arm.com>
14612 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
14613 parameter for the mode of "x". Remove the "known_x", "known_mode"
14614 and "known_ret" arguments. Change the type of the mode argument
14615 to scalar_int_mode.
14616 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
14617 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
14618 (reg_num_sign_bit_copies_for_combine): Likewise.
14619 * rtlanal.c (nonzero_bits1): Likewise.
14620 (num_sign_bit_copies1): Likewise.
14621 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
14622 (reg_num_sign_bit_copies_general): Likewise.
14623 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
14624 (reg_nonzero_bits_general): Likewise.
14626 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14627 Alan Hayward <alan.hayward@arm.com>
14628 David Sherwood <david.sherwood@arm.com>
14630 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
14631 than in subroutines. Return 1 for non-integer modes.
14632 (cached_num_sign_bit_copies): Change the type of the mode parameter
14633 to scalar_int_mode.
14634 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
14635 classes. Handle CONST_INT_P first and then check whether X also
14636 has a scalar integer mode. Check the same thing for inner registers
14637 of a SUBREG and for values that are being extended or truncated.
14639 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14640 Alan Hayward <alan.hayward@arm.com>
14641 David Sherwood <david.sherwood@arm.com>
14643 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
14644 in subroutines. Return the mode mask for non-integer modes.
14645 (cached_nonzero_bits): Change the type of the mode parameter
14646 to scalar_int_mode.
14647 (nonzero_bits1): Likewise. Remove early exit for other mode
14648 classes. Handle CONST_INT_P first and then check whether X
14649 also has a scalar integer mode.
14651 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14652 Alan Hayward <alan.hayward@arm.com>
14653 David Sherwood <david.sherwood@arm.com>
14655 * expr.c (widest_int_mode_for_size): Make the comment match the code.
14656 Return a scalar_int_mode and assert that the size is greater than
14658 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
14659 (op_by_pieces_d::op_by_pieces_d): Likewise.
14660 (op_by_pieces_d::run): Likewise.
14661 (can_store_by_pieces): Likewise.
14663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14664 Alan Hayward <alan.hayward@arm.com>
14665 David Sherwood <david.sherwood@arm.com>
14667 * combine.c (extract_left_shift): Add a mode argument and update
14669 (make_compound_operation_int): Change the type of the mode parameter
14670 to scalar_int_mode and update the call to extract_left_shift.
14672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14673 Alan Hayward <alan.hayward@arm.com>
14674 David Sherwood <david.sherwood@arm.com>
14676 * combine.c (simplify_and_const_int): Change the type of the mode
14677 parameter to scalar_int_mode.
14678 (simplify_and_const_int_1): Likewise. Update recursive call.
14680 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14681 Alan Hayward <alan.hayward@arm.com>
14682 David Sherwood <david.sherwood@arm.com>
14684 * combine.c (simplify_compare_const): Check that the mode is a
14685 scalar_int_mode (rather than VOIDmode) before testing its
14687 (simplify_comparison): Move COMPARISON_P handling out of the
14688 loop and restrict the latter part of the loop to scalar_int_modes.
14689 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
14690 and when considering SUBREG_REGs. Use is_int_mode instead of
14691 checking GET_MODE_CLASS against MODE_INT.
14693 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14694 Alan Hayward <alan.hayward@arm.com>
14695 David Sherwood <david.sherwood@arm.com>
14697 * combine.c (try_widen_shift_mode): Move check for equal modes to...
14698 (simplify_shift_const_1): ...here. Use scalar_int_mode for
14699 shift_unit_mode and for modes involved in scalar shifts.
14701 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14702 Alan Hayward <alan.hayward@arm.com>
14703 David Sherwood <david.sherwood@arm.com>
14705 * combine.c (force_int_to_mode): New function, split out from...
14706 (force_to_mode): ...here. Keep xmode up-to-date and use it
14707 instead of GET_MODE (x).
14709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14710 Alan Hayward <alan.hayward@arm.com>
14711 David Sherwood <david.sherwood@arm.com>
14713 * optabs-query.h (extraction_insn::struct_mode): Change type to
14714 opt_scalar_int_mode and update comment.
14715 (extraction_insn::field_mode): Change type to scalar_int_mode.
14716 (extraction_insn::pos_mode): Likewise.
14717 * combine.c (make_extraction): Update accordingly.
14718 * optabs-query.c (get_traditional_extraction_insn): Likewise.
14719 (get_optab_extraction_insn): Likewise.
14720 * recog.c (simplify_while_replacing): Likewise.
14721 * expmed.c (narrow_bit_field_mem): Change the type of the mode
14722 parameter to opt_scalar_int_mode.
14724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14725 Alan Hayward <alan.hayward@arm.com>
14726 David Sherwood <david.sherwood@arm.com>
14728 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
14729 to a scalar_int_mode instead of a machine_mode.
14730 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
14731 (get_best_mode): Return a boolean and use a pointer argument to store
14732 the selected mode. Replace the limit mode parameter with a bit limit.
14733 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
14734 for the values returned by bit_field_mode_iterator::next_mode.
14735 (store_bit_field): Update call to get_best_mode.
14736 (store_fixed_bit_field): Likewise.
14737 (extract_fixed_bit_field): Likewise.
14738 * expr.c (optimize_bitfield_assignment_op): Likewise.
14739 * fold-const.c (optimize_bit_field_compare): Likewise.
14740 (fold_truth_andor_1): Likewise.
14741 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
14742 Update for new type of m_mode.
14743 (get_best_mode): As above.
14745 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14746 Alan Hayward <alan.hayward@arm.com>
14747 David Sherwood <david.sherwood@arm.com>
14749 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
14750 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
14751 (store_bit_field): Check is_a <scalar_int_mode> before calling
14752 strict_volatile_bitfield_p.
14753 (extract_bit_field): Likewise.
14755 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14756 Alan Hayward <alan.hayward@arm.com>
14757 David Sherwood <david.sherwood@arm.com>
14759 * target.def (cstore_mode): Return a scalar_int_mode.
14760 * doc/tm.texi: Regenerate.
14761 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
14762 * targhooks.h (default_cstore_mode): Likewise.
14763 * targhooks.c (default_cstore_mode): Likewise, using a forced
14765 * expmed.c (emit_cstore): Expect the target of the cstore to be
14768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14769 Alan Hayward <alan.hayward@arm.com>
14770 David Sherwood <david.sherwood@arm.com>
14772 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
14774 (niter_desc): Likewise mode.
14775 (iv_analyze): Add a mode parameter.
14777 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
14778 and change its type to scalar_int_mode.
14779 * loop-iv.c: Update commentary at head of file.
14780 (iv_constant): Pass the mode paraeter before the rtx it describes
14781 and change its type to scalar_int_mode. Remove VOIDmode handling.
14782 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
14783 (iv_extend): Likewise.
14784 (shorten_into_mode): Likewise.
14785 (iv_add): Use scalar_int_mode.
14786 (iv_mult): Likewise.
14787 (iv_shift): Likewise.
14788 (canonicalize_iv_subregs): Likewise.
14789 (get_biv_step_1): Pass the outer_mode parameter before the rtx
14790 it describes and change its mode to scalar_int_mode. Also change
14791 the type of the returned inner_mode to scalar_int_mode.
14792 (get_biv_step): Likewise, turning outer_mode from a pointer
14793 into a direct parameter. Update call to get_biv_step_1.
14794 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
14795 iv_constant and get_biv_step.
14796 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
14797 and change its type to scalar_int_mode. Don't initialise iv->mode
14798 to VOIDmode and remove later checks for its still being VOIDmode.
14799 Update calls to iv_analyze_op and iv_analyze_expr. Check
14800 is_a <scalar_int_mode> when changing the mode under consideration.
14801 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
14802 Update call to iv_analyze_expr.
14803 (iv_analyze_op): Add a mode parameter. Reject subregs whose
14804 inner register is not also a scalar_int_mode. Update call to
14806 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
14807 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
14808 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
14809 separate mode class checks. Update calls to iv_analyze. Remove
14810 fix-up of VOIDmodes after iv_analyze_biv.
14811 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
14812 don't have a scalar_int_mode. Update call to biv_p.
14814 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14815 Alan Hayward <alan.hayward@arm.com>
14816 David Sherwood <david.sherwood@arm.com>
14818 * cfgexpand.c (convert_debug_memory_address): Use
14819 as_a <scalar_int_mode>.
14820 * combine.c (expand_compound_operation): Likewise.
14821 (make_extraction): Likewise.
14822 (change_zero_ext): Likewise.
14823 (simplify_comparison): Likewise.
14824 * cse.c (cse_insn): Likewise.
14825 * dwarf2out.c (minmax_loc_descriptor): Likewise.
14826 (mem_loc_descriptor): Likewise.
14827 (loc_descriptor): Likewise.
14828 * expmed.c (init_expmed_one_mode): Likewise.
14829 (synth_mult): Likewise.
14830 (emit_store_flag_1): Likewise.
14831 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
14832 of a comparison with size.
14833 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
14834 (reduce_to_bit_field_precision): Likewise.
14835 * function.c (expand_function_end): Likewise.
14836 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
14837 * loop-doloop.c (doloop_modify): Likewise.
14838 * optabs.c (expand_binop): Likewise.
14839 (expand_unop): Likewise.
14840 (expand_copysign_absneg): Likewise.
14841 (prepare_cmp_insn): Likewise.
14842 (maybe_legitimize_operand): Likewise.
14843 * recog.c (const_scalar_int_operand): Likewise.
14844 * rtlanal.c (get_address_mode): Likewise.
14845 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14846 (simplify_cond_clz_ctz): Likewise.
14847 * tree-nested.c (get_nl_goto_field): Likewise.
14848 * tree.c (build_vector_type_for_mode): Likewise.
14849 * var-tracking.c (use_narrower_mode): Likewise.
14851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14852 Alan Hayward <alan.hayward@arm.com>
14853 David Sherwood <david.sherwood@arm.com>
14855 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
14856 * builtins.c (expand_builtin_signbit): Use it.
14857 * cfgexpand.c (expand_debug_expr): Likewise.
14858 * dojump.c (do_jump): Likewise.
14859 (do_compare_and_jump): Likewise.
14860 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
14861 * expmed.c (make_tree): Likewise.
14862 * expr.c (expand_expr_real_2): Likewise.
14863 (expand_expr_real_1): Likewise.
14864 (try_casesi): Likewise.
14865 * fold-const-call.c (fold_const_call_ss): Likewise.
14866 * fold-const.c (unextend): Likewise.
14867 (extract_muldiv_1): Likewise.
14868 (fold_single_bit_test): Likewise.
14869 (native_encode_int): Likewise.
14870 (native_encode_string): Likewise.
14871 (native_interpret_int): Likewise.
14872 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
14873 * internal-fn.c (expand_addsub_overflow): Likewise.
14874 (expand_neg_overflow): Likewise.
14875 (expand_mul_overflow): Likewise.
14876 (expand_arith_overflow): Likewise.
14877 * match.pd: Likewise.
14878 * stor-layout.c (layout_type): Likewise.
14879 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14880 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
14881 * tree-ssanames.c (get_range_info): Likewise.
14882 * tree-switch-conversion.c (array_value_type) Likewise.
14883 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
14884 (vect_recog_divmod_pattern): Likewise.
14885 (vect_recog_mixed_size_cond_pattern): Likewise.
14886 * tree-vrp.c (extract_range_basic): Likewise.
14887 (simplify_float_conversion_using_ranges): Likewise.
14888 * tree.c (int_fits_type_p): Likewise.
14889 * ubsan.c (instrument_bool_enum_load): Likewise.
14890 * varasm.c (mergeable_string_section): Likewise.
14891 (narrowing_initializer_constant_valid_p): Likewise.
14892 (output_constant): Likewise.
14894 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14895 Alan Hayward <alan.hayward@arm.com>
14896 David Sherwood <david.sherwood@arm.com>
14898 * machmode.h (NARROWEST_INT_MODE): New macro.
14899 * expr.c (alignment_for_piecewise_move): Use it instead of
14900 GET_CLASS_NARROWEST_MODE (MODE_INT).
14901 (push_block): Likewise.
14902 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
14904 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
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 * postreload.c (move2add_valid_value_p): Change the type of the
14911 mode parameter to scalar_int_mode.
14912 (move2add_use_add2_insn): Add a mode parameter and use it instead
14914 (move2add_use_add3_insn): Likewise.
14915 (reload_cse_move2add): Update accordingly.
14917 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14918 Alan Hayward <alan.hayward@arm.com>
14919 David Sherwood <david.sherwood@arm.com>
14921 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
14923 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
14924 * optabs.c (expand_unop): Likewise.
14926 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14927 Alan Hayward <alan.hayward@arm.com>
14928 David Sherwood <david.sherwood@arm.com>
14930 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
14931 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
14932 (popcount_loc_descriptor): Likewise.
14933 (bswap_loc_descriptor): Likewise.
14934 (rotate_loc_descriptor): Likewise.
14935 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
14936 calling the functions above.
14938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14939 Alan Hayward <alan.hayward@arm.com>
14940 David Sherwood <david.sherwood@arm.com>
14942 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
14944 (try_combine): Likewise.
14945 (simplify_if_then_else): Likewise.
14946 * cse.c (cse_insn): Likewise.
14947 * dwarf2out.c (mem_loc_descriptor): Likewise.
14948 * emit-rtl.c (gen_lowpart_common): Likewise.
14949 * simplify-rtx.c (simplify_truncation): Likewise.
14950 (simplify_binary_operation_1): Likewise.
14951 (simplify_const_relational_operation): Likewise.
14952 (simplify_ternary_operation): Likewise.
14953 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
14955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14956 Alan Hayward <alan.hayward@arm.com>
14957 David Sherwood <david.sherwood@arm.com>
14959 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
14960 * reload.c (push_reload): Likewise.
14961 (find_reloads): Likewise.
14963 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14964 Alan Hayward <alan.hayward@arm.com>
14965 David Sherwood <david.sherwood@arm.com>
14967 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
14968 (make_compound_operation_int): Likewise.
14969 (change_zero_ext): Likewise.
14970 * expr.c (convert_move): Likewise.
14971 (convert_modes): Likewise.
14972 * fwprop.c (forward_propagate_subreg): Likewise.
14973 * loop-iv.c (get_biv_step_1): Likewise.
14974 * optabs.c (widen_operand): Likewise.
14975 * postreload.c (move2add_valid_value_p): Likewise.
14976 * recog.c (simplify_while_replacing): Likewise.
14977 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14978 (simplify_binary_operation_1): Likewise. Remove redundant
14979 mode equality check.
14981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14982 Alan Hayward <alan.hayward@arm.com>
14983 David Sherwood <david.sherwood@arm.com>
14985 * combine.c (combine_simplify_rtx): Add checks for
14986 is_a <scalar_int_mode>.
14987 (simplify_if_then_else): Likewise.
14988 (make_field_assignment): Likewise.
14989 (simplify_comparison): Likewise.
14990 * ifcvt.c (noce_try_bitop): Likewise.
14991 * loop-invariant.c (canonicalize_address_mult): Likewise.
14992 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14995 Alan Hayward <alan.hayward@arm.com>
14996 David Sherwood <david.sherwood@arm.com>
14998 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
14999 is_a <scalar_int_mode> instead of != BLKmode.
15001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15002 Alan Hayward <alan.hayward@arm.com>
15003 David Sherwood <david.sherwood@arm.com>
15005 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
15006 instead of != VOIDmode.
15007 * combine.c (if_then_else_cond): Likewise.
15008 (change_zero_ext): Likewise.
15009 * dwarf2out.c (mem_loc_descriptor): Likewise.
15010 (loc_descriptor): Likewise.
15011 * rtlanal.c (canonicalize_condition): Likewise.
15012 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
15014 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15015 Alan Hayward <alan.hayward@arm.com>
15016 David Sherwood <david.sherwood@arm.com>
15018 * simplify-rtx.c (simplify_binary_operation_1): Use
15019 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
15021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15022 Alan Hayward <alan.hayward@arm.com>
15023 David Sherwood <david.sherwood@arm.com>
15025 * wide-int.h (int_traits<unsigned char>) New class.
15026 (int_traits<unsigned short>) Likewise.
15027 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
15028 Use GET_MODE_UNIT_PRECISION and remove redundant test for
15030 * combine.c (set_nonzero_bits_and_sign_copies): Use
15031 is_a <scalar_int_mode>.
15032 (find_split_point): Likewise.
15033 (combine_simplify_rtx): Likewise.
15034 (simplify_logical): Likewise.
15035 (expand_compound_operation): Likewise.
15036 (expand_field_assignment): Likewise.
15037 (make_compound_operation): Likewise.
15038 (extended_count): Likewise.
15039 (change_zero_ext): Likewise.
15040 (simplify_comparison): Likewise.
15041 * dwarf2out.c (scompare_loc_descriptor): Likewise.
15042 (ucompare_loc_descriptor): Likewise.
15043 (minmax_loc_descriptor): Likewise.
15044 (mem_loc_descriptor): Likewise.
15045 (loc_descriptor): Likewise.
15046 * expmed.c (init_expmed_one_mode): Likewise.
15047 * lra-constraints.c (lra_constraint_offset): Likewise.
15048 * optabs.c (prepare_libcall_arg): Likewise.
15049 * postreload.c (move2add_note_store): Likewise.
15050 * reload.c (operands_match_p): Likewise.
15051 * rtl.h (load_extend_op): Likewise.
15052 * rtlhooks.c (gen_lowpart_general): Likewise.
15053 * simplify-rtx.c (simplify_truncation): Likewise.
15054 (simplify_unary_operation_1): Likewise.
15055 (simplify_binary_operation_1): Likewise.
15056 (simplify_const_binary_operation): Likewise.
15057 (simplify_const_relational_operation): Likewise.
15058 (simplify_subreg): Likewise.
15059 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15060 * var-tracking.c (adjust_mems): Likewise.
15061 (prepare_call_arguments): Likewise.
15063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15064 Alan Hayward <alan.hayward@arm.com>
15065 David Sherwood <david.sherwood@arm.com>
15067 * machmode.h (is_int_mode): New fuction.
15068 * combine.c (find_split_point): Use it.
15069 (combine_simplify_rtx): Likewise.
15070 (simplify_if_then_else): Likewise.
15071 (simplify_set): Likewise.
15072 (simplify_shift_const_1): Likewise.
15073 (simplify_comparison): Likewise.
15074 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15075 * cse.c (notreg_cost): Likewise.
15076 (cse_insn): Likewise.
15077 * cselib.c (cselib_lookup_1): Likewise.
15078 * dojump.c (do_jump_1): Likewise.
15079 (do_compare_rtx_and_jump): Likewise.
15080 * dse.c (get_call_args): Likewise.
15081 * dwarf2out.c (rtl_for_decl_init): Likewise.
15082 (native_encode_initializer): Likewise.
15083 * expmed.c (emit_store_flag_1): Likewise.
15084 (emit_store_flag): Likewise.
15085 * expr.c (convert_modes): Likewise.
15086 (store_field): Likewise.
15087 (expand_expr_real_1): Likewise.
15088 * fold-const.c (fold_read_from_constant_string): Likewise.
15089 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15090 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15091 * optabs.c (expand_binop): Likewise.
15092 (expand_unop): Likewise.
15093 (expand_abs_nojump): Likewise.
15094 (expand_one_cmpl_abs_nojump): Likewise.
15095 * simplify-rtx.c (mode_signbit_p): Likewise.
15096 (val_signbit_p): Likewise.
15097 (val_signbit_known_set_p): Likewise.
15098 (val_signbit_known_clear_p): Likewise.
15099 (simplify_relational_operation_1): Likewise.
15100 * tree.c (vector_type_mode): Likewise.
15102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15103 Alan Hayward <alan.hayward@arm.com>
15104 David Sherwood <david.sherwood@arm.com>
15106 * machmode.h (smallest_mode_for_size): Fix formatting.
15107 (smallest_int_mode_for_size): New function.
15108 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15109 instead of smallest_mode_for_size.
15110 * combine.c (make_extraction): Likewise.
15111 * config/arc/arc.c (arc_expand_movmem): Likewise.
15112 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15113 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15114 * config/s390/s390.c (s390_expand_insv): Likewise.
15115 * config/sparc/sparc.c (assign_int_registers): Likewise.
15116 * config/spu/spu.c (spu_function_value): Likewise.
15117 (spu_function_arg): Likewise.
15118 * coverage.c (get_gcov_type): Likewise.
15119 (get_gcov_unsigned_t): Likewise.
15120 * dse.c (find_shift_sequence): Likewise.
15121 * expmed.c (store_bit_field_1): Likewise.
15122 * expr.c (convert_move): Likewise.
15123 (store_field): Likewise.
15124 * internal-fn.c (expand_arith_overflow): Likewise.
15125 * optabs-query.c (get_best_extraction_insn): Likewise.
15126 * optabs.c (expand_twoval_binop_libfunc): Likewise.
15127 * stor-layout.c (layout_type): Likewise.
15128 (initialize_sizetypes): Likewise.
15129 * targhooks.c (default_get_mask_mode): Likewise.
15130 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
15132 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15133 Alan Hayward <alan.hayward@arm.com>
15134 David Sherwood <david.sherwood@arm.com>
15136 * machmode.h (opt_mode::else_blk): New function.
15137 (int_mode_for_mode): Declare.
15138 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
15139 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
15141 * cfgexpand.c (expand_debug_expr): Likewise.
15142 * combine.c (gen_lowpart_or_truncate): Likewise.
15143 (gen_lowpart_for_combine): Likewise.
15144 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
15145 * config/avr/avr.c (avr_to_int_mode): Likewise.
15146 (avr_out_plus_1): Likewise.
15147 (avr_out_plus): Likewise.
15148 (avr_out_round): Likewise.
15149 * config/i386/i386.c (ix86_split_to_parts): Likewise.
15150 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15151 (s390_expand_vcond): Likewise.
15152 * config/spu/spu.c (spu_split_immediate): Likewise.
15153 (spu_expand_mov): Likewise.
15154 * dse.c (get_stored_val): Likewise.
15155 * expmed.c (store_bit_field_1): Likewise.
15156 (convert_extracted_bit_field): Use int_mode_for_mode instead of
15158 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
15159 (extract_low_bits): Likewise.
15160 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
15161 handling rather than repeating the check.
15162 (emit_group_store): Likewise.
15163 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
15164 * optabs.c (expand_absneg_bit): Likewise.
15165 (expand_copysign_absneg): Likewise.
15166 (expand_copysign_bit): Likewise.
15167 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
15168 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
15169 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
15170 * var-tracking.c (prepare_call_arguments): Likewise.
15171 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
15172 int_mode_for_mode instead of mode_for_size.
15173 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15175 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15176 Alan Hayward <alan.hayward@arm.com>
15177 David Sherwood <david.sherwood@arm.com>
15179 * machmode.h (int_mode_for_size): New function.
15180 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
15181 instead of mode_for_size.
15182 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
15184 * combine.c (expand_field_assignment): Use int_mode_for_size
15185 instead of mode_for_size.
15186 (make_extraction): Likewise.
15187 (simplify_shift_const_1): Likewise.
15188 (simplify_comparison): Likewise.
15189 * dojump.c (do_jump): Likewise.
15190 * dwarf2out.c (mem_loc_descriptor): Likewise.
15191 * emit-rtl.c (init_derived_machine_modes): Likewise.
15192 * expmed.c (flip_storage_order): Likewise.
15193 (convert_extracted_bit_field): Likewise.
15194 * expr.c (copy_blkmode_from_reg): Likewise.
15195 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
15196 * internal-fn.c (expand_mul_overflow): Likewise.
15197 * lower-subreg.c (simple_move): Likewise.
15198 * optabs-libfuncs.c (init_optabs): Likewise.
15199 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15200 * tree.c (vector_type_mode): Likewise.
15201 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
15202 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
15203 * tree-vect-generic.c (expand_vector_parallel): Likewise.
15204 * tree-vect-stmts.c (vectorizable_load): Likewise.
15205 (vectorizable_store): Likewise.
15207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15208 Alan Hayward <alan.hayward@arm.com>
15209 David Sherwood <david.sherwood@arm.com>
15211 * coretypes.h (pod_mode): New type.
15212 (scalar_int_mode_pod): New typedef.
15213 * machmode.h (pod_mode): New class.
15214 (int_n_data_t::m): Change type to scalar_int_mode_pod.
15215 * genmodes.c (emit_mode_int_n): Update accordingly.
15216 * lower-subreg.h (target_lower_subreg): Change type to
15217 scalar_int_mode_pod.
15218 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
15219 scalar_int_mode_pod.
15221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15222 Alan Hayward <alan.hayward@arm.com>
15223 David Sherwood <david.sherwood@arm.com>
15225 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
15226 machine_mode to scalar_int_mode.
15227 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
15228 (rs6000_option_override_internal): Remove cast to int.
15229 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
15230 machine_mode to scalar_int_mode.
15231 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
15232 (rs6000_option_override_internal): Remove cast to int.
15233 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
15234 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
15236 * config/s390/s390.c (s390_expand_builtin): Likewise.
15237 * coretypes.h (scalar_int_mode): New type.
15238 (opt_scalar_int_mode): New typedef.
15239 * machmode.h (scalar_int_mode): New class.
15240 (scalar_int_mode::includes_p): New function.
15241 (byte_mode): Change type to scalar_int_mode.
15242 (word_mode): Likewise.
15243 (ptr_mode): Likewise.
15244 * emit-rtl.c (byte_mode): Likewise.
15245 (word_mode): Likewise.
15246 (ptr_mode): Likewise.
15247 (init_derived_machine_modes): Update accordingly.
15248 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
15249 and MODE_PARTIAL_INT.
15250 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
15251 opt_scalar_int_mode.
15253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15254 Alan Hayward <alan.hayward@arm.com>
15255 David Sherwood <david.sherwood@arm.com>
15257 * target.def (libgcc_floating_mode_supported_p): Take a
15259 * doc/tm.texi: Regenerate.
15260 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
15262 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15263 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
15266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15267 Alan Hayward <alan.hayward@arm.com>
15268 David Sherwood <david.sherwood@arm.com>
15270 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
15271 * doc/tm.texi: Regenerate.
15272 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
15273 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
15274 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
15275 * targhooks.h (default_floatn_mode): Likewise.
15276 * targhooks.c (default_floatn_mode): Likewise.
15277 * tree.c (build_common_tree_nodes): Update accordingly.
15279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15280 Alan Hayward <alan.hayward@arm.com>
15281 David Sherwood <david.sherwood@arm.com>
15283 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
15284 (mode_iterator::iterate_p): Likewise.
15285 (mode_iterator::get_wider): Likewise.
15286 * expr.c (init_expr_target): Use 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 * coretypes.h (opt_scalar_float_mode): New typedef.
15293 * machmode.h (float_mode_for_size): New function.
15294 * emit-rtl.c (double_mode): Delete.
15295 (init_emit_once): Use float_mode_for_size.
15296 * stor-layout.c (layout_type): Likewise.
15297 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
15299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15300 Alan Hayward <alan.hayward@arm.com>
15301 David Sherwood <david.sherwood@arm.com>
15303 * output.h (assemble_real): Take a scalar_float_mode.
15304 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
15305 * config/arm/arm.md (consttable_4): Likewise.
15306 (consttable_8): Likewise.
15307 (consttable_16): Likewise.
15308 * config/mips/mips.md (consttable_float): Likewise.
15309 * config/s390/s390.c (s390_output_pool_entry): Likewise.
15310 * varasm.c (assemble_real): Take a scalar_float_mode.
15311 (output_constant_pool_2): Update accordingly.
15312 (output_constant): Likewise.
15314 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15315 Alan Hayward <alan.hayward@arm.com>
15316 David Sherwood <david.sherwood@arm.com>
15318 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
15319 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
15320 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
15321 (native_encode_real): Likewise.
15322 (native_interpret_real): Likewise.
15323 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
15324 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15326 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15327 Alan Hayward <alan.hayward@arm.com>
15328 David Sherwood <david.sherwood@arm.com>
15330 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
15331 <scalar_float_mode>. Simplify.
15332 (gen_extend_conv_libfunc): Likewise.
15334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15335 Alan Hayward <alan.hayward@arm.com>
15336 David Sherwood <david.sherwood@arm.com>
15338 * coretypes.h (scalar_float_mode): New type.
15339 * machmode.h (mode_traits::from_int): Use machine_mode if
15340 USE_ENUM_MODES is defined.
15341 (is_a): New function.
15343 (dyn_cast): Likewise.
15344 (scalar_float_mode): New class.
15345 (scalar_float_mode::includes_p): New function.
15346 (is_float_mode): Likewise.
15347 * gdbhooks.py (MachineModePrinter): New class.
15348 (build_pretty_printer): Use it for scalar_float_mode.
15349 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
15350 (format_helper::format_helper): Turn into a template.
15351 * genmodes.c (get_mode_class): New function.
15352 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
15353 or machine_mode if none.
15354 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
15355 as_a <scalar_float_mode>.
15356 * dwarf2out.c (mem_loc_descriptor): Likewise.
15357 (insert_float): Likewise.
15358 (add_const_value_attribute): Likewise.
15359 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15360 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
15361 (expand_unop): Update accordingly.
15362 (expand_abs_nojump): Likewise.
15363 (expand_copysign_absneg): Take a scalar_float_mode.
15364 (expand_copysign_bit): Likewise.
15365 (expand_copysign): Update accordingly.
15367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15368 Alan Hayward <alan.hayward@arm.com>
15369 David Sherwood <david.sherwood@arm.com>
15371 * coretypes.h (opt_mode): New class.
15372 * machmode.h (opt_mode): Likewise.
15373 (opt_mode::else_void): New function.
15374 (opt_mode::require): Likewise.
15375 (opt_mode::exists): Likewise.
15376 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
15377 (GET_MODE_2XWIDER_MODE): Likewise.
15378 (mode_iterator::get_wider): Update accordingly.
15379 (mode_iterator::get_2xwider): Likewise.
15380 (mode_iterator::get_known_wider): Likewise, turning into a template.
15381 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
15382 forcing a wider mode to exist.
15383 * config/cr16/cr16.h (LONG_REG_P): Likewise.
15384 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15385 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
15386 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
15387 * lower-subreg.c (init_lower_subreg): Likewise.
15388 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
15389 on the final iteration.
15390 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
15391 a wider mode exists before asking for a move pattern.
15392 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
15393 forcing a wider mode to exist.
15394 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
15395 returning false if no such mode exists.
15396 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
15397 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
15398 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
15399 Avoid checking for a MODE_INT if we already know the mode is not a
15401 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
15402 forcing a wider mode to exist.
15403 (expmed_mult_highpart_optab): Likewise.
15404 (expmed_mult_highpart): Likewise.
15405 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
15407 * lto-streamer-in.c (lto_input_mode_table): Likewise.
15408 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
15409 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15410 * internal-fn.c (expand_mul_overflow): Update use of
15411 GET_MODE_2XWIDER_MODE.
15412 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15413 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
15414 GET_MODE_WIDER_MODE.
15415 (convert_plusminus_to_widen): Likewise.
15416 * tree-switch-conversion.c (array_value_type): Likewise.
15417 * var-tracking.c (emit_note_insn_var_location): Likewise.
15418 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15419 Return false inside rather than outside the loop if no wider mode
15421 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
15422 and GET_MODE_2XWIDER_MODE
15423 (can_compare_p): Use else_void.
15424 * gdbhooks.py (OptMachineModePrinter): New class.
15425 (build_pretty_printer): Use it for opt_mode.
15427 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15428 Alan Hayward <alan.hayward@arm.com>
15429 David Sherwood <david.sherwood@arm.com>
15431 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
15432 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
15434 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15435 Alan Hayward <alan.hayward@arm.com>
15436 David Sherwood <david.sherwood@arm.com>
15438 * machmode.h (mode_traits): New structure.
15439 (get_narrowest_mode): New function.
15440 (mode_iterator::start): Likewise.
15441 (mode_iterator::iterate_p): Likewise.
15442 (mode_iterator::get_wider): Likewise.
15443 (mode_iterator::get_known_wider): Likewise.
15444 (mode_iterator::get_2xwider): Likewise.
15445 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
15446 (FOR_EACH_MODE): Likewise.
15447 (FOR_EACH_MODE_FROM): Likewise.
15448 (FOR_EACH_MODE_UNTIL): Likewise.
15449 (FOR_EACH_WIDER_MODE): Likewise.
15450 (FOR_EACH_2XWIDER_MODE): Likewise.
15451 * builtins.c (expand_builtin_strlen): Use new mode iterators.
15452 * combine.c (simplify_comparison): Likewise
15453 * config/i386/i386.c (type_natural_mode): Likewise.
15454 * cse.c (cse_insn): Likewise.
15455 * dse.c (find_shift_sequence): Likewise.
15456 * emit-rtl.c (init_derived_machine_modes): Likewise.
15457 (init_emit_once): Likewise.
15458 * explow.c (hard_function_value): Likewise.
15459 * expmed.c (extract_fixed_bit_field_1): Likewise.
15460 (extract_bit_field_1): Likewise.
15461 (expand_divmod): Likewise.
15462 (emit_store_flag_1): Likewise.
15463 * expr.c (init_expr_target): Likewise.
15464 (convert_move): Likewise.
15465 (alignment_for_piecewise_move): Likewise.
15466 (widest_int_mode_for_size): Likewise.
15467 (emit_block_move_via_movmem): Likewise.
15468 (copy_blkmode_to_reg): Likewise.
15469 (set_storage_via_setmem): Likewise.
15470 (compress_float_constant): Likewise.
15471 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15472 * optabs-query.c (get_best_extraction_insn): Likewise.
15473 * optabs.c (expand_binop): Likewise.
15474 (expand_twoval_unop): Likewise.
15475 (expand_twoval_binop): Likewise.
15476 (widen_leading): Likewise.
15477 (widen_bswap): Likewise.
15478 (expand_parity): Likewise.
15479 (expand_unop): Likewise.
15480 (prepare_cmp_insn): Likewise.
15481 (prepare_float_lib_cmp): Likewise.
15482 (expand_float): Likewise.
15483 (expand_fix): Likewise.
15484 (expand_sfix_optab): Likewise.
15485 * postreload.c (move2add_use_add2_insn): Likewise.
15486 * reg-stack.c (reg_to_stack): Likewise.
15487 * reginfo.c (choose_hard_reg_mode): Likewise.
15488 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15489 * stor-layout.c (mode_for_size): Likewise.
15490 (smallest_mode_for_size): Likewise.
15491 (mode_for_vector): Likewise.
15492 (finish_bitfield_representative): Likewise.
15493 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
15494 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
15495 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15496 * var-tracking.c (prepare_call_arguments): Likewise.
15498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15499 Alan Hayward <alan.hayward@arm.com>
15500 David Sherwood <david.sherwood@arm.com>
15502 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
15503 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
15504 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
15505 * machmode.h (mode_size): Move earlier in file.
15506 (mode_precision): Likewise.
15507 (mode_inner): Likewise.
15508 (mode_nunits): Likewise.
15509 (mode_unit_size): Likewise.
15510 (unit_unit_precision): Likewise.
15511 (mode_wider): Likewise.
15512 (mode_2xwider): Likewise.
15513 (machine_mode): New class.
15514 (mode_to_bytes): New function.
15515 (mode_to_bits): Likewise.
15516 (mode_to_precision): Likewise.
15517 (mode_to_inner): Likewise.
15518 (mode_to_unit_size): Likewise.
15519 (mode_to_unit_precision): Likewise.
15520 (mode_to_nunits): Likewise.
15521 (GET_MODE_SIZE): Use mode_to_bytes.
15522 (GET_MODE_BITSIZE): Use mode_to_bits.
15523 (GET_MODE_PRECISION): Use mode_to_precision.
15524 (GET_MODE_INNER): Use mode_to_inner.
15525 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
15526 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
15527 (GET_MODE_NUNITS): Use mode_to_nunits.
15528 * system.h (ALWAYS_INLINE): New macro.
15529 * config/powerpcspe/powerpcspe-c.c
15530 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
15531 int for arg1_mode and arg2_mode.
15533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15534 Alan Hayward <alan.hayward@arm.com>
15535 David Sherwood <david.sherwood@arm.com>
15537 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15538 Prefix mode names with E_ in case statements.
15539 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15540 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
15541 (aarch64_split_simd_move): Likewise.
15542 (aarch64_gen_storewb_pair): Likewise.
15543 (aarch64_gen_loadwb_pair): Likewise.
15544 (aarch64_gen_store_pair): Likewise.
15545 (aarch64_gen_load_pair): Likewise.
15546 (aarch64_get_condition_code_1): Likewise.
15547 (aarch64_constant_pool_reload_icode): Likewise.
15548 (get_rsqrte_type): Likewise.
15549 (get_rsqrts_type): Likewise.
15550 (get_recpe_type): Likewise.
15551 (get_recps_type): Likewise.
15552 (aarch64_gimplify_va_arg_expr): Likewise.
15553 (aarch64_simd_container_mode): Likewise.
15554 (aarch64_emit_load_exclusive): Likewise.
15555 (aarch64_emit_store_exclusive): Likewise.
15556 (aarch64_expand_compare_and_swap): Likewise.
15557 (aarch64_gen_atomic_cas): Likewise.
15558 (aarch64_emit_bic): Likewise.
15559 (aarch64_emit_atomic_swap): Likewise.
15560 (aarch64_emit_atomic_load_op): Likewise.
15561 (aarch64_evpc_trn): Likewise.
15562 (aarch64_evpc_uzp): Likewise.
15563 (aarch64_evpc_zip): Likewise.
15564 (aarch64_evpc_ext): Likewise.
15565 (aarch64_evpc_rev): Likewise.
15566 (aarch64_evpc_dup): Likewise.
15567 (aarch64_gen_ccmp_first): Likewise.
15568 (aarch64_gen_ccmp_next): Likewise.
15569 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15570 (alpha_emit_xfloating_libcall): Likewise.
15571 (emit_insxl): Likewise.
15572 (alpha_arg_type): Likewise.
15573 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
15574 (arc_preferred_simd_mode): Likewise.
15575 (arc_secondary_reload): Likewise.
15576 (get_arc_condition_code): Likewise.
15577 (arc_print_operand): Likewise.
15578 (arc_legitimate_constant_p): Likewise.
15579 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15580 * config/arc/arc.md (casesi_load): Likewise.
15581 (casesi_compact_jump): Likewise.
15582 * config/arc/predicates.md (proper_comparison_operator): Likewise.
15583 (cc_use_register): Likewise.
15584 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15585 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
15586 (arm_init_iwmmxt_builtins): Likewise.
15587 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
15588 (neon_expand_vector_init): Likewise.
15589 (arm_attr_length_move_neon): Likewise.
15590 (maybe_get_arm_condition_code): Likewise.
15591 (arm_emit_vector_const): Likewise.
15592 (arm_preferred_simd_mode): Likewise.
15593 (arm_output_iwmmxt_tinsr): Likewise.
15594 (thumb1_output_casesi): Likewise.
15595 (thumb2_output_casesi): Likewise.
15596 (arm_emit_load_exclusive): Likewise.
15597 (arm_emit_store_exclusive): Likewise.
15598 (arm_expand_compare_and_swap): Likewise.
15599 (arm_evpc_neon_vuzp): Likewise.
15600 (arm_evpc_neon_vzip): Likewise.
15601 (arm_evpc_neon_vrev): Likewise.
15602 (arm_evpc_neon_vtrn): Likewise.
15603 (arm_evpc_neon_vext): Likewise.
15604 (arm_validize_comparison): Likewise.
15605 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
15606 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
15607 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
15608 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
15609 (c6x_preferred_simd_mode): Likewise.
15610 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
15611 (epiphany_rtx_costs): Likewise.
15612 * config/epiphany/predicates.md (proper_comparison_operator):
15614 * config/frv/frv.c (condexec_memory_operand): Likewise.
15615 (frv_emit_move): Likewise.
15616 (output_move_single): Likewise.
15617 (output_condmove_single): Likewise.
15618 (frv_hard_regno_mode_ok): Likewise.
15619 (frv_matching_accg_mode): Likewise.
15620 * config/h8300/h8300.c (split_adds_subs): Likewise.
15621 (h8300_rtx_costs): Likewise.
15622 (h8300_print_operand): Likewise.
15623 (compute_mov_length): Likewise.
15624 (output_logical_op): Likewise.
15625 (compute_logical_op_length): Likewise.
15626 (compute_logical_op_cc): Likewise.
15627 (h8300_shift_needs_scratch_p): Likewise.
15628 (output_a_shift): Likewise.
15629 (compute_a_shift_length): Likewise.
15630 (compute_a_shift_cc): Likewise.
15631 (expand_a_rotate): Likewise.
15632 (output_a_rotate): Likewise.
15633 * config/i386/i386.c (classify_argument): Likewise.
15634 (function_arg_advance_32): Likewise.
15635 (function_arg_32): Likewise.
15636 (function_arg_64): Likewise.
15637 (function_value_64): Likewise.
15638 (ix86_gimplify_va_arg): Likewise.
15639 (ix86_legitimate_constant_p): Likewise.
15640 (put_condition_code): Likewise.
15641 (split_double_mode): Likewise.
15642 (ix86_avx256_split_vector_move_misalign): Likewise.
15643 (ix86_expand_vector_logical_operator): Likewise.
15644 (ix86_split_idivmod): Likewise.
15645 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
15646 (ix86_build_const_vector): Likewise.
15647 (ix86_build_signbit_mask): Likewise.
15648 (ix86_match_ccmode): Likewise.
15649 (ix86_cc_modes_compatible): Likewise.
15650 (ix86_expand_branch): Likewise.
15651 (ix86_expand_sse_cmp): Likewise.
15652 (ix86_expand_sse_movcc): Likewise.
15653 (ix86_expand_int_sse_cmp): Likewise.
15654 (ix86_expand_vec_perm_vpermi2): Likewise.
15655 (ix86_expand_vec_perm): Likewise.
15656 (ix86_expand_sse_unpack): Likewise.
15657 (ix86_expand_int_addcc): Likewise.
15658 (ix86_split_to_parts): Likewise.
15659 (ix86_vectorize_builtin_gather): Likewise.
15660 (ix86_vectorize_builtin_scatter): Likewise.
15661 (avx_vpermilp_parallel): Likewise.
15662 (inline_memory_move_cost): Likewise.
15663 (ix86_tieable_integer_mode_p): Likewise.
15664 (x86_maybe_negate_const_int): Likewise.
15665 (ix86_expand_vector_init_duplicate): Likewise.
15666 (ix86_expand_vector_init_one_nonzero): Likewise.
15667 (ix86_expand_vector_init_one_var): Likewise.
15668 (ix86_expand_vector_init_concat): Likewise.
15669 (ix86_expand_vector_init_interleave): Likewise.
15670 (ix86_expand_vector_init_general): Likewise.
15671 (ix86_expand_vector_set): Likewise.
15672 (ix86_expand_vector_extract): Likewise.
15673 (emit_reduc_half): Likewise.
15674 (ix86_emit_i387_round): Likewise.
15675 (ix86_mangle_type): Likewise.
15676 (ix86_expand_round_sse4): Likewise.
15677 (expand_vec_perm_blend): Likewise.
15678 (canonicalize_vector_int_perm): Likewise.
15679 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
15680 (expand_vec_perm_1): Likewise.
15681 (expand_vec_perm_interleave3): Likewise.
15682 (expand_vec_perm_even_odd_pack): Likewise.
15683 (expand_vec_perm_even_odd_1): Likewise.
15684 (expand_vec_perm_broadcast_1): Likewise.
15685 (ix86_vectorize_vec_perm_const_ok): Likewise.
15686 (ix86_expand_vecop_qihi): Likewise.
15687 (ix86_expand_mul_widen_hilo): Likewise.
15688 (ix86_expand_sse2_abs): Likewise.
15689 (ix86_expand_pextr): Likewise.
15690 (ix86_expand_pinsr): Likewise.
15691 (ix86_preferred_simd_mode): Likewise.
15692 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
15693 * config/i386/sse.md (*andnot<mode>3): Likewise.
15694 (<mask_codefor><code><mode>3<mask_name>): Likewise.
15695 (*<code><mode>3): Likewise.
15696 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
15697 (ia64_expand_atomic_op): Likewise.
15698 (ia64_arg_type): Likewise.
15699 (ia64_mode_to_int): Likewise.
15700 (ia64_scalar_mode_supported_p): Likewise.
15701 (ia64_vector_mode_supported_p): Likewise.
15702 (expand_vec_perm_broadcast): Likewise.
15703 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
15704 (iq2000_function_arg_advance): Likewise.
15705 (iq2000_function_arg): Likewise.
15706 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
15707 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
15708 (m68k_libcall_value): Likewise.
15709 (m68k_function_value): Likewise.
15710 (sched_attr_op_type): Likewise.
15711 * config/mcore/mcore.c (mcore_output_move): Likewise.
15712 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
15714 (microblaze_function_arg): Likewise.
15715 * config/mips/mips.c (mips16_build_call_stub): Likewise.
15716 (mips_print_operand): Likewise.
15717 (mips_mode_ok_for_mov_fmt_p): Likewise.
15718 (mips_vector_mode_supported_p): Likewise.
15719 (mips_preferred_simd_mode): Likewise.
15720 (mips_expand_vpc_loongson_even_odd): Likewise.
15721 (mips_expand_vec_unpack): Likewise.
15722 (mips_expand_vi_broadcast): Likewise.
15723 (mips_expand_vector_init): Likewise.
15724 (mips_expand_vec_reduc): Likewise.
15725 (mips_expand_msa_cmp): Likewise.
15726 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
15727 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
15728 (cc_flags_for_mode): Likewise.
15729 * config/msp430/msp430.c (msp430_print_operand): Likewise.
15730 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
15731 (nds32_output_casesi_pc_relative): Likewise.
15732 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15733 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
15734 (nvptx_gen_unpack): Likewise.
15735 (nvptx_gen_pack): Likewise.
15736 (nvptx_gen_shuffle): Likewise.
15737 (nvptx_gen_wcast): Likewise.
15738 (nvptx_preferred_simd_mode): Likewise.
15739 * config/pa/pa.c (pa_secondary_reload): Likewise.
15740 * config/pa/predicates.md (base14_operand): Likewise.
15741 * config/powerpcspe/powerpcspe-c.c
15742 (altivec_resolve_overloaded_builtin): Likewise.
15743 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
15745 (rs6000_preferred_simd_mode): Likewise.
15746 (output_vec_const_move): Likewise.
15747 (rs6000_expand_vector_extract): Likewise.
15748 (rs6000_split_vec_extract_var): Likewise.
15749 (reg_offset_addressing_ok_p): Likewise.
15750 (rs6000_legitimate_offset_address_p): Likewise.
15751 (rs6000_legitimize_address): Likewise.
15752 (rs6000_emit_set_const): Likewise.
15753 (rs6000_const_vec): Likewise.
15754 (rs6000_emit_move): Likewise.
15755 (spe_build_register_parallel): Likewise.
15756 (rs6000_darwin64_record_arg_recurse): Likewise.
15757 (swap_selector_for_mode): Likewise.
15758 (spe_init_builtins): Likewise.
15759 (paired_init_builtins): Likewise.
15760 (altivec_init_builtins): Likewise.
15761 (do_load_for_compare): Likewise.
15762 (rs6000_generate_compare): Likewise.
15763 (rs6000_expand_float128_convert): Likewise.
15764 (emit_load_locked): Likewise.
15765 (emit_store_conditional): Likewise.
15766 (rs6000_output_function_epilogue): Likewise.
15767 (rs6000_handle_altivec_attribute): Likewise.
15768 (rs6000_function_value): Likewise.
15769 (emit_fusion_gpr_load): Likewise.
15770 (emit_fusion_p9_load): Likewise.
15771 (emit_fusion_p9_store): Likewise.
15772 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
15773 (fusion_gpr_mem_load): Likewise.
15774 (fusion_addis_mem_combo_load): Likewise.
15775 (fusion_addis_mem_combo_store): Likewise.
15776 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
15777 (fusion_gpr_mem_load): Likewise.
15778 (fusion_addis_mem_combo_load): Likewise.
15779 (fusion_addis_mem_combo_store): Likewise.
15780 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15782 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
15783 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15784 (rs6000_preferred_simd_mode): Likewise.
15785 (output_vec_const_move): Likewise.
15786 (rs6000_expand_vector_extract): Likewise.
15787 (rs6000_split_vec_extract_var): Likewise.
15788 (reg_offset_addressing_ok_p): Likewise.
15789 (rs6000_legitimate_offset_address_p): Likewise.
15790 (rs6000_legitimize_address): Likewise.
15791 (rs6000_emit_set_const): Likewise.
15792 (rs6000_const_vec): Likewise.
15793 (rs6000_emit_move): Likewise.
15794 (rs6000_darwin64_record_arg_recurse): Likewise.
15795 (swap_selector_for_mode): Likewise.
15796 (paired_init_builtins): Likewise.
15797 (altivec_init_builtins): Likewise.
15798 (rs6000_expand_float128_convert): Likewise.
15799 (emit_load_locked): Likewise.
15800 (emit_store_conditional): Likewise.
15801 (rs6000_output_function_epilogue): Likewise.
15802 (rs6000_handle_altivec_attribute): Likewise.
15803 (rs6000_function_value): Likewise.
15804 (emit_fusion_gpr_load): Likewise.
15805 (emit_fusion_p9_load): Likewise.
15806 (emit_fusion_p9_store): Likewise.
15807 * config/rx/rx.c (rx_gen_move_template): Likewise.
15808 (flags_from_mode): Likewise.
15809 * config/s390/predicates.md (s390_alc_comparison): Likewise.
15810 (s390_slb_comparison): Likewise.
15811 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
15812 (s390_vector_mode_supported_p): Likewise.
15813 (s390_cc_modes_compatible): Likewise.
15814 (s390_match_ccmode_set): Likewise.
15815 (s390_canonicalize_comparison): Likewise.
15816 (s390_emit_compare_and_swap): Likewise.
15817 (s390_branch_condition_mask): Likewise.
15818 (s390_rtx_costs): Likewise.
15819 (s390_secondary_reload): Likewise.
15820 (__SECONDARY_RELOAD_CASE): Likewise.
15821 (s390_expand_cs): Likewise.
15822 (s390_preferred_simd_mode): Likewise.
15823 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
15824 * config/sh/sh.c (sh_print_operand): Likewise.
15825 (dump_table): Likewise.
15826 (sh_secondary_reload): Likewise.
15827 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15828 * config/sh/sh.md (casesi_worker_1): Likewise.
15829 (casesi_worker_2): Likewise.
15830 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
15831 (fcc_comparison_operator): Likewise.
15832 * config/sparc/sparc.c (sparc_expand_move): Likewise.
15833 (emit_soft_tfmode_cvt): Likewise.
15834 (sparc_preferred_simd_mode): Likewise.
15835 (output_cbranch): Likewise.
15836 (sparc_print_operand): Likewise.
15837 (sparc_expand_vec_perm_bmask): Likewise.
15838 (vector_init_bshuffle): Likewise.
15839 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15840 (spu_vector_mode_supported_p): Likewise.
15841 (spu_expand_insv): Likewise.
15842 (spu_emit_branch_or_set): Likewise.
15843 (spu_handle_vector_attribute): Likewise.
15844 (spu_builtin_splats): Likewise.
15845 (spu_builtin_extract): Likewise.
15846 (spu_builtin_promote): Likewise.
15847 (spu_expand_sign_extend): Likewise.
15848 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15849 (tilegx_simd_int): Likewise.
15850 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
15851 (tilepro_simd_int): Likewise.
15852 * config/v850/v850.c (const_double_split): Likewise.
15853 (v850_print_operand): Likewise.
15854 (ep_memory_offset): Likewise.
15855 * config/vax/vax.c (vax_rtx_costs): Likewise.
15856 (vax_output_int_move): Likewise.
15857 (vax_output_int_add): Likewise.
15858 (vax_output_int_subtract): Likewise.
15859 * config/visium/predicates.md (visium_branch_operator): Likewise.
15860 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
15861 (visium_print_operand_address): Likewise.
15862 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15863 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
15864 (xtensa_expand_conditional_branch): Likewise.
15865 (xtensa_copy_incoming_a7): Likewise.
15866 (xtensa_output_literal): Likewise.
15867 * dfp.c (decimal_real_maxval): Likewise.
15868 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15871 Alan Hayward <alan.hayward@arm.com>
15872 David Sherwood <david.sherwood@arm.com>
15874 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
15875 (mode_nunits_inline): Likewise.
15876 (mode_inner_inline): Likewise.
15877 (mode_unit_size_inline): Likewise.
15878 (mode_unit_precision_inline): Likewise.
15879 (emit_insn_modes_h): Likewise. Also emit a #define of the
15881 (emit_mode_wider): Add an E_ prefix to mode names.
15882 (emit_mode_complex): Likewise.
15883 (emit_mode_inner): Likewise.
15884 (emit_mode_adjustments): Likewise.
15885 (emit_mode_int_n): Likewise.
15886 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
15887 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
15888 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
15889 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
15890 (CRC32_BUILTIN, ENTRY): Likewise.
15891 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
15892 (aarch64_pop_regs): Likewise.
15893 (aarch64_process_components): Likewise.
15894 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
15895 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
15896 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
15897 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
15898 * config/arm/arm.c (arm_init_libfuncs): Likewise.
15899 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
15901 * config/i386/i386-builtin.def (pcmpestr): Likewise.
15902 (pcmpistr): Likewise.
15903 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
15904 * config/mmix/mmix.c (mmix_output_condition): Likewise.
15905 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15907 * config/rl78/rl78.c (mduc_regs): Likewise.
15908 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15909 (htm_expand_builtin): Likewise.
15910 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
15911 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
15912 * config/xtensa/xtensa.c (print_operand): Likewise.
15913 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
15914 (NUM_MODE_VECTOR_INT): Likewise.
15915 * genoutput.c (null_operand): Likewise.
15916 (output_operand_data): Likewise.
15917 * genrecog.c (print_parameter_value): Likewise.
15918 * lra.c (debug_operand_data): Likewise.
15920 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15922 * dwarf2out.c (scompare_loc_descriptor_wide)
15923 (scompare_loc_descriptor_narrow): New functions, split out from...
15924 (scompare_loc_descriptor): ...here.
15925 * expmed.c (emit_store_flag_int): New function, split out from...
15926 (emit_store_flag): ...here.
15928 2017-08-30 Richard Biener <rguenther@suse.de>
15930 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
15931 (dwarf2out_early_finish): Move setting of AT_pubnames from
15932 early debug output to early finish.
15934 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15936 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
15937 and -mdata-region to the assembler.
15939 2017-08-30 Richard Biener <rguenther@suse.de>
15941 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
15943 (gen_subprogram_die): Add DW_AT_object_pointer only early.
15944 (dwarf2out_early_global_decl): Only generate a DIE for the
15945 abstract origin if it doesn't already exist or is a declaration DIE.
15946 (resolve_addr): Do not add the linkage name twice when
15947 generating a stub DIE for the DW_TAG_GNU_call_site target.
15949 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15951 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15952 Use machine_mode rather than int for arg1_mode.
15954 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
15957 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
15958 that the second argument of the built-in functions to unpack
15959 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
15960 switch statement instead a lot of if statements.
15961 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
15962 Allow 64-bit values to be in Altivec registers as well as
15963 traditional floating point registers.
15964 (pack<mode>, FMOVE128_VSX iterator): Likewise.
15966 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
15968 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
15969 MAX_REGS_PER_ADDRESS == 1.
15971 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
15973 * config/i386/i386.opt (flag_fentry): Do not init to -1.
15974 * config/i386/i386.c (ix86_option_override_internal): Simplify
15975 setting of opts->x_flag_entry.
15977 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15978 Jakub Jelinek <jakub@redhat.com>
15979 Richard Biener <rguenther@suse.de>
15981 PR tree-optimization/81503
15982 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
15983 folded constant fits in the target type; reorder tests for clarity.
15985 2017-08-29 Martin Liska <mliska@suse.cz>
15987 * passes.def: Include pass_lower_switch.
15988 * stmt.c (dump_case_nodes): Remove and move to
15989 tree-switch-conversion.
15990 (case_values_threshold): Likewise.
15991 (expand_switch_as_decision_tree_p): Likewise.
15992 (emit_case_decision_tree): Likewise.
15993 (expand_case): Likewise.
15994 (balance_case_nodes): Likewise.
15995 (node_has_low_bound): Likewise.
15996 (node_has_high_bound): Likewise.
15997 (node_is_bounded): Likewise.
15998 (emit_case_nodes): Likewise.
15999 (struct simple_case_node): New struct.
16000 (add_case_node): Remove.
16001 (emit_case_dispatch_table): Use vector instead of case_list.
16002 (reset_out_edges_aux): Remove.
16003 (compute_cases_per_edge): Likewise.
16004 (expand_case): Build list of simple_case_node.
16005 (expand_sjlj_dispatch_table): Use it.
16006 * tree-switch-conversion.c (struct case_node): Moved from
16007 stmt.c and adjusted.
16008 (emit_case_nodes): Likewise.
16009 (node_has_low_bound): Likewise.
16010 (node_has_high_bound): Likewise.
16011 (node_is_bounded): Likewise.
16012 (case_values_threshold): Likewise.
16013 (reset_out_edges_aux): Likewise.
16014 (compute_cases_per_edge): Likewise.
16015 (add_case_node): Likewise.
16016 (dump_case_nodes): Likewise.
16017 (balance_case_nodes): Likewise.
16018 (expand_switch_as_decision_tree_p): Likewise.
16019 (emit_jump): Likewise.
16020 (emit_case_decision_tree): Likewise.
16021 (try_switch_expansion): Likewise.
16022 (do_jump_if_equal): Likewise.
16023 (emit_cmp_and_jump_insns): Likewise.
16024 (fix_phi_operands_for_edge): New function.
16025 (record_phi_operand_mapping): Likewise.
16026 (class pass_lower_switch): New pass.
16027 (pass_lower_switch::execute): New function.
16028 (make_pass_lower_switch): Likewise.
16029 (conditional_probability):
16030 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
16031 * tree-pass.h: Add make_pass_lower_switch.
16033 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16036 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
16039 2017-08-29 Richard Biener <rguenther@suse.de>
16041 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
16042 we do not add a DW_AT_inline attribute twice.
16043 (gen_subprogram_die): Remove code setting DW_AT_inline on
16044 DECL_ABSTRACT_P nodes.
16046 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
16048 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
16049 calls to internal functions.
16050 (gimplify_modify_expr): Likewise.
16051 * tree-call-cdce.c (use_internal_fn): Likewise.
16052 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
16053 (convert_to_divmod): Set the nothrow flag.
16054 * tree-if-conv.c (predicate_mem_writes): Likewise.
16055 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16056 (vectorizable_call): Likewise.
16057 (vectorizable_store): Likewise.
16058 (vectorizable_load): Likewise.
16059 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16060 (vect_recog_mask_conversion_pattern): Likewise.
16062 2017-08-29 Martin Liska <mliska@suse.cz>
16065 * gcc.c (driver_handle_option): Add new argument.
16066 * opts-common.c (handle_option): Pass
16067 target_option_override_hook.
16068 * opts-global.c (lang_handle_option): Add new option.
16069 (set_default_handlers): Add new argument.
16070 (decode_options): Likewise.
16071 * opts.c (target_handle_option): Likewise.
16072 (common_handle_option): Call target_option_override_hook.
16073 * opts.h (struct cl_option_handler_func): Add hook for
16074 target option override.
16075 (struct cl_option_handlers): Likewise.
16076 (set_default_handlers): Add new argument.
16077 (decode_options): Likewise.
16078 (common_handle_option): Likewise.
16079 (target_handle_option): Likewise.
16080 * toplev.c (toplev::main): Pass targetm.target_option.override
16083 2017-08-29 Richard Biener <rguenther@suse.de>
16084 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16086 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16087 life to the active subtree.
16089 2017-08-28 Jeff Law <law@redhat.com>
16091 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16092 derive_equivalences.
16093 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16094 Code moved into....
16095 (edge_info::derive_equivalences): New private member function
16097 * tree-ssa-dom.c (class edge_info): Changed from a struct
16098 to a class. Add ctor/dtor, methods and data members.
16099 (edge_info::edge_info): Renamed from allocate_edge_info.
16100 Initialize additional members.
16101 (edge_info::~edge_info): New.
16102 (free_dom_edge_info): Delete the edge info.
16103 (record_edge_info): Use new class & associated member functions.
16104 Tighten forms for testing for edge equivalences.
16105 (record_temporary_equivalences): Iterate over the simple
16106 equivalences rather than assuming there's only one per edge.
16107 (cprop_into_successor_phis): Iterate over the simple
16108 equivalences rather than assuming there's only one per edge.
16109 (optimize_stmt): Use operand_equal_p rather than pointer
16110 equality for mini-DSE code.
16112 2017-08-28 Nathan Sidwell <nathan@acm.org>
16114 * gcc.c (execute): Fold SIGPIPE handling into switch
16115 statement. Adjust internal error message.
16117 2017-08-28 Richard Biener <rguenther@suse.de>
16120 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
16121 Do nothing for removed DIEs.
16123 2017-08-28 Richard Biener <rguenther@suse.de>
16125 PR tree-optimization/81977
16126 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
16129 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
16132 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
16133 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
16134 using targetm.gen_mem_thread_fence.
16136 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
16139 * config/i386/i386.md (*<btsc><mode>): Change operand 2
16140 predicate to register_operand. Reorder operands.
16141 (*btr<mode>): Ditto.
16142 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
16143 (*btr<mode>_mask): Ditto.
16145 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
16147 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
16148 * config/rs6000/xmmintrin.h: New file.
16149 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
16151 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16154 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
16155 parameter and_insn and return it.
16156 (recombine_lvx_pattern): Insert a copy to ensure availability of
16157 the base register of the copied masking operation at the point of
16158 the instruction replacement.
16159 (recombine_stvx_pattern): Likewise.
16161 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16163 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
16164 undocumented switches.
16165 (-mpower9-dform-vector): Likewise.
16166 (-mpower9-dform): Likewise.
16167 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
16168 comments to delete references to -mpower9-dform* switches.
16169 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
16170 Delete reference to -mpower9-dform* switches, test for
16171 -mpower9-vector instead.
16172 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
16173 (OTHER_P9_VECTOR_MASKS): Likewise.
16174 (POWERPC_MASKS): Likewise.
16175 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
16176 tests against -mpower9-dform* to -mpower9-vector. Delete code
16177 that checked for -mpower9-dform* consistancy with other options.
16178 Add test for -mpower9-misc to enable other power9 switches.
16179 (rs6000_init_hard_regno_mode_ok): Likewise.
16180 (rs6000_option_override_internal): Likewise.
16181 (rs6000_emit_prologue): Likewise.
16182 (rs6000_emit_epilogue): Likewise.
16183 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
16184 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
16185 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
16187 (emit_fusion_p9_store): Likewise.
16188 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
16189 resetting these macros if the assembler does not support ISA 3.0
16191 (TARGET_P9_DFORM_VECTOR): Likewise.
16192 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
16193 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
16195 2017-08-25 Alan Modra <amodra@gmail.com>
16197 PR rtl-optimization/81747
16198 * cse.c (cse_extended_basic_block): Don't attempt to record
16199 equivalences for degenerate conditional branches that branch
16200 to their fall-through.
16202 2017-08-24 Martin Sebor <msebor@redhat.com>
16204 PR middle-end/81908
16205 * gimple-fold.c (size_must_be_zero_p): New function.
16206 (gimple_fold_builtin_memory_op): Call it.
16208 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
16210 * config/rs6000/mm_malloc.h: New file.
16212 2017-08-24 Bin Cheng <bin.cheng@arm.com>
16214 PR tree-optimization/81913
16215 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
16216 analysis when either IVs in condition can wrap.
16218 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
16220 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
16221 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
16223 2017-08-24 Richard Biener <rguenther@suse.de>
16226 * targhooks.c (default_target_can_inline_p): Properly
16227 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
16228 is present and always compare.
16229 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
16230 infer -mfpmath=sse from TARGET_SSE_P.
16231 (ix86_can_inline_p): Properly use target_option_default_node when
16232 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
16234 2017-08-24 Richard Biener <rguenther@suse.de>
16237 * dwarf2out.c (output_die): Handle flag_generate_offload like
16239 (output_comp_unit): Likewise.
16240 (gen_array_type_die): Likewise.
16241 (dwarf2out_early_finish): Likewise.
16242 (note_variable_value_in_expr): Likewise.
16243 (dwarf2out_finish): Likewise. Adjust assert.
16244 * cgraphunit.c (symbol_table::compile): Move setting of
16245 flag_generate_offload earlier ...
16246 (symbol_table::finalize_compilation_unit): ... here, before
16247 early debug finalization.
16249 2017-08-24 Richard Biener <rguenther@suse.de>
16251 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
16252 and ipa-fnsummary.h.
16253 (ix86_can_inline_p): When ix86_fpmath flags do not match
16254 check whether the callee uses FP math at all.
16256 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
16258 PR middle-end/81931
16259 * tree-ssanames.c (get_nonzero_bits): Use element_precision
16260 instead of TYPE_PRECISION.
16262 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
16263 Alan Hayward <alan.hayward@arm.com>
16264 David Sherwood <david.sherwood@arm.com>
16266 * combine.c (make_extraction): Use subreg_offset_from_lsb.
16268 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
16270 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
16272 (ix86_frame::stack_realign_allocate): New field.
16273 (struct machine_frame_state): Modify comments.
16274 (machine_frame_state::sp_realigned_fp_end): New field.
16275 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
16276 layout calculation.
16277 (sp_valid_at): Add assertion to assure no attempt to access invalid
16278 offset of a realigned stack.
16279 (fp_valid_at): Likewise.
16280 (choose_baseaddr): Modify comments.
16281 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
16282 ix86_expand_prologue.
16283 (ix86_expand_prologue): Modify stack realignment and allocation.
16284 (ix86_expand_epilogue): Modify comments.
16285 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
16286 avx2_runtime, avx512f, and avx512f_runtime.
16288 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
16290 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
16291 (mstackrealign): Do not init to -1.
16292 * config/i386/i386.c (ix86_option_override_internal):
16293 Check opts_set, not opts when setting default value of
16294 opts->x_ix86_force_align_arg_pointer.
16296 2017-08-23 Richard Biener <rguenther@suse.de>
16298 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
16299 lang_hooks.decl_printable_name.
16300 * print-rtl-function.c (print_rtx_function): Likewise.
16301 * tree-pretty-print.c (dump_function_header): Likewise.
16303 2017-08-23 Richard Biener <rguenther@suse.de>
16306 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
16307 -g0 at compile-time.
16309 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16311 PR middle-end/19706
16312 * doc/sourcebuild.texi (Other hardware attributes):
16315 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16317 PR middle-end/19706
16318 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
16319 Fix single-use check.
16321 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
16323 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
16325 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16327 * doc/install.texi: Modify to add more details on running selected
16330 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16332 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
16333 is combined with -mabi=ms.
16334 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
16337 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16339 PR tree-optimization/81488
16340 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
16341 and cached_basis fields.
16342 (MAX_SPREAD): New constant.
16343 (alloc_cand_and_find_basis): Initialize new fields.
16344 (clear_visited): New function.
16345 (create_phi_basis_1): Rename from create_phi_basis, set visited
16346 and cached_basis fields.
16347 (create_phi_basis): New wrapper function.
16348 (phi_add_costs_1): Rename from phi_add_costs, add spread
16349 parameter, set visited field, short-circuit when limits reached.
16350 (phi_add_costs): New wrapper function.
16351 (record_phi_increments_1): Rename from record_phi_increments, set
16353 (record_phi_increments): New wrapper function.
16354 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
16355 (phi_incr_cost): New wrapper function.
16356 (all_phi_incrs_profitable_1): Rename from
16357 all_phi_incrs_profitable, set visited field.
16358 (all_phi_incrs_profitable): New wrapper function.
16360 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16361 Alan Hayward <alan.hayward@arm.com>
16362 David Sherwood <david.sherwood@arm.com>
16364 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
16365 that takes the outer and inner modes.
16366 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
16367 comparison as the canonical test for a paradoxical subreg.
16368 * combine.c (simplify_set): Use paradoxical_subreg_p.
16369 (make_extraction): Likewise.
16370 (force_to_mode): Likewise.
16371 (rtx_equal_for_field_assignment_p): Likewise.
16372 (gen_lowpart_for_combine): Likewise.
16373 (simplify_comparison): Likewise.
16374 * cse.c (equiv_constant): Likewise.
16375 * expmed.c (store_bit_field_1): Likewise.
16376 * final.c (alter_subreg): Likewise.
16377 * fwprop.c (propagate_rtx): Likewise.
16378 (forward_propagate_subreg): Likewise.
16379 * ira-conflicts.c (ira_build_conflicts): Likewise.
16380 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
16381 * lra-constraints.c (curr_insn_transform): Likewise.
16382 (split_reg): Likewise.
16383 * lra-eliminations.c (move_plus_up): Likewise.
16384 (lra_eliminate_regs_1): Likewise.
16385 * recog.c (general_operand): Likewise.
16386 * ree.c (combine_reaching_defs): Likewise.
16387 * reload.c (push_reload): Likewise.
16388 (find_reloads): Likewise.
16389 * reload1.c (elimination_effects): Likewise.
16390 (compute_reload_subreg_offset): Likewise.
16391 (choose_reload_regs): Likewise.
16392 * rtlanal.c (subreg_lsb_1): Likewise.
16393 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16394 (simplify_subreg): Likewise.
16395 * var-tracking.c (track_loc_p): Likewise.
16396 * emit-rtl.c (byte_lowpart_offset): Likewise.
16397 (paradoxical_subreg_p): Delete out-of-line definition.
16399 2017-08-22 Jeff Law <law@redhat.com>
16401 PR tree-optimization/81741
16402 PR tree-optimization/71947
16403 * tree-ssa-dom.c: Include tree-inline.h.
16404 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
16405 equivalences if one is more expensive to compute than the other.
16406 * tree-ssa-scopedtables.h (class const_or_copies): Make
16407 record_const_or_copy_raw method private.
16408 (class avail_exprs_stack): New method simplify_binary_operation.
16409 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
16410 avail_exprs_stack::simplify_binary_operation as needed.
16411 (avail_exprs_stack::simplify_binary_operation): New function.
16413 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
16415 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
16416 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
16417 (DOT_SYMBOLS): Likewise.
16418 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
16419 (RELOCATABLE_NEEDS_FIXUP): Likewise.
16420 (RS6000_ABI_NAME): Likewise.
16421 (TARGET_CMODEL): Likewise.
16422 (TOC_SECTION_ASM_OP): Likewise.
16423 (SET_CMODEL): New macro.
16424 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
16426 2017-08-22 Richard Biener <rguenther@suse.de>
16428 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
16429 to allow for free-lang-data replacements similar to verify_type_variant.
16431 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16432 Alan Hayward <alan.hayward@arm.com>
16433 David Sherwood <david.sherwood@arm.com>
16435 * config/aarch64/aarch64.md (casesi): Use DImode rather than
16436 VOIDmode for the LABEL_REF.
16438 2017-08-22 Richard Biener <rguenther@suse.de>
16440 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
16442 2017-08-22 Richard Biener <rguenther@suse.de>
16444 * common.opt (feliminate-dwarf2-dups): Ignore.
16445 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
16446 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
16447 same_die_p_wrap, compute_section_prefix,
16448 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
16449 (comdat_symbol_id, comdat_symbol_number): Likewise.
16450 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
16452 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
16453 (output_die): Mark unreachable path unreachable.
16454 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
16455 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
16456 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
16457 (dwarf2out_early_finish): Likewise.
16459 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
16461 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
16463 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
16466 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
16467 not VAR_P. Filter attribute warnings with OPT_Wattributes.
16468 (avr_attribute_table) <io, io_low, address>: Initialize
16469 .decl_required with true.
16471 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
16473 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
16474 undocumented debugging options.
16475 (-mvsx-scalar-double): Likewise.
16476 (-mallow-df-permute): Likewise.
16477 (-mvectorize-builtins): Likewise.
16478 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16479 (rs6000_builtin_vectorized_function): Likewise.
16480 (rs6000_builtin_md_vectorized_function): Likewise.
16481 (rs6000_opt_vars): Likewise.
16483 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
16486 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
16487 (*btrq_imm): Rename from *btrq.
16488 (*btcq_imm): Rename from *btcq.
16489 (btsc): New code attribute.
16490 (*<btsc><mode>): New insn pattern.
16491 (*btr<mode>): Ditto.
16492 (*<btsc><mode>_mask): New insn_and_split pattern.
16493 (*btr<mode>_mask): Ditto.
16495 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16496 Alan Hayward <alan.hayward@arm.com>
16497 David Sherwood <david.sherwood@arm.com>
16499 * function.c (pad_below): Simplify padding calculation.
16501 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16502 Alan Hayward <alan.hayward@arm.com>
16503 David Sherwood <david.sherwood@arm.com>
16505 * target.def (function_prologue): Remove frame size argument.
16506 (function_epilogue): Likewise.
16507 * doc/tm.texi: Regenerate.
16508 * final.c (final_start_function): Update call to function_prologue.
16509 (final_end_function): Update call to function_epilogue.
16510 (default_function_pro_epilogue): Remove frame size argument.
16511 * output.h (default_function_pro_epilogue): Likewise.
16512 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
16513 (arm_output_function_prologue): Likewise.
16514 * config/frv/frv.c (frv_function_prologue): Likewise.
16515 (frv_function_epilogue): Likewise.
16516 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
16517 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
16518 (ia64_output_function_epilogue): Likewise.
16519 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
16520 (m32r_output_function_epilogue): Likewise.
16521 * config/microblaze/microblaze.c (microblaze_function_prologue)
16522 (microblaze_function_epilogue): Likewise.
16523 * config/mips/mips.c (mips_output_function_prologue): Likewise.
16524 (mips_output_function_epilogue): Likewise.
16525 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
16526 (mmix_target_asm_function_epilogue): Likewise.
16527 * config/msp430/msp430.c (msp430_start_function): Likewise.
16528 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
16529 (nds32_asm_function_epilogue): Likewise.
16530 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
16531 * config/pa/pa.c (pa_output_function_prologue): Likewise.
16532 (pa_output_function_epilogue): Likewise.
16533 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
16534 (rs6000_output_function_epilogue): Likewise.
16535 * config/rl78/rl78.c (rl78_start_function): Likewise.
16536 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
16537 (rs6000_output_function_epilogue): Likewise.
16538 * config/rx/rx.c (rx_output_function_prologue): Likewise.
16539 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
16540 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
16541 (sparc_asm_function_epilogue): Likewise.
16543 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16545 * tree.h (type_has_mode_precision_p): New function.
16546 * convert.c (convert_to_integer_1): Use it.
16547 * expr.c (expand_expr_real_2): Likewise.
16548 (expand_expr_real_1): Likewise.
16549 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
16550 * match.pd: Likewise.
16551 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
16552 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
16553 * tree-tailcall.c (process_assignment): Likewise.
16554 * tree-vect-loop.c (vectorizable_reduction): Likewise.
16555 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
16556 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
16557 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16558 (vectorizable_assignment): Likewise.
16559 (vectorizable_shift): Likewise.
16560 (vectorizable_operation): Likewise.
16561 * tree-vrp.c (register_edge_assert_for_2): Likewise.
16563 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
16565 * match.pd: Add pow (C, x) simplification.
16567 2017-08-21 Richard Biener <rguenther@suse.de>
16569 PR tree-optimization/81900
16570 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
16571 for blocks with abnormal predecessors.
16572 (compute_antic): Do not set visited flag prematurely.
16574 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
16577 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
16579 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16581 * stor-layout.h (vector_type_mode): Move to...
16582 * tree.h (vector_type_mode): ...here.
16583 * stor-layout.c (vector_type_mode): Move to...
16584 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
16586 2017-08-21 Richard Biener <rguenther@suse.de>
16588 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
16589 register_external_die hooks.
16590 (debug_false_tree_charstarstar_uhwistar): Declare.
16591 (debug_nothing_tree_charstar_uhwi): Likewise.
16592 * debug.c (do_nothing_debug_hooks): Adjust.
16593 (debug_false_tree_charstarstar_uhwistar): New do nothing.
16594 (debug_nothing_tree_charstar_uhwi): Likewise.
16595 * dbxout.c (dbx_debug_hooks): Adjust.
16596 (xcoff_debug_hooks): Likewise.
16597 * sdbout.c (sdb_debug_hooks): Likewise.
16598 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16599 * dwarf2out.c (macinfo_label_base): New global.
16600 (dwarf2out_register_external_die): New function for the
16601 register_external_die hook.
16602 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
16603 (dwarf2_debug_hooks): Use them.
16604 (dwarf2_lineno_debug_hooks): Adjust.
16605 (struct die_struct): Add with_offset flag.
16606 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
16607 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
16608 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
16609 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
16610 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
16611 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
16612 defining section names for the early LTO debug variants.
16613 (reset_indirect_string): New helper.
16614 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
16615 (print_dw_val): Add support for offsetted symbol references.
16616 (get_ultimate_context): Split out from is_cxx.
16617 (is_cxx): Use get_ultimate_context.
16618 (is_fortran): Add decl overload.
16619 (compute_comp_unit_symbol): Split out worker from
16620 compute_section_prefix.
16621 (compute_section_prefix): Call compute_comp_unit_symbol and
16622 set comdat_type_p here.
16623 (output_die): Skip DIE symbol output for the LTO added one.
16624 Handle DIE symbol references with offset.
16625 (output_comp_unit): Guard section name mangling properly.
16626 For LTO debug sections emit a symbol at the section beginning
16627 which we use to refer to its DIEs.
16628 (add_abstract_origin_attribute): For DIEs registered via
16629 dwarf2out_register_external_die directly refer to the early
16630 DIE rather than indirectly through the shadow one we created.
16631 Remove obsolete call to dwarf2out_abstract_function for
16632 non-function/block origins.
16633 (gen_array_type_die): When generating early LTO debug do
16634 not emit DW_AT_string_length.
16635 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
16636 late when in LTO. As suggested place a gcc_unreachable for
16637 the DECL_ABSTRACT_P case.
16638 (gen_subprogram_die): Avoid another specification DIE
16639 for early built declarations/definitions for the late LTO case.
16640 (gen_variable_die): Add type references for late duplicated VLA dies
16642 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
16643 we have the abstract instance already.
16644 (process_scope_var): Adjust decl DIE contexts in LTO which
16645 first puts them in limbo.
16646 (gen_decl_die): Do not generate type DIEs late apart from
16647 types for VLAs or for decls we do not yet have a DIE. Do not
16648 call dwarf2out_abstract_function late.
16649 (dwarf2out_early_global_decl): Make sure to create DIEs
16650 for abstract instances of a decl first.
16651 (dwarf2out_late_global_decl): Adjust comment.
16652 (output_macinfo_op): With multiple macro sections use
16653 macinfo_label_base to distinguish labels.
16654 (output_macinfo): Likewise. Update macinfo_label_base.
16655 Pass in the line info label.
16656 (note_variable_value_in_expr): When generating LTO resolve
16657 all variable values here by generating DIEs as needed.
16658 (init_sections_and_labels): Add early LTO debug flag parameter
16659 and generate different sections and names if set. Add generation
16660 counter for the labels so we can have multiple of them.
16661 (reset_dies): Helper to allow DIEs to be output multiple times.
16662 (dwarf2out_finish): When outputting DIEs to the fat part of an
16663 LTO object first reset DIEs.
16664 (dwarf2out_early_finish): Output early DIEs when generating LTO.
16665 (modified_type_die): Check for decl_ultimate_origin being self
16667 (gen_type_die_with_usage): Likewise.
16668 (gen_typedef_die): Allow decl_ultimate_origin being self.
16669 (set_decl_abstract_flags): Remove.
16670 (set_block_abstract_flags): Likewise.
16671 (dwarf2out_abstract_function): Treat the early generated DIEs
16672 as the abstract copy and only add DW_AT_inline and
16673 DW_AT_artificial here and call set_decl_origin_self.
16674 If the DIE has an abstract origin don't do anything.
16675 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
16676 if we have none yet (Go fails to build one, PR78628).
16677 (variably_modified_type_p): Prevent endless recursion for Ada
16678 cyclic pointer types.
16679 * lto-streamer-in.c: Include debug.h.
16680 (dref_queue): New global.
16681 (lto_read_tree_1): Stream in DIE references.
16682 (lto_input_tree): Register DIE references.
16683 (input_function): Stream DECL_DEBUG_ARGS.
16684 * lto-streamer-out.c: Include debug.h.
16685 (lto_write_tree_1): Output DIE references.
16686 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
16687 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
16688 (output_function): Stream DECL_DEBUG_ARGS.
16689 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
16690 Stream DECL_ABSTRACT_ORIGIN.
16691 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
16692 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
16693 DECL_CONTEXT for file-scope decls.
16694 * lto-streamer.h (struct dref_entry): Declare.
16695 (dref_queue): Likewise.
16696 * cfgexpand.c (pass_expand::execute): Do not call the
16697 outlining_inline_function hook here.
16698 * lto-wrapper.c (debug_obj): New global.
16699 (tool_cleanup): Unlink it if required.
16700 (debug_objcopy): New function.
16701 (run_gcc): Handle early debug sections in the IL files by
16702 extracting them to separate files, partially linkin them and
16703 feeding the result back as result to the linker.
16704 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
16705 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
16706 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
16707 sections into a separate segment.
16708 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
16710 (darwin_asm_dwarf_section): Likewise.
16711 (darwin_asm_output_dwarf_offset): Likewise.
16712 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
16714 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16715 Alan Hayward <alan.hayward@arm.com>
16716 David Sherwood <david.sherwood@arm.com>
16718 * read-md.h (md_reader::record_potential_iterator_use): Replace
16719 pointer argument with an rtx and an index.
16720 * read-rtl.c (iterator_group::apply_iterator): Likewise.
16721 (apply_mode_iterator): Likewise.
16722 (apply_code_iterator): Likewise.
16723 (apply_int_iterator): Likewise.
16724 (apply_subst_iterator): Likewise.
16725 (record_iterator_use): Likewise.
16726 (record_attribute_use): Likewise.
16727 (md_reader::record_potential_iterator_use): Likewise. Update calls
16728 to record_iterator_use and apply_iterator.
16729 (iterator_use): Replace ptr with x and index.
16730 (attribute_use): Likewise.
16731 (apply_attribute_uses): Update calls to apply_iterator.
16732 (apply_iterators): Likewise. Update initialization of iterator_use.
16733 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
16734 and record_potential_iterator_use.
16735 (rtx_reader::read_rtx_operand): Likewise.
16737 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16738 Alan Hayward <alan.hayward@arm.com>
16739 David Sherwood <david.sherwood@arm.com>
16741 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
16744 2017-08-21 Richard Biener <rguenther@suse.de>
16746 PR middle-end/81884
16747 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
16748 at struct end conservatively when comparing common bases.
16750 2017-08-21 Richard Biener <rguenther@suse.de>
16752 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
16753 (mem_ref_in_stmt): Remove.
16754 (determine_max_movement): Use ref index to get at the reference.
16755 (invariantness_dom_walker::before_dom_children): Deal with
16756 lim data already initialized.
16757 (gather_mem_refs_stmt): Initialize lim data and record ref index.
16759 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16761 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
16762 (TARGET_ISA_ROUND): Ditto.
16763 (TARGET_ROUND): Ditto.
16764 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
16765 * config/i386/i386.md: Ditto.
16766 * config/i386/sse.md: Ditto.
16767 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
16768 with OPTION_MASK_ISA_SSE4_1.
16770 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16773 * doc/extend.texi (x86 Built-in Functions): Correct the name of
16774 __builtin_ia32_lzcnt_u16.
16776 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
16779 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
16780 (rs6000_set_current_function): Rewrite function to use it.
16782 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
16785 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
16786 and TYPE_WARN_IF_NOT_ALIGN.
16787 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
16788 (handle_warn_if_not_align): New.
16789 (place_union_field): Call handle_warn_if_not_align.
16790 (place_field): Call handle_warn_if_not_align.
16791 Copy TYPE_WARN_IF_NOT_ALIGN.
16792 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
16793 (layout_type): Likewise.
16794 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
16796 (tree_decl_common): Add warn_if_not_align.
16797 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
16798 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
16799 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
16800 (DECL_WARN_IF_NOT_ALIGN): Likewise.
16801 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
16802 * doc/extend.texi: Document warn_if_not_aligned attribute.
16803 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
16805 2017-08-17 Martin Liska <mliska@suse.cz>
16808 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
16809 (get_data_dependence): Use it as pointer type.
16810 (distribute_loop): Likewise.
16812 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16814 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
16815 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
16816 (p8_vmrgow_<mode>_direct): New define_insn.
16817 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
16818 handle endianness for vmrgew and vmrgow permute patterns.
16820 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16822 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
16823 * config/rs6000/rs6000-cpus.def: Remove comment.
16824 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
16825 (POWERPC_MASKS): Likewise.
16826 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
16827 use of TARGET_VSX_TIMODE.
16828 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
16829 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
16830 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
16831 (rs6000_option_override_internal): Remove dead code.
16832 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
16833 (rs6000_legitimize_reload_address): Likewise.
16834 (rs6000_legitimate_address_p): Likewise.
16835 (rs6000_opt_masks): Delete "vsx-timode".
16836 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
16837 from function comment.
16838 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
16839 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
16840 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
16842 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
16843 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
16844 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
16845 (VSX_TI): Likewise.
16847 (define_peephole2): Likewise.
16849 2017-08-17 Martin Sebor <msebor@redhat.com>
16852 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
16853 past the end of an array.
16854 (test_pp_format): Add test cases.
16856 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
16858 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
16859 missing ECF_NOTHROW flags.
16861 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16864 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
16865 operands residing in integer registers.
16866 (*vsx_le_perm_load_<mode>): Likewise.
16867 (*vsx_le_perm_store_<mode>): Likewise.
16868 (define_peephole2): Add peepholes to optimize the above.
16870 2017-08-17 Marek Polacek <polacek@redhat.com>
16872 PR middle-end/81814
16873 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
16874 to mimic what shorten_compare did. Change the return type to bool.
16875 (fold_cond_expr_with_comparison): Update call to
16876 operand_equal_for_comparison_p.
16877 (fold_ternary_loc): Likewise.
16879 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
16881 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
16883 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
16885 2017-08-17 Richard Biener <rguenther@suse.de>
16887 * tree-ssa-structalias.c (solve_graph): When propagating
16888 to successors update the graphs succ edges and avoid duplicate work.
16890 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
16893 * config/i386/i386.c (ix86_option_override_internal): Save target
16894 specific options after ix86_stack_protector_guard_reg was changed.
16896 2017-08-17 Richard Biener <rguenther@suse.de>
16898 PR tree-optimization/81827
16899 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
16900 (new_var_info): Initialize it conservatively.
16901 (get_call_vi): Mark register vars.
16902 (new_scalar_tmp_constraint_exp): Likewise.
16903 (handle_rhs_call): Likewise.
16904 (handle_const_call): Likewise.
16905 (create_function_info_for): Likewise.
16906 (solve_constraints): Sort varinfos to separate register from
16907 non-register vars to pack points-to solution bitmaps during
16910 2017-08-17 Marek Polacek <polacek@redhat.com>
16912 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
16914 2017-08-17 Richard Biener <rguenther@suse.de>
16916 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
16917 to true when overflow is undefined and we saturated the result.
16919 2017-08-17 Alan Modra <amodra@gmail.com>
16922 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
16923 Don't use store multiple if only one reg needs saving.
16924 (interesting_frame_related_regno): New function.
16925 (rs6000_frame_related): Don't emit frame info for regs that
16927 (rs6000_emit_epilogue): Likewise.
16929 2017-08-16 Nathan Sidwell <nathan@acm.org>
16931 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
16932 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
16933 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
16934 * tree.c (free_lang_data_in_type): Use else-if chain. Always
16935 clear TYPE_LANG_1. Remove obsolete member-function stripping.
16936 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
16937 (verify_type): Adjust for TYPE_BINFO move.
16938 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
16939 process TYPE_BINFO directly.
16940 (hash_tree): Likewise.
16941 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16943 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16946 2017-08-16 David Malcolm <dmalcolm@redhat.com>
16948 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
16950 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
16953 * config/i386/i386.md (*anddi_1_btr): Change predicates of
16954 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
16955 Add ix86_binary_operator_ok to insn constraint.
16956 (*iordi_1_bts): Ditto.
16957 (*xordi_1_btc): Ditto.
16958 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
16959 Update corresponding peephole2 pattern.
16963 2017-08-16 Bin Cheng <bin.cheng@arm.com>
16965 PR tree-optimization/81832
16966 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
16967 copy loop header which has IFN_LOOP_DIST_ALIAS call.
16969 2017-08-16 Marek Polacek <polacek@redhat.com>
16972 * fold-const.c (fold_indirect_ref_1): Restore original behavior
16973 regarding size_zero_node.
16975 2017-08-16 Martin Liska <mliska@suse.cz>
16978 * config.gcc: Respect previously set extra_objs in case
16981 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
16983 PR tree-optimization/81835
16984 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
16985 the phi SSA_NAME. Check that the condition in a COND_EXPR does
16986 not depend on the phi.
16988 2017-08-16 Alan Modra <amodra@gmail.com>
16990 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
16993 2017-08-16 Alan Modra <amodra@gmail.com>
16995 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
16996 (save_reg_p): ..into this. Update all callers.
16997 (first_reg_to_save): Simplify.
16999 2017-08-16 Alan Modra <amodra@gmail.com>
17001 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
17004 2017-08-15 Joseph Myers <joseph@codesourcery.com>
17008 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
17009 constant count if that count is less than 32.
17011 2017-08-15 Nathan Sidwell <nathan@acm.org>
17013 * gcc.c (execute): Emit friendlier message if inferior is killed
17014 by an external cause.
17016 2017-08-15 Richard Biener <rguenther@suse.de>
17018 PR tree-optimization/81790
17019 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
17020 CONSTRUCTORs from simplifying and VN.
17022 2017-08-14 Martin Sebor <msebor@redhat.com>
17024 * builtin-attrs.def: Add comments.
17026 2017-08-14 Martin Sebor <msebor@redhat.com>
17029 * doc/extend.texi (attribute nonstring): Document new attribute.
17031 2017-08-14 Martin Sebor <msebor@redhat.com>
17034 * tree-diagnostic.c (default_tree_printer): Handle %G.
17035 * gimple-pretty-print.h (percent_G_format): Declare new function.
17036 * gimple-pretty-print.c (percent_G_format): Define.
17037 * tree-pretty-print.c (percent_K_format): Add argument.
17039 2017-08-14 Martin Sebor <msebor@redhat.com>
17041 PR translation/79998
17042 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
17043 Remove a stray space.
17045 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
17048 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
17049 (*iordi_1_bts): Ditto.
17050 (*xordi_1_btc): Ditto.
17052 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17055 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17056 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17058 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17059 quoted strings, and make more translator-friendly.
17060 (darwin_rs6000_override_options): Likewise.
17061 (rs6000_option_override_internal): Likewise.
17062 (rs6000_return_in_memory): Fix overlong line.
17063 (init_cmulative_args): Use quoted strings, and make more
17064 translator-friendly.
17065 (rs6000_pass_by_reference): Fix overlong line.
17066 (def_builtin): Use quoted strings.
17067 (altivec_expand_predicate_builtin): Use quoted strings, and make
17068 more translator-friendly.
17069 (htm_expand_builtin): Use quoted strings.
17070 (cpu_expand_builtin): Use quoted strings, and make more
17071 translator-friendly.
17072 (altivec_expand_builtin): Likewise.
17073 (paired_expand_predicate_builtin): Likewise.
17074 (rs6000_invalid_builtin): Likewise.
17075 (builtin_function_type): Use quoted strings.
17076 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17077 more translator-friendly.
17078 (rs6000_trampoline_init): Likewise.
17079 (rs6000_handle_altivec_attribute): Likewise.
17080 (rs6000_inner_target_options): Use quoted strings.
17081 (rs6000_disable_incompatible_switches): Likewise.
17082 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17083 strings, and make more translator-friendly.
17084 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17086 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17088 PR tree-optimization/81799
17089 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17090 cond_expr to simple gimple operand.
17092 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17094 PR middle-end/46932
17095 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17097 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17101 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17102 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17103 TARGET_GASISR_PROLOGUES.
17104 * config/avr/avr.c (avr_option_override): Same.
17105 (avr_pass_pre_proep::execute): Same.
17107 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17110 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17111 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17113 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
17115 * config/i386/i386.md (*load_tp_<mode>): Redefine as
17116 define_insn_and_split. Split to a memory load from 0 in
17117 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
17118 using PTR mode iterator.
17119 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
17120 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
17121 (*add_tp_<mode>): Redefine as define_insn_and_split.
17122 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
17123 address space. Merge with *add_tp_x32 using PTR mode iterator.
17124 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
17125 Split to an add with a memory load from 0 in
17126 DEFAULT_TLS_SEG_REG address space.
17128 2017-08-12 Andrew Pinski <apinski@cavium.com>
17130 * config/aarch64/aarch64-option-extensions.def (rdma):
17131 Fix feature string to what Linux prints out in /proc/cpuinfo.
17133 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
17136 * dwarf2out.c (modified_type_die): For C typedef types that have
17137 an ultimate origin, process the ultimate origin instead of the
17139 (gen_typedef_die): Assert that input DECLs have no ultimate
17141 (gen_type_die_with_usage): For typedef variants that have an
17142 ultimate origin, just call gen_decl_die on the original DECL.
17143 (process_scope_var): Avoid creating DIEs for local typedefs and
17144 concrete static variables.
17146 2017-08-12 Alan Modra <amodra@gmail.com>
17150 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
17151 match gnu-user.h startfile.
17152 (ENDFILE_LINUX_SPEC): Similarly.
17154 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
17157 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
17159 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
17161 2017-08-11 Tamar Christina <tamar.christina@arm.com>
17162 * config/aarch64/aarch64.md (mov<mode>): Change.
17163 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17164 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
17165 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
17167 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
17169 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
17170 for storage order barriers.
17172 2017-08-11 Martin Liska <mliska@suse.cz>
17175 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
17176 variables of void type.
17178 2017-08-11 Martin Liska <mliska@suse.cz>
17180 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
17181 TARGET_SUPPORTS_ALIASES.
17182 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
17183 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
17184 (optimize_weakref): Likewise.
17185 * symtab.c (symtab_node::noninterposable_alias): Likewise.
17186 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
17187 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
17189 2017-08-11 Martin Liska <mliska@suse.cz>
17192 * config/i386/i386.c (make_resolver_func): Do complete
17193 refactoring of the function.
17195 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17198 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
17199 * config/i386/i386.c (ix86_stack_protect_guard): Use
17200 ix86_stack_protect_guard_symbol_str to generate varible declaration.
17201 * doc/invoke.texi (x86 Options): Document
17202 -mstack-protector-guard-symbol= option.
17204 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17206 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
17207 * config/i386/i386.c (ix86_split_stack_guard): New function.
17208 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
17209 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
17210 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
17211 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
17212 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
17213 (split_stack_space_check): Call ix86_split_stack_guard.
17215 2017-08-10 Martin Sebor <msebor@redhat.com>
17217 * print-tree.c (print_node): Print location using the established
17219 Replace spaces with colons.
17220 (debug_raw, debug): Ditto.
17222 2017-08-10 Martin Sebor <msebor@redhat.com>
17225 * pretty-print.c (pp_format): Correct the handling of %s precision.
17227 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
17230 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17232 (ix86_finalize_stack_frame_flags): This. Also clear
17233 frame_pointer_needed if -fno-omit-frame-pointer is used without
17235 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17236 with ix86_finalize_stack_frame_flags.
17237 (ix86_expand_epilogue): Likewise.
17238 (ix86_expand_split_stack_prologue): Likewise.
17239 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
17241 2017-08-10 Martin Liska <mliska@suse.cz>
17244 * c-attribs.c (handle_target_attribute):
17245 Report warning for an empty string argument of target attribute.
17247 2017-08-09 Jakub Jelinek <jakub@redhat.com>
17250 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
17252 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
17253 or DECL_NONLOCAL labels.
17254 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
17255 or DECL_NONLOCAL labels here.
17257 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
17259 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
17260 to indicate when early gimple folding has been disabled.
17261 (rs6000_gimple_fold_builtin): Add debug content.
17262 (rs6000_invalid_builtin): Fix whitespace.
17263 (rs6000_expand_builtin): Fix whitespace.
17264 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
17266 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
17269 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
17270 SAVE_MULTIPLE if not all the registers that saves, should be saved.
17272 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
17274 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
17275 (qdf24xx): Likewise.
17276 * config/aarch64/aarch64.md: Include falkor.md.
17277 * config/aarch64/falkor.md: New.
17279 2017-08-09 Marek Polacek <polacek@redhat.com>
17282 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
17283 * diagnostic.c (emit_diagnostic): Add a comment.
17284 (emit_diagnostic_valist): New function.
17286 2017-08-09 Marek Polacek <polacek@redhat.com>
17289 * input.c (make_location): New overload.
17290 * input.h (make_location): Declare.
17292 2017-08-08 Alan Modra <amodra@gmail.com>
17293 H.J. Lu <hongjiu.lu@intel.com>
17296 * gcc.c (NO_PIE_SPEC): Delete.
17297 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
17299 (LINK_PIE_SPEC): ..to here.
17300 (LINK_COMMAND_SPEC): Support -no-pie.
17301 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
17302 chain of crtbegin*.o selection, update for PIE_SPEC changes and
17304 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
17305 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
17306 (ENDFILE_CRTEND_SPEC): Similarly.
17308 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
17311 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
17312 (mstack-protector-guard-offset=): Ditto.
17313 * config/i386/i386.c (ix86_option_override): Handle
17314 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
17316 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
17317 ix86_stack_protect_guard_offset variables.
17318 (TARGET_STACK_PROTECT_GUARD): Always define.
17319 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
17320 and -mstack-protector-guard-offset= options.
17322 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17324 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
17325 boundary case for the last candidate.
17327 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17329 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
17330 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
17332 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17334 PR middle-end/19706
17335 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
17336 * config/aarch64/aarch64-builtins.c
17337 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
17338 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
17339 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
17341 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17342 Andrew Pinski <pinskia@gmail.com>
17344 PR middle-end/19706
17345 * internal-fn.def (XORSIGN): New.
17346 * optabs.def (xorsign_optab): New.
17347 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
17348 (convert_expand_mult_copysign): New.
17349 (pass_optimize_widening_mul::execute): Call
17350 convert_expand_mult_copysign.
17352 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17354 PR tree-optimization/81354
17355 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
17356 Insert on edges rather than explicitly creating landing pads.
17357 (analyze_candidates_and_replace): Commit edge inserts.
17359 2017-08-08 Richard Biener <rguenther@suse.de>
17361 PR middle-end/81719
17362 * tree-ssa-loop-niter.c: Include tree-dfa.h.
17363 (expand_simple_operations): Also look through ADDR_EXPRs with
17364 MEM_REF bases treating them as POINTER_PLUS_EXPR.
17366 2017-08-08 Richard Biener <rguenther@suse.de>
17368 PR tree-optimization/81723
17369 * tree-vect-slp.c (struct bst_traits): New hash traits.
17370 (bst_fail): New global.
17371 (vect_build_slp_tree_2): New worker, split out from ...
17372 (vect_build_slp_tree): ... this now wrapping it with using
17373 bst_fail set to cache SLP tree build fails. Properly handle
17375 (vect_analyze_slp_instance): Allocate and free bst_fail.
17377 2017-08-08 Martin Liska <mliska@suse.cz>
17380 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
17381 LABEL_DECLs that can be from a different function.
17383 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17385 PR tree-optimization/81744
17386 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
17387 loop's number of iterations.
17389 2017-08-08 Martin Liska <mliska@suse.cz>
17391 * asan.c: Include header files.
17392 * attribs.c (build_decl_attribute_variant): New function moved
17394 (build_type_attribute_qual_variant): Likewise.
17395 (cmp_attrib_identifiers): Likewise.
17396 (simple_cst_list_equal): Likewise.
17397 (omp_declare_simd_clauses_equal): Likewise.
17398 (attribute_value_equal): Likewise.
17399 (comp_type_attributes): Likewise.
17400 (build_type_attribute_variant): Likewise.
17401 (lookup_ident_attribute): Likewise.
17402 (remove_attribute): Likewise.
17403 (merge_attributes): Likewise.
17404 (merge_type_attributes): Likewise.
17405 (merge_decl_attributes): Likewise.
17406 (merge_dllimport_decl_attributes): Likewise.
17407 (handle_dll_attribute): Likewise.
17408 (attribute_list_equal): Likewise.
17409 (attribute_list_contained): Likewise.
17410 * attribs.h (lookup_attribute): New function moved from tree.[ch].
17411 (lookup_attribute_by_prefix): Likewise.
17412 * bb-reorder.c: Include header files.
17413 * builtins.c: Likewise.
17414 * calls.c: Likewise.
17415 * cfgexpand.c: Likewise.
17416 * cgraph.c: Likewise.
17417 * cgraphunit.c: Likewise.
17418 * convert.c: Likewise.
17419 * dwarf2out.c: Likewise.
17420 * final.c: Likewise.
17421 * fold-const.c: Likewise.
17422 * function.c: Likewise.
17423 * gimple-expr.c: Likewise.
17424 * gimple-fold.c: Likewise.
17425 * gimple-pretty-print.c: Likewise.
17426 * gimple.c: Likewise.
17427 * gimplify.c: Likewise.
17428 * hsa-common.c: Likewise.
17429 * hsa-gen.c: Likewise.
17430 * internal-fn.c: Likewise.
17431 * ipa-chkp.c: Likewise.
17432 * ipa-cp.c: Likewise.
17433 * ipa-devirt.c: Likewise.
17434 * ipa-fnsummary.c: Likewise.
17435 * ipa-inline.c: Likewise.
17436 * ipa-visibility.c: Likewise.
17438 * lto-cgraph.c: Likewise.
17439 * omp-expand.c: Likewise.
17440 * omp-general.c: Likewise.
17441 * omp-low.c: Likewise.
17442 * omp-offload.c: Likewise.
17443 * omp-simd-clone.c: Likewise.
17444 * opts-global.c: Likewise.
17445 * passes.c: Likewise.
17446 * predict.c: Likewise.
17447 * sancov.c: Likewise.
17448 * sanopt.c: Likewise.
17449 * symtab.c: Likewise.
17450 * toplev.c: Likewise.
17451 * trans-mem.c: Likewise.
17452 * tree-chkp.c: Likewise.
17453 * tree-eh.c: Likewise.
17454 * tree-into-ssa.c: Likewise.
17455 * tree-object-size.c: Likewise.
17456 * tree-parloops.c: Likewise.
17457 * tree-profile.c: Likewise.
17458 * tree-ssa-ccp.c: Likewise.
17459 * tree-ssa-live.c: Likewise.
17460 * tree-ssa-loop.c: Likewise.
17461 * tree-ssa-sccvn.c: Likewise.
17462 * tree-ssa-structalias.c: Likewise.
17463 * tree-ssa.c: Likewise.
17464 * tree-streamer-in.c: Likewise.
17465 * tree-vectorizer.c: Likewise.
17466 * tree-vrp.c: Likewise.
17467 * tsan.c: Likewise.
17468 * ubsan.c: Likewise.
17469 * varasm.c: Likewise.
17470 * varpool.c: Likewise.
17471 * tree.c: Remove functions moved to attribs.[ch].
17472 * tree.h: Likewise.
17473 * config/aarch64/aarch64.c: Add attrs.h header file.
17474 * config/alpha/alpha.c: Likewise.
17475 * config/arc/arc.c: Likewise.
17476 * config/arm/arm.c: Likewise.
17477 * config/avr/avr.c: Likewise.
17478 * config/bfin/bfin.c: Likewise.
17479 * config/c6x/c6x.c: Likewise.
17480 * config/cr16/cr16.c: Likewise.
17481 * config/cris/cris.c: Likewise.
17482 * config/darwin.c: Likewise.
17483 * config/epiphany/epiphany.c: Likewise.
17484 * config/fr30/fr30.c: Likewise.
17485 * config/frv/frv.c: Likewise.
17486 * config/ft32/ft32.c: Likewise.
17487 * config/h8300/h8300.c: Likewise.
17488 * config/i386/winnt.c: Likewise.
17489 * config/ia64/ia64.c: Likewise.
17490 * config/iq2000/iq2000.c: Likewise.
17491 * config/lm32/lm32.c: Likewise.
17492 * config/m32c/m32c.c: Likewise.
17493 * config/m32r/m32r.c: Likewise.
17494 * config/m68k/m68k.c: Likewise.
17495 * config/mcore/mcore.c: Likewise.
17496 * config/microblaze/microblaze.c: Likewise.
17497 * config/mips/mips.c: Likewise.
17498 * config/mmix/mmix.c: Likewise.
17499 * config/mn10300/mn10300.c: Likewise.
17500 * config/moxie/moxie.c: Likewise.
17501 * config/msp430/msp430.c: Likewise.
17502 * config/nds32/nds32-isr.c: Likewise.
17503 * config/nds32/nds32.c: Likewise.
17504 * config/nios2/nios2.c: Likewise.
17505 * config/nvptx/nvptx.c: Likewise.
17506 * config/pa/pa.c: Likewise.
17507 * config/pdp11/pdp11.c: Likewise.
17508 * config/powerpcspe/powerpcspe.c: Likewise.
17509 * config/riscv/riscv.c: Likewise.
17510 * config/rl78/rl78.c: Likewise.
17511 * config/rx/rx.c: Likewise.
17512 * config/s390/s390.c: Likewise.
17513 * config/sh/sh.c: Likewise.
17514 * config/sol2.c: Likewise.
17515 * config/sparc/sparc.c: Likewise.
17516 * config/spu/spu.c: Likewise.
17517 * config/stormy16/stormy16.c: Likewise.
17518 * config/tilegx/tilegx.c: Likewise.
17519 * config/tilepro/tilepro.c: Likewise.
17520 * config/v850/v850.c: Likewise.
17521 * config/vax/vax.c: Likewise.
17522 * config/visium/visium.c: Likewise.
17523 * config/xtensa/xtensa.c: Likewise.
17525 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17528 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
17529 constraints since the -mupper-regs-* switches have been
17531 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
17532 into a vector from a double word element that was extracted from
17533 another vector, and eliminate extra XXPERMDI instructions.
17534 (vsx_concat_<mode>_2): Likewise.
17535 (vsx_concat_<mode>_3): Likewise.
17536 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
17537 concat to allow optimizing inserts from previous extracts.
17539 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
17541 * config/i386/i386.c (ix86_stack_protect_guard): Generate
17542 memory reference to a SSP offset in TLS address space.
17543 (ix86_print_operand) <case '@'>: Remove.
17544 (ix86_print_operand_punct_valid_p): Remove '@' code.
17545 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
17546 UNSPEC_SP_TLS_TEST.
17547 (stack_tls_protect_set_<mode>): Remove.
17548 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
17549 (stack_tls_protect_test_<mode>): Remove.
17550 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
17552 2017-08-07 Olivier Hainque <hainque@adacore.com>
17555 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
17557 2017-08-07 Douglas Rupp <rupp@adacore.com>
17559 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
17560 variable was referenced as multidir in command.
17562 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17565 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
17568 2017-08-07 Martin Liska <mliska@suse.cz>
17570 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
17571 * config/rl78/rl78.c: Add include of attribs.h.
17572 * config/sh/sh.c: Likewise.
17573 * config/v850/v850.c: Likewise.
17575 2017-08-07 Tom de Vries <tom@codesourcery.com>
17577 PR middle-end/78266
17578 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
17580 2017-08-07 Martin Liska <mliska@suse.cz>
17582 * config/mips/mips.c: Include attribs.h.
17584 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
17587 * doc/invoke.texi: Document change in behvaior for -Ofast for
17590 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
17592 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
17594 (aarch64_pop_regs): Likewise.
17595 (aarch64_gen_load_pair): Likewise.
17596 (aarch64_save_callee_saves): Likewise.
17597 (aarch64_restore_callee_saves): Likewise.
17599 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17601 * config/i386/i386.c: Revert the last change.
17603 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17606 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17608 (ix86_finalize_stack_frame_flags): This. Also clear
17609 frame_pointer_needed if -fno-omit-frame-pointer is used without
17611 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17612 with ix86_finalize_stack_frame_flags.
17613 (ix86_expand_epilogue): Likewise.
17614 (ix86_expand_split_stack_prologue): Likewise.
17616 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17619 * config/i386/i386.c (get_builtin_code_for_version): Set priority
17620 to P_AES for Westmere.
17622 2017-08-07 Jonathan Yong <10walls@gmail.com>
17624 * config/i386/mingw.opt (fset-stack-executable): Removed.
17625 * config/i386/cygming.opt (fset-stack-executable): Moved
17627 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
17629 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
17631 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
17633 2017-08-07 Marek Polacek <polacek@redhat.com>
17635 PR middle-end/81737
17636 * fold-const.c (fold_indirect_ref_1): Check type_domain.
17638 2017-08-07 Martin Liska <mliska@suse.cz>
17640 * attribs.h (canonicalize_attr_name): New function.
17641 (cmp_attribs): Move from c-format.c and adjusted.
17642 (is_attribute_p): Moved from tree.h.
17643 * tree-inline.c: Add new includes.
17644 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
17645 (private_is_attribute_p): Remove.
17646 (private_lookup_attribute): Likewise.
17647 (private_lookup_attribute_by_prefix): Simplify.
17648 (remove_attribute): Use is_attribute_p.
17649 * tree.h: Remove removed declarations.
17651 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17653 PR middle-end/81698
17654 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
17655 instead of computing it in the function. Formatting fix.
17656 (expand_case): Don't rely on default_edge being the first edge,
17657 clear it if removing it, pass default_edge to
17658 emit_case_dispatch_table.
17659 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
17662 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
17664 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
17665 insn in the function, emit NOP after the insn.
17667 2017-08-06 Tom de Vries <tom@codesourcery.com>
17669 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
17672 2017-08-06 Tom de Vries <tom@codesourcery.com>
17674 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
17677 2017-08-04 Yury Gribov <tetra2005@gmail.com>
17679 PR tree-optimization/57371
17680 * match.pd: New pattern.
17682 2017-08-04 Marek Polacek <polacek@redhat.com>
17684 PR middle-end/81695
17685 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
17686 perform the computation in offset_int.
17688 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17690 PR tree-optimization/81136
17691 * tree-vectorizer.h: Include tree-hash-traits.h.
17692 (vec_base_alignments): New typedef.
17693 (vec_info): Add a base_alignments field.
17694 (vect_record_base_alignments): Declare.
17695 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
17697 (DR_IS_CONDITIONAL_IN_STMT): New macro.
17698 (create_data_ref): Add an is_conditional_in_stmt argument.
17699 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
17700 the is_conditional_in_stmt field.
17701 (data_ref_loc): Add an is_conditional_in_stmt field.
17702 (get_references_in_stmt): Set the is_conditional_in_stmt field.
17703 (find_data_references_in_stmt): Update call to create_data_ref.
17704 (graphite_find_data_references_in_stmt): Likewise.
17705 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
17706 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
17707 (vect_record_base_alignment): New function.
17708 (vect_record_base_alignments): Likewise.
17709 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
17710 for nested statements even if we fail to compute a misalignment.
17711 Use pooled base alignments for unconditional references.
17712 (vect_find_same_alignment_drs): Compare base addresses instead
17714 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
17715 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
17717 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17719 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
17720 Add an explicit name for the enum. Use auto_vec for slp_instances
17721 and grouped_stores.
17722 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
17724 (_bb_vec_info): Add a constructor and destructor.
17725 (vinfo_for_stmt): Return NULL for uids of -1 as well.
17726 (destroy_loop_vec_info): Delete.
17727 (vect_destroy_datarefs): Likewise.
17728 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
17729 (vec_info::vec_info): New function.
17730 (vec_info::~vec_info): Likewise.
17731 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
17732 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
17733 destroy_loop_vec_info.
17734 * tree-vect-loop.c (new_loop_vec_info): Replace with...
17735 (_loop_vec_info::_loop_vec_info): ...this.
17736 (destroy_loop_vec_info): Replace with...
17737 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
17738 the stmt_vec_infos. Leave handling of vec_info information to its
17739 destructor. Remove explicit vector releases.
17740 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
17741 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
17742 * tree-vect-slp.c (new_bb_vec_info): Replace with...
17743 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
17744 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
17745 (destroy_bb_vec_info): Replace with...
17746 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
17747 information to its destructor.
17748 (vect_slp_analyze_bb_1): Use new and delete instead of
17749 new_bb_vec_info and destroy_bb_vec_info.
17750 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
17753 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17755 * tree-data-ref.h (subscript): Add access_fn field.
17756 (data_dependence_relation): Add could_be_independent_p.
17757 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
17758 (same_access_functions): Move to tree-data-ref.c.
17759 * tree-data-ref.c (ref_contains_union_access_p): New function.
17760 (access_fn_component_p): Likewise.
17761 (access_fn_components_comparable_p): Likewise.
17762 (dr_analyze_indices): Add a reference to access_fn_component_p.
17763 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
17765 (constant_access_functions): Likewise.
17766 (add_other_self_distances): Likewise.
17767 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
17768 (initialize_data_dependence_relation): Use XCNEW and remove
17769 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
17770 of access functions that have the same type. Allow the
17771 subsequence to end with different bases in some circumstances.
17772 Record the chosen access functions in SUB_ACCESS_FN.
17773 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
17774 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
17775 (subscript_dependence_tester_1): Likewise dra and drb.
17776 (build_classic_dist_vector): Update calls accordingly.
17777 (subscript_dependence_tester): Likewise.
17778 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
17779 DDR_COULD_BE_INDEPENDENT_P.
17780 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
17781 comp_alias_ddrs instead of may_alias_ddrs.
17782 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
17784 (vect_analyze_data_ref_dependence): Use it if
17785 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
17786 distance vectors if that fails.
17787 (dependence_distance_ge_vf): New function.
17788 (vect_prune_runtime_alias_test_list): Use it. Don't clear
17789 LOOP_VINFO_MAY_ALIAS_DDRS.
17791 2017-08-04 Richard Biener <rguenther@suse.de>
17793 PR middle-end/81705
17794 * fold-const.c (fold_binary_loc): Properly restrict
17795 minus_var0 && minus_var1 case when associating undefined overflow
17798 2017-08-04 Tom de Vries <tom@codesourcery.com>
17800 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
17802 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17804 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17805 Don't start diagnostic messages with a capital letter.
17806 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17808 (rs6000_invalid_builtin): Likewise.
17809 (rs6000_trampoline_init): Likewise.
17811 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17814 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
17815 after setting changeable df flags.
17817 2017-08-03 Richard Biener <rguenther@suse.de>
17819 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
17820 up if the use is in USE - X.
17822 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
17824 * toplev.c (dumpfile.h): New include.
17825 (internal_error_reentered): New static function. Use it...
17826 (internal_error_function): ...here to handle reentered internal_error.
17828 2017-08-03 Richard Biener <rguenther@suse.de>
17830 PR middle-end/81148
17831 * fold-const.c (split_tree): Add minus_var and minus_con
17832 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
17833 here but always use minus_*.
17834 (associate_trees): Assert we never associate with MINUS_EXPR
17835 and NULL first operand. Do not recurse for PLUS_EXPR operands
17836 when associating as MINUS_EXPR either.
17837 (fold_binary_loc): Track minus_var and minus_con.
17839 2017-08-03 Tom de Vries <tom@codesourcery.com>
17842 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
17843 ACCEL_COMPILER, apply finish_options on
17844 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
17846 2017-08-03 Tom de Vries <tom@codesourcery.com>
17849 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
17850 function_entry_patch_area_size > 0.
17852 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17855 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
17856 instead of 10??LU, perform unit multiplication in wide_int,
17857 don't change alloc_object_size_limit if the limit is larger
17860 PR tree-optimization/81655
17861 PR tree-optimization/81588
17862 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
17863 the case when ranges[i].low and high are 1 for unsigned type with
17866 PR middle-end/81052
17867 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
17868 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
17870 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17872 * tree-vrp.h: Add include guard.
17874 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
17877 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
17878 (ud2): New insn pattern.
17879 * config/i386/i386.c (ix86_expand_epilogue):
17880 For naked functions, generate ud2 instead of trap insn.
17882 2017-08-02 Marek Polacek <polacek@redhat.com>
17885 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
17887 2017-08-02 Tom de Vries <tom@codesourcery.com>
17888 Cesar Philippidis <cesar@codesourcery.com>
17890 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
17891 Add missing edge probabilities.
17893 2017-08-02 Tamar Christina <tamar.christina@arm.com>
17895 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
17896 Correct endianness.
17898 2017-08-02 Jakub Jelinek <jakub@redhat.com>
17900 PR middle-end/79499
17901 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
17902 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
17903 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
17904 prologue_seq sequences - if any.
17906 2017-08-02 Richard Biener <rguenther@suse.de>
17908 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
17909 via vectors if supported, integer extracts via punning if supported
17910 or otherwise vector extracts.
17912 2017-08-02 Richard Biener <rguenther@suse.de>
17914 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
17916 (bitmap_insert_into_set): ... this.
17918 2017-08-02 Richard Biener <rguenther@suse.de>
17920 PR tree-optimization/81633
17922 2015-08-17 Alan Hayward <alan.hayward@arm.com>
17924 PR tree-optimization/71752
17925 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
17927 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
17929 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
17930 (machine_function::call_ms2sysv_pad_out): Remove field.
17931 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
17932 (ix86_compute_frame_layout): Likewise.
17934 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
17937 * config/i386/i386.c (ix86_set_func_type): Disallow naked
17938 attribute with interrupt attribute.
17940 2017-08-01 Andrew Pinski <apinski@cavium.com>
17942 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
17943 BIT_INSERT_EXPR's operand 1
17944 to see if the types precision matches.
17946 2017-08-01 Martin Liska <mliska@suse.cz>
17948 PR middle-end/70140
17949 * builtins.c (expand_builtin_memcpy_args): Remove.
17950 (expand_builtin_memcpy): Call newly added function
17951 expand_builtin_memory_copy_args.
17952 (expand_builtin_memcpy_with_bounds): Likewise.
17953 (expand_builtin_mempcpy): Remove last argument.
17954 (expand_builtin_mempcpy_with_bounds): Likewise.
17955 (expand_builtin_memory_copy_args): New function created from
17956 expand_builtin_mempcpy_args with small modifications.
17957 (expand_builtin_mempcpy_args): Remove.
17958 (expand_builtin_stpcpy): Remove unused argument.
17959 (expand_builtin): Likewise.
17960 (expand_builtin_with_bounds): Likewise.
17962 2017-08-01 Martin Liska <mliska@suse.cz>
17965 Make mempcpy more optimal (PR middle-end/70140).
17967 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17970 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
17971 __builtin_vec_cmpne verify both arguments are compatible vectors
17972 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
17973 verify arg1_type is a pointer or array type. For __builtin_vec_st,
17974 move computation of aligned to after checking the argument types.
17978 * config/rs6000/vsx.md (vextract_fp_from_shorth,
17979 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
17982 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
17983 Doug Rupp <rupp@adacore.com>
17984 Olivier Hainque <hainque@adacore.com>
17986 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
17987 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
17989 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
17990 for TARGET_OS_CPP_BUILTIN.
17991 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
17992 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
17994 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
17995 passing required abi options to the assembler for EABI configurations.
17996 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
17997 of .text.hot and .text.unlikely sections for kernel modules when
17998 using ARM style exceptions.
17999 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
18000 options. Add EXTRA_CC1_SPEC.
18001 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
18003 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
18005 (ARM_TARGET2_DWARF_FORMAT): Define.
18006 * config/arm/t-vxworks: Adjust multilib control to removal of the
18007 Diab command line options.
18009 2017-08-01 Martin Liska <mliska@suse.cz>
18011 PR gcov-profile/81561
18012 * gcov.c (unblock): Make unblocking safe as we need to preserve
18013 index correspondence of blocks and block_lists.
18015 2017-08-01 Richard Biener <rguenther@suse.de>
18017 PR tree-optimization/81181
18018 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
18019 (compute_antic): ... end of iteration here.
18021 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
18023 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
18024 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
18025 (ftree-slp-vectorize): Likewise.
18026 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
18027 can no longer be set independent of flag_tree_loop_vectorize.
18028 * omp-general.c (emp_max_vf): Likewise.
18029 * opts.c (enable_fdo_optimizations): Remove references to
18030 flag_tree_vectorize, these are now implicit.
18031 (common_handle_option): Remove handling for OPT_ftree_vectorize,
18032 and leave it for the options machinery.
18034 2017-08-01 Martin Liska <mliska@suse.cz>
18036 PR middle-end/70140
18037 * builtins.c (expand_builtin_memcpy_args): Remove.
18038 (expand_builtin_memcpy): Call newly added function
18039 expand_builtin_memory_copy_args.
18040 (expand_builtin_memcpy_with_bounds): Likewise.
18041 (expand_builtin_mempcpy): Remove last argument.
18042 (expand_builtin_mempcpy_with_bounds): Likewise.
18043 (expand_builtin_memory_copy_args): New function created from
18044 expand_builtin_mempcpy_args with small modifications.
18045 (expand_builtin_mempcpy_args): Remove.
18046 (expand_builtin_stpcpy): Remove unused argument.
18047 (expand_builtin): Likewise.
18048 (expand_builtin_with_bounds): Likewise.
18050 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18053 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18054 print "ds:" only for immediates in generic address space.
18056 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18059 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18060 (ix86_function_ok_for_sibcall): Return false for naked functions.
18062 2017-08-01 Richard Biener <rguenther@suse.de>
18064 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18065 (compute_antic): Seed worklist with exit block predecessors.
18066 * cfganal.c (dfs_find_deadend): For a cycle return the source
18067 of the edge closing it.
18069 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18071 * config/aarch64/aarch64.c
18072 (aarch64_can_const_movi_rtx_p): Move 0 check.
18074 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18076 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18077 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18079 * match.pd: Ditto in address comparison pattern.
18081 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18083 PR tree-optimization/81627
18084 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18085 closed ssa form for store-store chain.
18087 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18089 PR tree-optimization/81620
18090 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18091 for store-store chain.
18093 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18095 PR tree-optimization/81588
18096 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18097 ranges[i].in_p, invert comparison code ccode. For >/>=,
18098 swap rhs1 and rhs2 and comparison code unconditionally,
18099 for </<= don't do that. Don't swap rhs1/rhs2 again if
18100 ranges[i].in_p, instead invert comparison code ccode if
18101 opcode or oe->rank is BIT_IOR_EXPR.
18104 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18105 a direct optab to conversion optab.
18106 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18107 with GET_MODE_INNER as last argument instead of optab_handler.
18108 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18109 vector extraction if possible and optab is available.
18110 * expr.c (store_constructor): Use convert_optab_handler instead
18111 of optab_handler. Use vector initialization from smaller
18112 vectors if possible and optab is available.
18113 * tree-vect-stmts.c (vectorizable_load): Likewise.
18114 * doc/md.texi (vec_extract, vec_init): Document that the optabs
18115 now have two modes.
18116 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
18117 of vec_init from half-sized vectors with the same element mode.
18118 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
18119 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
18120 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
18121 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
18122 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
18123 after mode in gen_vec_extract* calls.
18124 (vec_extract<mode>): Renamed to ...
18125 (vec_extract<mode><ssescalarmodelower>): ... this.
18126 (vec_extract<mode><ssehalfvecmodelower>): New expander.
18127 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
18128 element mode after mode in gen_vec_init* calls.
18129 (VEC_INIT_HALF_MODE): New mode iterator.
18130 (vec_init<mode>): Renamed to ...
18131 (vec_init<mode><ssescalarmodelower>): ... this.
18132 (vec_init<mode><ssehalfvecmodelower>): New expander.
18133 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
18134 (vec_extractv2sfsf): ... this.
18135 (vec_initv2sf): Renamed to ...
18136 (vec_initv2sfsf): ... this.
18137 (vec_extractv2si): Renamed to ...
18138 (vec_extractv2sisi): ... this.
18139 (vec_initv2si): Renamed to ...
18140 (vec_initv2sisi): ... this.
18141 (vec_extractv4hi): Renamed to ...
18142 (vec_extractv4hihi): ... this.
18143 (vec_initv4hi): Renamed to ...
18144 (vec_initv4hihi): ... this.
18145 (vec_extractv8qi): Renamed to ...
18146 (vec_extractv8qiqi): ... this.
18147 (vec_initv8qi): Renamed to ...
18148 (vec_initv8qiqi): ... this.
18149 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
18150 (vec_init<mode>): Renamed to ...
18151 (vec_init<mode><VEC_base_l>): ... this.
18152 (vec_extract<mode>): Renamed to ...
18153 (vec_extract<mode><VEC_base_l>): ... this.
18154 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
18155 (vec_initv2sfsf): ... this.
18156 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
18157 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18158 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
18159 element mode after mode in gen_vec_init* calls.
18160 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
18161 (vec_init<mode><Vel>): ... this.
18162 (vec_extract<mode>): Renamed to ...
18163 (vec_extract<mode><Vel>): ... this.
18164 * config/aarch64/iterators.md (Vel): New mode attribute.
18165 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
18166 Add element mode after mode in gen_vec_extract* calls.
18167 * config/s390/vector.md (non_vec_l): New mode attribute.
18168 (vec_extract<mode>): Renamed to ...
18169 (vec_extract<mode><non_vec_l>): ... this.
18170 (vec_init<mode>): Renamed to ...
18171 (vec_init<mode><non_vec_l>): ... this.
18172 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
18173 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
18175 * config/arm/iterators.md (V_elem_l): New mode attribute.
18176 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
18177 (vec_extract<mode><V_elem_l>): ... this.
18178 (vec_extractv2di): Renamed to ...
18179 (vec_extractv2didi): ... this.
18180 (vec_init<mode>): Renamed to ...
18181 (vec_init<mode><V_elem_l>): ... this.
18182 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
18183 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
18184 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
18185 Add element mode after gen_vec_extract* calls.
18186 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
18187 (vec_init<mode><unitmode>): ... this.
18188 (vec_extract<mode>): Renamed to ...
18189 (vec_extract<mode><unitmode>): ... this.
18190 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
18191 (vec_init<mode><unitmode>): ... this.
18192 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
18193 (vec_initv2sfsf): ... this.
18194 (vec_extractv2sf): Renamed to ...
18195 (vec_extractv2sfsf): ... this.
18196 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
18197 Add element mode after gen_vec_extract* calls.
18198 * config/mips/mips.md (unitmode): New mode iterator.
18199 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
18200 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
18201 * config/spu/spu.md (inner_l): New mode attribute.
18202 (vec_init<mode>): Renamed to ...
18203 (vec_init<mode><inner_l>): ... this.
18204 (vec_extract<mode>): Renamed to ...
18205 (vec_extract<mode><inner_l>): ... this.
18206 * config/sparc/sparc.md (veltmode): New mode iterator.
18207 (vec_init<VMALL:mode>): Renamed to ...
18208 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
18209 * config/ia64/vect.md (vec_initv2si): Renamed to ...
18210 (vec_initv2sisi): ... this.
18211 (vec_initv2sf): Renamed to ...
18212 (vec_initv2sfsf): ... this.
18213 (vec_extractv2sf): Renamed to ...
18214 (vec_extractv2sfsf): ... this.
18215 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
18216 (vec_init<mode>): Renamed to ...
18217 (vec_init<mode><VEC_base_l>): ... this.
18218 (vec_extract<mode>): Renamed to ...
18219 (vec_extract<mode><VEC_base_l>): ... this.
18220 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
18221 (vec_initv2sfsf): ... this.
18222 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
18223 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18224 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
18225 gen_vec_init* calls.
18227 2017-08-01 Richard Biener <rguenther@suse.de>
18229 PR tree-optimization/81297
18230 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
18231 TREE_OVERFLOW from INTEGER_CSTs.
18233 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
18235 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
18237 2017-07-31 Carl Love <cel@us.ibm.com>
18239 * config/rs6000/rs6000-c: Add support for built-in functions
18240 vector signed char vec_xl_be (signed long long, signed char *);
18241 vector unsigned char vec_xl_be (signed long long, unsigned char *);
18242 vector signed int vec_xl_be (signed long long, signed int *);
18243 vector unsigned int vec_xl_be (signed long long, unsigned int *);
18244 vector signed long long vec_xl_be (signed long long, signed long long *);
18245 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
18246 vector signed short vec_xl_be (signed long long, signed short *);
18247 vector unsigned short vec_xl_be (signed long long, unsigned short *);
18248 vector double vec_xl_be (signed long long, double *);
18249 vector float vec_xl_be (signed long long, float *);
18250 * config/rs6000/altivec.h (vec_xl_be): Add #define.
18251 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
18252 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
18254 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
18255 (altivec_expand_builtin): Add switch statement to call
18256 altivec_expand_xl_be for each builtin.
18257 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
18258 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
18259 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
18260 __builtin_vsx_le_be_v16qi.
18261 * doc/extend.texi: Update the built-in documentation file for the
18262 new built-in functions.
18264 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18267 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
18269 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18271 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18273 * config.gcc: Add z14.
18274 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
18275 CPU model numbers for z13s and z14.
18276 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
18278 * config/s390/s390-opts.h (enum processor_type): Rename
18279 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18280 * config/s390/s390.c (processor_table): Add field for CPU name to
18281 be passed to Binutils.
18282 (s390_asm_output_machine_for_arch): Use the new field in
18283 processor_table for Binutils.
18284 (s390_expand_builtin): Replace arch12 with z14.
18285 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18286 (s390_get_sched_attrmask): Likewise.
18287 (s390_get_unit_mask): Likewise.
18288 * config/s390/s390.opt: Add z14 to processor_type enum.
18290 2017-07-31 Martin Jambor <mjambor@suse.cz>
18293 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
18294 regardless of optimization level.
18296 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
18297 Martin Liska <mliska@suse.cz>
18299 * predict.def: Remove old comment and adjust probability.
18300 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
18301 PREDICT statements.
18303 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18306 * config/i386/i386.c (ix86_function_naked): New function.
18307 (ix86_can_use_return_insn_p): Return false for naked functions.
18308 (ix86_expand_prologue): Skip prologue for naked functions.
18309 (ix86_expand_epilogue): Skip epilogue for naked functions
18310 and emit trap instruction.
18311 (ix86_warn_func_return): New function.
18312 (ix86_attribute_table): Add "naked" attribute specification.
18313 (TARGET_WARN_FUNC_RETURN): Define.
18314 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
18316 2017-07-31 Martin Liska <mliska@suse.cz>
18318 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
18319 (dump_gimple_bb_header): Always dump BB info.
18320 (pp_cfg_jump): Do not append info about BB when dumping a jump.
18322 2017-07-31 Martin Liska <mliska@suse.cz>
18325 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
18326 also with current_function_decl non-null equality.
18328 2017-07-31 Jakub Jelinek <jakub@redhat.com>
18331 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
18332 change type to the element type, instead add eltype variable and
18333 use it where we are interested in the element type.
18335 PR tree-optimization/81603
18336 * ipa-polymorphic-call.c
18337 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
18338 offset arithmetic in offset_int, bail out if the resulting bit offset
18339 doesn't fit into shwi.
18341 2017-07-31 Martin Liska <mliska@suse.cz>
18343 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
18344 (gimplify_save_expr): Fix comment.
18346 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
18349 * config/i386/i386.c (ix86_function_arg): Update arguments for
18351 (ix86_compute_frame_layout): Set the initial stack offset to
18352 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
18353 INCOMING_FRAME_SP_OFFSET.
18354 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
18355 stack before exception handler returns.
18356 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
18357 the 'ERROR_CODE' for exception handler.
18359 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
18361 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
18362 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
18363 (ASM_OUTPUT_REG_POP): Ditto.
18364 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
18365 instead of asm_fprintf to output pure string.
18367 2017-07-29 Jakub Jelinek <jakub@redhat.com>
18369 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
18370 to imported_module_or_decl hook.
18371 (debug_nothing_tree_tree_tree_bool): Remove.
18372 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
18373 * debug.c (do_nothing_debug_hooks): Use
18374 debug_nothing_tree_tree_tree_bool_bool instead of
18375 debug_nothing_tree_tree_tree_bool.
18376 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18377 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
18378 * sdbout.c (sdb_debug_hooks): Likewise.
18379 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
18380 (gen_namespace_die): Add DW_AT_export_symbols attribute if
18382 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
18383 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
18384 attribute, don't add anything.
18386 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18388 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
18389 (fold_build2_stat_loc): Likewise.
18390 (fold_build3_stat_loc): Likewise.
18391 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
18392 (fold_build1_loc): Remove macro.
18393 (fold_build2_loc): Likewise.
18394 (fold_build3_loc): Likewise.
18396 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18398 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
18399 (gimple_build_debug_bind_source_stat): Likewise.
18400 * gimple.h (gimple_build_debug_bind): Remove macro.
18401 (gimple_build_debug_bind_source): Likewise.
18403 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18405 * bitmap.c (bitmap_alloc): Adjust.
18406 (bitmap_gc_alloc): Likewise.
18407 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
18409 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18411 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
18412 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
18413 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
18414 (bitmap_gc_alloc_stat): Likewise.
18415 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
18417 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18419 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
18420 * rtl.h (shallow_copy_rtx): Remove macro.
18422 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18424 * emit-rtl.c (gen_raw_REG): Adjust.
18425 * gengenrtl.c (gendef): Likewise.
18426 * rtl.c (rtx_alloc_stat): Remove _stat from name.
18427 * rtl.h (rtx_alloc): Remove macro.
18429 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18431 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
18432 (build_tree_list_stat): Likewise.
18433 * tree.h (build_tree_list): Remove macro.
18434 (build_tree_list_vec): Likewise.
18436 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18438 * tree.c (make_vector_stat): Remove _stat from name.
18439 (build_vector_stat): Likewise.
18440 * tree.h (make_vector_stat): Remove macro.
18441 (build_vector_stat): Likewise.
18443 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18445 * tree.h (build_var_debug_value): Remove prototype.
18447 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18449 * tree.c (tree_cons_stat): Remove _stat from name.
18450 * tree.h (tree_cons): Remove macro.
18452 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18454 * tree.c (build_vl_exp_stat): Remove _stat from name.
18455 * tree.h (build_vl_exp): Remove macro.
18457 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18459 * tree.c (build_decl_stat): Remove _stat from name.
18460 * tree.h (build_decl): Remove macro.
18462 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18464 * gimple.c (gimple_build_with_ops_stat): Adjust.
18465 (gimple_alloc_stat): Remove _stat from name.
18466 * gimple.h (gimple_alloc): Remove macro.
18468 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18470 * tree.c (make_tree_vec_stat): Remove _stat from name.
18471 (grow_tree_vec_stat): Likewise.
18472 * tree.h (make_tree_vec_stat): Adjust prototype.
18473 (grow_tree_vec_stat): Likewise.
18474 (make_tree_vec): Remove macro.
18475 (grow_tree_vec): Likewise.
18477 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18479 * fold-const.c (fold_build1_stat_loc): Adjust.
18480 (fold_build2_stat_loc): Likewise.
18481 (fold_build3_stat_loc): Likewise.
18482 * tree.c (build0_stat): Remove _stat from name.
18483 (build1_stat): Likewise.
18484 (build2_stat): Likewise.
18485 (build3_stat): Likewise.
18486 (build4_stat): Likewise.
18487 (build5_stat): Likewise.
18488 * tree.h (build1_loc): Remove macro, and rename _stat function
18490 (build2_loc): Likewise.
18491 (build3_loc): Likewise.
18492 (build4_loc): Likewise.
18493 (build5_loc): Likewise.
18495 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18497 * tree.c (make_int_cst_stat): Remove _stat from name.
18498 * tree.h (make_int_cst_stat): Adjust prototype.
18499 (make_int_cst): Remove macro.
18501 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18503 * tree.c (make_tre_binfo_stat): Remove _stat from name.
18504 * tree.h (make_tree_binfo_stat): Adjust prototype.
18505 (make_tree_binfo): Remove.
18507 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18509 * tree.c (copy_node_stat): Rename to copy_node.
18510 (build_distinct_type_copy): Adjust.
18511 * tree.h (copy_node_stat): Adjust prototype.
18512 (copy_node): Remove macro.
18514 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18516 * tree.c (make_node_stat): rename to make_node.
18517 (build_tree_list_stat): Adjust.
18518 (build0_stat): Likewise.
18519 (build2_stat): Likewise.
18520 (build3_stat): Likewise.
18521 (build4_stat): Likewise.
18522 (build5_stat): Likewise.
18523 (build_decl_stat): Likewise.
18524 * tree.h (make_node_stat): Adjust prototype.
18525 (make_node): remove macro.
18527 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
18529 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
18530 (PPC_FEATURE2_SCV): Likewise.
18531 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
18533 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18535 * config/aarch64/aarch64.c
18536 (aarch64_internal_mov_immediate): Add new special pattern.
18537 * config/aarch64/aarch64.md (*movdi_aarch64):
18538 Add reg/32bit const mov case.
18540 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18541 Richard Sandiford <richard.sandiford@linaro.org>
18543 * config/aarch64/aarch64.md (mov<mode>): Generalize.
18544 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18545 Add integer and movi cases.
18546 (movi-split-hf-df-sf split, fp16): New.
18547 (enabled): Added TARGET_FP_F16INST.
18548 * config/aarch64/iterators.md (GPF_HF): New.
18549 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
18551 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18553 * config/aarch64/aarch64.c
18554 (aarch64_simd_container_mode): Add prototype.
18555 (aarch64_expand_mov_immediate): Add HI support.
18556 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
18557 (aarch64_can_const_movi_rtx_p): New.
18558 (aarch64_preferred_reload_class):
18559 Remove restrictions of using FP registers for certain SIMD operations.
18560 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
18561 (aarch64_valid_floating_const): Add integer move validation.
18562 (aarch64_simd_imm_scalar_p): Remove.
18563 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
18564 (aarch64_legitimate_constant_p): Expand list of supported cases.
18565 * config/aarch64/aarch64-protos.h
18566 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
18567 (aarch64_reinterpret_float_as_int): New.
18568 (aarch64_simd_imm_scalar_p): Remove.
18569 * config/aarch64/constraints.md (Uvi): New.
18570 (Dd): Split into Ds and new Dd.
18571 * config/aarch64/aarch64.md (*movsi_aarch64):
18573 (*movdi_aarch64): Add SIMD mov case.
18575 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18577 * tree-predcom.c: (struct chain): Handle store-store chain in which
18578 stores for elimination only store loop invariant values.
18579 (execute_pred_commoning_chain): Ditto.
18580 (prepare_initializers_chain_store_elim): Ditto.
18581 (prepare_finalizers): Ditto.
18582 (is_inv_store_elimination_chain): New function.
18583 (initialize_root_vars_store_elim_1): New function.
18585 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18587 * tree-predcom.c: Revise general description of the pass.
18588 (enum chain_type): New enum type for store elimination.
18589 (struct chain): New field supporting store elimination.
18590 (struct component): Ditto.
18591 (dump_chain): Dump store-stores chain.
18592 (release_chain): Release resources.
18593 (split_data_refs_to_components): Compute and create component
18594 contains only stores for elimination.
18595 (get_chain_last_ref_at): New function.
18596 (make_invariant_chain): Initialization.
18597 (make_rooted_chain): Specify chain type in parameter and record it.
18598 (add_looparound_copies): Skip for store-stores chain.
18599 (determine_roots_comp): Compute type of chain and pass it to
18601 (initialize_root_vars_store_elim_2): New function.
18602 (finalize_eliminated_stores): New function.
18603 (remove_stmt): Handle store for elimination.
18604 (execute_pred_commoning_chain): Execute predictive commoning on
18605 store-store chains.
18606 (determine_unroll_factor): Skip unroll for store-stores chain.
18607 (prepare_initializers_chain_store_elim): New function.
18608 (prepare_initializers_chain): Hanlde store-store chain.
18609 (prepare_finalizers_chain, prepare_finalizers): New function.
18610 (tree_predictive_commoning_loop): Return integer value indicating
18611 if loop is unrolled or lcssa form is corrupted.
18612 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
18614 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18616 * tree-predcom.c (initialize_root): Delete.
18617 (execute_pred_commoning_chain): Initialize root vars and replace
18618 reference of non-combined chain directly, rather than call above
18621 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18623 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
18624 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
18626 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18628 * tree-predcom.c (struct chain): New field init_seq.
18629 (release_chain): Release init_seq.
18630 (prepare_initializers_chain): Record intialization stmts in above
18632 (insert_init_seqs): New function.
18633 (tree_predictive_commoning_loop): Call insert_init_seqs.
18635 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18637 * tree-predcom.c (determine_roots_comp): Skip trivial components.
18639 2017-07-28 Richard Biener <rguenther@suse.de>
18641 * match.pd: Remove superfluous :c.
18642 * genmatch.c (simplify::id): Add member.
18643 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
18645 (current_id): New global.
18646 (dt_node::parent): Move from ...
18647 (dt_operand::parent): ... here. Add for_id member.
18648 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
18649 (decision_tree::find_node): Relax order requirement when
18650 merging DT_TRUE nodes to ones inbetween the current simplify
18651 and the one we try to merge with. Add diagnostic whenever
18652 we need to enforce pattern order by not merging.
18653 (decision_tree::insert): Set current_id.
18654 (decision_tree::print_node): Dump parent node and for_id.
18655 (parser::last_id): Add member.
18656 (parser::push_simplify): Assign unique id.
18657 (parser::parser): Initialize last_id.
18659 2017-07-28 Martin Liska <mliska@suse.cz>
18662 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
18663 gimple_build_debug_bind.
18665 2017-07-28 Richard Biener <rguenther@suse.de>
18667 PR tree-optimization/81502
18668 * match.pd: Add pattern combining BIT_INSERT_EXPR with
18670 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
18672 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
18673 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
18674 for BIT_FIELD_REF args.
18675 * fold-const.c (make_bit_field_ref): Likewise.
18676 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
18678 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18681 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
18682 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
18683 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
18684 Or it into SANITIZER_UNDEFINED.
18685 * ubsan.c: Include gimple-fold.h and varasm.h.
18686 (ubsan_expand_ptr_ifn): New function.
18687 (instrument_pointer_overflow): New function.
18688 (maybe_instrument_pointer_overflow): New function.
18689 (instrument_object_size): Formatting fix.
18690 (pass_ubsan::execute): Call instrument_pointer_overflow
18691 and maybe_instrument_pointer_overflow.
18692 * internal-fn.c (expand_UBSAN_PTR): New function.
18693 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
18694 * sanitizer.def (__ubsan_handle_pointer_overflow,
18695 __ubsan_handle_pointer_overflow_abort): New builtins.
18696 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
18697 * internal-fn.def (UBSAN_PTR): New internal function.
18698 * opts.c (sanitizer_opts): Add pointer-overflow.
18699 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
18700 * fold-const.c (build_range_check): Compute pointer range check in
18701 integral type if pointer arithmetics would be needed. Formatting
18704 2017-07-28 Martin Liska <mliska@suse.cz>
18707 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
18708 parameters that are of a variable-length.
18710 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
18712 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
18714 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
18715 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
18716 (CRT_CALL_STATIC_FUNCTION): Likewise.
18717 (ASM_DEFAULT_SPEC): New define.
18718 (ASM_SPEC32): Likewise.
18719 (ASM_SPEC64): Likewise.
18720 (ASM_SPEC_COMMON): Likewise.
18721 (ASM_SPEC): Likewise.
18722 (INVALID_64BIT): Likewise.
18723 (LINK_OS_DEFAULT_SPEC): Likewise.
18724 (LINK_OS_SPEC32): Likewise.
18725 (LINK_OS_SPEC64): Likewise.
18726 (POWERPC_LINUX): Likewise.
18727 (PTRDIFF_TYPE): Likewise.
18728 (RESTORE_FP_PREFIX): Likewise.
18729 (RESTORE_FP_SUFFIX): Likewise.
18730 (SAVE_FP_PREFIX): Likewise.
18731 (SAVE_FP_SUFFIX): Likewise.
18732 (SIZE_TYPE): Likewise.
18733 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
18734 (TARGET_64BIT): Likewise.
18735 (TARGET_64BIT): Likewise.
18736 (TARGET_AIX): Likewise.
18737 (WCHAR_TYPE_SIZE): Likewise.
18738 (WCHAR_TYPE): Undefine.
18739 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
18740 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
18741 (CPP_OS_RTEMS_SPEC): Delete.
18742 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
18743 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
18745 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
18747 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18749 PR tree-optimization/81578
18750 * tree-parloops.c (build_new_reduction): Bail out if
18751 reduction_code isn't one of the standard OpenMP reductions.
18752 Move the details printing after that decision.
18754 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18756 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
18757 related to reload_in_progress.
18758 (splat_input_operand): Likewise.
18759 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
18761 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
18763 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
18764 (TARGET_INSTANTIATE_DECLS): Likewise.
18765 (legitimate_indexed_address_p): Delete reload_in_progress code.
18766 (rs6000_debug_legitimate_address_p): Likewise.
18767 (rs6000_eliminate_indexed_memrefs): Likewise.
18768 (rs6000_emit_le_vsx_store): Likewise.
18769 (rs6000_emit_move_si_sf_subreg): Likewise.
18770 (rs6000_emit_move): Likewise.
18771 (register_to_reg_type): Likewise.
18772 (rs6000_pre_atomic_barrier): Likewise.
18773 (rs6000_machopic_legitimize_pic_address): Likewise.
18774 (rs6000_allocate_stack_temp): Likewise.
18775 (rs6000_address_for_fpconvert): Likewise.
18776 (rs6000_address_for_altivec): Likewise.
18777 (rs6000_secondary_memory_needed_rtx): Delete function.
18778 (rs6000_check_sdmode): Likewise.
18779 (rs6000_alloc_sdmode_stack_slot): Likewise.
18780 (rs6000_instantiate_decls): Likewise.
18781 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
18782 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
18783 Delete reload_in_progress.
18784 (*vec_reload_and_plus_<mptrsize>): Likewise.
18785 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
18786 (vsx_div_v2di): Likewise.
18787 (vsx_udiv_v2di): Likewise.
18789 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18791 * config/rs6000/rs6000.opt (mlra): Replace with stub.
18792 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
18793 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
18794 (rs6000_debug_reg_global): Delete print of LRA status.
18795 (rs6000_option_override_internal): Delete dead LRA related code.
18796 (rs6000_lra_p): Delete function.
18797 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
18799 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
18801 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
18802 * config/riscv/rtems.h: New file.
18804 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18805 Sudakshina Das <sudi.das@arm.com>
18807 * config/aarch64/aarch64.md
18808 (define_split for and<mode>3nr_compare): Move
18809 non aarch64_logical_operand to a register.
18810 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
18811 register immediate operand to a register.
18812 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
18814 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18816 PR middle-end/81564
18817 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
18819 2017-07-27 Richard Biener <rguenther@suse.de>
18821 PR tree-optimization/81573
18822 PR tree-optimization/81494
18823 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
18824 multi defuse cycle case.
18826 2017-07-27 Richard Biener <rguenther@suse.de>
18828 PR tree-optimization/81571
18829 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
18832 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
18834 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
18835 earlier and only if MASK_FPU is set. Adjust formatting.
18837 2017-07-27 Martin Liska <mliska@suse.cz>
18839 * opt-functions.awk: Add validation of value of Init.
18840 * optc-gen.awk: Pass new argument.
18842 2017-07-27 Martin Liska <mliska@suse.cz>
18844 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
18845 Fix wrong condition.
18847 2017-07-27 Martin Liska <mliska@suse.cz>
18849 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
18850 BBs and edges seen by autoFDO.
18852 2017-07-27 Richard Biener <rguenther@suse.de>
18854 PR tree-optimization/81502
18855 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
18856 with incompatible but same sized type.
18857 (execute_update_addresses_taken): Likewise.
18859 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
18861 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
18862 flag_tree_loop_vectorize rather than flag_tree_vectorize.
18864 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18867 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
18868 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
18869 Change s_operand to memory_operand.
18871 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
18873 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
18874 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
18875 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
18876 Emit instructions rather than returning an expression. Handle TFmode
18877 and KFmode by casting to TImode.
18878 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
18879 (rs6000_emit_le_vsx_store): Likewise.
18880 * config/rs6000/vsx.md (VSX_TI): New iterator.
18881 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
18882 (*vsx_le_undo_permute_<mode>): Likewise.
18883 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
18884 emit the split sequence.
18885 (*vsx_le_perm_store_<mode>): Likewise.
18887 2017-07-27 Jakub Jelinek <jakub@redhat.com>
18889 PR tree-optimization/81555
18890 PR tree-optimization/81556
18891 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
18892 if true, force CHANGED for the recursive invocation.
18893 (reassociate_bb): Remember original length of ops array, pass
18894 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
18896 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
18897 attributes for noipa attribute. For naked attribute use
18898 lookup_attribute first before lookup_attribute_spec.
18899 * final.c (rest_of_handle_final): Disable IPA RA for functions with
18901 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
18902 for functions with noipa attribute.
18903 (cgraph_externally_visible_p): Return true for functions with noipa
18905 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
18906 for functions with noipa attribute.
18907 * doc/extend.texi: Document noipa function attribute.
18908 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
18909 also for functions with noipa attribute.
18910 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
18912 2017-07-26 Andrew Pinski <apinski@cavium.com>
18914 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
18915 vec_unalign_load_cost and vec_unalign_store_cost.
18917 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
18919 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
18920 -mvsx-small-integer option.
18921 (ISA_3_0_MASKS_IEEE): Likewise.
18922 (OTHER_VSX_VECTOR_MASKS): Likewise.
18923 (POWERPC_MASKS): Likewise.
18924 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
18925 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
18926 code, only testing for DImode being allowed in non-VSX floating
18928 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
18929 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
18931 (rs6000_option_override_internal): Delete -mvsx-small-integer.
18932 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
18933 TARGET_P8_VECTOR test.
18934 (rs6000_secondary_reload_simple_move): Likewise.
18935 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
18936 since TARGET_P9_VECTOR was already tested.
18937 (rs6000_opt_masks): Remove -mvsx-small-integer.
18938 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
18939 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18941 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
18942 test for TARGET_VEXTRACTUB was used, and that uses
18944 (p9 extract splitter): Likewise.
18945 (vsx_extract_<mode>_di_p9): Likewise.
18946 (vsx_extract_<mode>_store_p9): Likewise.
18947 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
18948 for TARGET_P9_VECTOR was used. Delete code that is now dead with
18949 the elimination of TARGET_VSX_SMALL_INTEGER.
18950 (vsx_extract_<mode>_p8): Likewise.
18951 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
18952 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
18953 (vsx_set_<mode>_p9): Likewise.
18954 (vsx_set_v4sf_p9): Likewise.
18955 (vsx_set_v4sf_p9_zero): Likewise.
18956 (vsx_insert_extract_v4sf_p9): Likewise.
18957 (vsx_insert_extract_v4sf_p9_2): Likewise.
18958 * config/rs6000/rs6000.md (sign extend splitter): Change
18959 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
18960 (floatsi<mode>2_lfiwax_mem): Likewise.
18961 (floatunssi<mode>2_lfiwzx_mem): Likewise.
18962 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
18963 since a test for TARGET_P9_VECTOR was used.
18964 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18965 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
18966 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18967 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18968 TARGET_P8_VECTOR test.
18969 (fix_trunc<mode>si2_stfiwx): Likewise.
18970 (fix_trunc<mode>si2_internal): Likewise.
18971 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
18972 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18974 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18975 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18976 TARGET_P8_VECTOR test.
18977 (fixuns_trunc<mode>si2_stfiwx): Likewise.
18978 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
18979 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18981 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18982 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
18983 since a test for TARGET_P9_VECTOR was used.
18984 (splitter for loading small constants): Likewise.
18986 2017-07-26 Andrew Pinski <apinski@cavium.com>
18988 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
18991 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
18994 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
18995 (fp_valid_at): Likewise.
18997 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18999 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
19000 (qdf24xx_addrcost_table): Likewise.
19001 (cortexa57_tunings): Update to use generic_branch_cost.
19002 (cortexa72_tunings): Likewise.
19003 (cortexa73_tunings): Likewise.
19004 (qdf24xx_tunings): Likewise.
19006 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
19008 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
19009 (thunderx2t99_branch_cost): Likewise.
19010 (cortexa35_tunings): Update to use generic_branch_cost.
19011 (cortexa53_tunings): Likewise.
19012 (cortexa57_tunings): Likewise.
19013 (cortexa72_tunings): Likewise.
19014 (cortexa73_tunings): Likewise.
19015 (thunderx2t99_tunings): Likewise.
19017 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19019 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
19020 (sparc_option_override): Honour MASK_FSMULD.
19021 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
19022 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
19023 * config/sparc/sparc.opt (mfsmuld): New option.
19024 * doc/invoke.texi (mfsmuld): Document option.
19026 2017-07-26 Marek Polacek <polacek@redhat.com>
19028 PR middle-end/70992
19029 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
19031 2017-07-26 Richard Biener <rguenther@suse.de>
19033 * gimple-match-head.c (do_valueize): Return OP if valueize
19035 (get_def): New helper to get at the def stmt of a SSA name
19036 if valueize allows.
19037 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
19038 do_valueize to get at the def stmt.
19039 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
19041 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
19043 PR middle-end/46932
19044 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
19046 2017-07-26 Martin Liska <mliska@suse.cz>
19049 * function.c (expand_function_start): Make expansion of
19050 nonlocal_goto_save_area after parm_birth_insn.
19052 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19054 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19055 from all CPU target flags enable members.
19057 2017-07-26 Richard Biener <rguenther@suse.de>
19059 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19060 (decision_tree::gen): Make 'type' const.
19061 (write_predicate): Likewise.
19063 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19065 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19066 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19067 (rs6000_option_override_internal): Likewise.
19068 (rs6000_expand_vector_set): Likewise.
19069 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19070 (TARGET_UPPER_REGS_SF): Likewise.
19071 (TARGET_UPPER_REGS_DI): Likewise.
19072 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19073 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19074 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19075 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19076 (Splitters for DI constants in Altivec registers): Likewise.
19077 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19078 (vsx_set_v4sf_p9): Likewise.
19079 (vsx_set_v4sf_p9_zero): Likewise.
19080 (vsx_insert_extract_v4sf_p9): Likewise.
19081 (vsx_insert_extract_v4sf_p9_2): Likewise.
19083 2017-07-25 Carl Love <cel@us.ibm.com>
19085 * doc/extend.texi: Update the built-in documentation file for the
19086 existing built-in functions
19087 vector signed char vec_cnttz (vector signed char);
19088 vector unsigned char vec_cnttz (vector unsigned char);
19089 vector signed short vec_cnttz (vector signed short);
19090 vector unsigned short vec_cnttz (vector unsigned short);
19091 vector signed int vec_cnttz (vector signed int);
19092 vector unsigned int vec_cnttz (vector unsigned int);
19093 vector signed long long vec_cnttz (vector signed long long);
19094 vector unsigned long long vec_cnttz (vector unsigned long long);
19096 2017-07-25 Andrew Pinski <apinski@cavium.com>
19098 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19099 accesses where the use is for the first operand of a BIT_INSERT.
19101 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19104 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19105 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19107 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19109 * config/i386/gstabs.h: Delete.
19110 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19112 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19114 * config/i386/i386.c (ix86_decompose_address): Do not check for
19115 register RTX when looking at index_reg or base_reg.
19116 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
19118 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
19120 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
19121 to update EH info here.
19123 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19125 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
19127 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19129 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
19131 2017-07-25 Torsten Duwe <duwe@suse.de>
19133 * common.opt: Introduce -fpatchable-function-entry
19134 command line option, and its variables function_entry_patch_area_size
19135 and function_entry_patch_area_start.
19136 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
19137 including a two-value parser.
19138 * target.def (print_patchable_function_entry): New target hook.
19139 * targhooks.h (default_print_patchable_function_entry): New function.
19140 * targhooks.c (default_print_patchable_function_entry): Likewise.
19141 * toplev.c (process_options): Switch off IPA-RA if
19142 patchable function entries are being generated.
19143 * varasm.c (assemble_start_function): Look at the
19144 patchable-function-entry command line switch and current
19145 function attributes and maybe generate NOP instructions by
19146 calling the print_patchable_function_entry hook.
19147 * doc/extend.texi: Document patchable_function_entry attribute.
19148 * doc/invoke.texi: Document -fpatchable_function_entry
19149 command line option.
19150 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
19152 * doc/tm.texi: Re-generate.
19154 2017-07-25 Jakub Jelinek <jakub@redhat.com>
19157 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
19158 TARGET_AVX512DQ rather than TARGET_AVX512BW.
19160 2017-07-25 Tamar Christina <tamar.christina@arm.com>
19162 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
19164 2017-07-25 Richard Biener <rguenther@suse.de>
19166 PR tree-optimization/81455
19167 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
19168 not walk in cycles when looking for guards.
19170 2017-07-25 Richard Biener <rguenther@suse.de>
19172 PR tree-optimization/81529
19173 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
19174 when optimizing backedge uses.
19176 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
19178 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
19180 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
19181 to dl_section_ref. On AIX, append an expression to subtract
19182 the size of the section length to dl_section_ref.
19184 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
19186 * configure.ac: If any of the config.* scripts fail, exit 1.
19187 * configure: Regenerate.
19189 2017-07-25 Richard Biener <rguenther@suse.de>
19191 PR middle-end/81546
19192 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
19193 of immediate uses, be more verbose on errors.
19195 2017-07-25 Richard Biener <rguenther@suse.de>
19197 PR tree-optimization/81510
19198 * tree-vect-loop.c (vect_is_simple_reduction): When the
19199 reduction stmt is not inside the loop bail out.
19201 2017-07-25 Richard Biener <rguenther@suse.de>
19203 PR tree-optimization/81303
19204 * tree-vect-loop-manip.c (vect_loop_versioning): Build
19205 profitability check against LOOP_VINFO_NITERSM1.
19207 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19209 * domwalk.c (cmp_bb_postorder): Simplify.
19210 (sort_bbs_postorder): New function. Use it...
19211 (dom_walker::walk): ...here to optimize common cases.
19213 2017-07-25 Martin Liska <mliska@suse.cz>
19216 * ipa-visibility.c (function_and_variable_visibility): Make the
19217 redirection just on target that supports aliasing.
19218 Fix GNU coding style.
19220 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
19223 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
19225 * config/arm/rtems.h: Likewise.
19226 * config/bfin/rtems.h: Likewise.
19227 * config/i386/rtemself.h: Likewise.
19228 * config/lm32/rtems.h: Likewise.
19229 * config/m32c/rtems.h: Likewise.
19230 * config/m68k/rtemself.h: Likewise.
19231 * config/microblaze/rtems.h: Likewise.
19232 * config/mips/rtems.h: Likewise.
19233 * config/moxie/rtems.h: Likewise.
19234 * config/nios2/rtems.h: Likewise.
19235 * config/powerpcspe/rtems.h: Likewise.
19236 * config/rs6000/rtems.h: Likewise.
19237 * config/rtems.h: Likewise.
19238 * config/sh/rtems.h: Likewise.
19239 * config/sh/rtemself.h: Likewise.
19240 * config/sparc/rtemself.h: Likewise.
19242 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19245 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
19246 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
19247 * tree-ssa-structalias.c (alias_get_name): Same.
19249 2017-07-25 Bin Cheng <bin.cheng@arm.com>
19252 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
19253 instructions if no du chain is found.
19255 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19257 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
19259 2017-07-25 Richard Biener <rguenther@suse.de>
19261 PR middle-end/81505
19262 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
19265 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19267 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
19268 upper-regs options.
19269 (ISA_2_7_MASKS_SERVER): Likewise.
19270 (ISA_3_0_MASKS_IEEE): Likewise.
19271 (OTHER_P8_VECTOR_MASKS): Likewise.
19272 (OTHER_VSX_VECTOR_MASKS): Likewise.
19273 (POWERPC_MASKS): Likewise.
19274 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
19275 duplicate list of options.
19276 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
19277 explicit -mupper-regs options.
19278 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
19279 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
19280 alias for -mupper-regs-df.
19281 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19282 (rs6000_init_hard_regno_mode_ok): Likewise.
19283 (rs6000_option_override_internal): Likewise.
19284 (rs6000_opt_masks): Likewise.
19285 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
19286 options in terms of whether -mvsx or -mpower8-vector was used.
19287 (TARGET_UPPER_REGS_DI): Likewise.
19288 (TARGET_UPPER_REGS_SF): Likewise.
19289 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
19290 -mupper-regs-* options.
19292 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19294 * passes.c (emergency_dump_function): Print some empty lines and a
19295 header before the RTL dump.
19297 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19299 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
19301 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
19304 * config/aarch64/aarch64.c (aarch64_classify_symbol):
19305 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
19307 2017-07-24 Carl Love <cel@us.ibm.com>
19309 * config/rs6000/rs6000-c.c: Add support for built-in functions
19310 vector float vec_extract_fp32_from_shorth (vector unsigned short);
19311 vector float vec_extract_fp32_from_shortl (vector unsigned short);
19312 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
19313 vec_extract_fp_from_shortl): Add defines for the two builtins.
19314 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
19315 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
19317 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
19318 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
19319 * doc/extend.texi: Update the built-in documentation file for the
19320 new built-in function.
19322 2017-07-24 Jakub Jelinek <jakub@redhat.com>
19325 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
19327 * doc/generic.texi: Likewise.
19328 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19329 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19331 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
19333 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
19334 (aarch64_mls_elt_merge<mode>): Likewise.
19336 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
19338 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19339 having __cxa_atexit.
19341 2017-07-23 Michael Collison <michael.collison@arm.com>
19343 * config/arm/arm.c (arm_option_override): Deprecate
19344 use of -mstructure-size-boundary.
19345 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
19346 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
19348 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19351 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19352 Reduce cost estimate for direct moves.
19354 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
19357 * config/i386/i386.c (ix86_option_override_internal): Disable
19358 BMI, BMI2 and TBM instructions for -m16.
19360 2017-07-21 Carl Love <cel@us.ibm.com>
19362 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19363 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19364 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19365 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19366 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19367 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19368 VMULOSW): New enum "unspec" values.
19369 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19370 altivec_vmulosw): New patterns.
19371 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19372 VMULOSW): Add definitions.
19374 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
19376 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
19377 (qdf24xx): Likewise.
19378 * config/aarch64/aarch64-options-extensions.def (rdma); New.
19379 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
19380 (AARCH64_FL_V8_1): Renumber.
19381 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
19382 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
19383 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
19384 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
19385 rdma to feature modifiers list.
19387 2017-07-21 Yury Gribov <tetra2005@gmail.com>
19389 PR middle-end/56727
19390 * ipa-visibility (function_and_variable_visibility): Convert
19391 recursive PLT call to direct call if appropriate.
19393 2017-07-21 Andrew Pinski <apinski@cavium.com>
19395 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
19396 operand 1 to see if the types precision matches.
19397 * fold-const.c (operand_equal_p): Likewise.
19399 2017-07-21 Richard Biener <rguenther@suse.de>
19401 PR tree-optimization/81303
19402 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
19403 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
19404 (vect_peeling_hash_get_lowest_cost): Adjust.
19405 (vect_enhance_data_refs_alignment): Likewise. Use
19406 vect_get_peeling_costs_all_drs to compute the penalty for no
19407 peeling to match up costs.
19409 2017-07-21 Richard Biener <rguenther@suse.de>
19411 PR tree-optimization/81500
19412 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
19413 we didn't identify a reduction path.
19415 2017-07-21 Tom de Vries <tom@codesourcery.com>
19416 Cesar Philippidis <cesar@codesourcery.com>
19418 PR gcov-profile/81442
19419 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
19422 2017-07-21 Tom de Vries <tom@codesourcery.com>
19425 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
19427 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
19428 nvptx_override_options_after_change.
19430 2017-07-21 Ulrich Drepper <drepper@redhat.com>
19432 * dwarf2out.c (output_file_names): Avoid double testing for
19433 dwarf_version >= 5.
19435 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
19437 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
19439 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19441 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
19443 (try_crossjump_to_edge): Do not punt on partitioned functions.
19445 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19447 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
19448 Put all BBs reachable only via paths crossing cold region to cold
19450 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
19452 2016-07-21 Richard Biener <rguenther@suse.de>
19454 PR tree-optimization/81303
19455 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
19456 into account prologue and epilogue iterations when raising
19457 min_profitable_iters to sth at least covering one vector iteration.
19459 2017-07-21 Tamar Christina <tamar.christina@arm.com>
19461 * config/arm/arm.c (arm_test_cpu_arch_dat):
19464 2017-07-20 Nathan Sidwell <nathan@acm.org>
19466 Remove TYPE_METHODS.
19467 * tree.h (TYPE_METHODS): Delete.
19468 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
19469 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
19470 (dbxout_type_methods): Scan TYPE_FIELDS.
19471 (dbxout_type): Don't check TYPE_METHODS here.
19472 * function.c (use_register_for_decl): Always ignore register for
19473 class types when not optimizing.
19474 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
19475 * tree.c (free_lang_data_in_type): Stitch out member functions and
19476 templates from TYPE_FIELDS.
19477 (build_distinct_type_copy, verify_type_variant,
19478 verify_type): Member fns are on TYPE_FIELDS.
19479 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
19480 * tree-pretty-print.c (dump_generic_node): Likewise.
19482 2017-07-20 Jakub Jelinek <jakub@redhat.com>
19485 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
19486 V2TImode and V4TImode.
19487 (ix86_expand_vector_extract): Likewise.
19488 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
19489 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
19490 (ssescalarmode): Handle V4TImode and V2TImode.
19491 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
19492 (*vec_extractv2ti, *vec_extractv4ti): New insns.
19493 (VEXTRACTI128_MODE): New mode iterator.
19494 (splitter for *vec_extractv?ti first element): New.
19495 (VEC_INIT_MODE): New mode iterator.
19496 (vec_init<mode>): Consolidate 3 expanders into one using
19497 VEC_INIT_MODE mode iterator.
19499 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19501 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
19502 non_spilled_static_chain_regno_p.
19504 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19506 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
19508 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
19510 * bb-reorder.c (connect_traces): Allow copying of blocks within
19513 2017-07-20 Richard Biener <rguenther@suse.de>
19515 * gimple.h (gimple_phi_result): Add gphi * overload.
19516 (gimple_phi_result_ptr): Likewise.
19517 (gimple_phi_arg): Likewise. Adjust index assert to only
19518 allow actual argument accesses rather than all slots available
19520 (gimple_phi_arg_def): Add gphi * overload.
19521 * tree-phinodes.c (make_phi_node): Initialize only actual
19523 (resize_phi_node): Clear memory not covered by old node,
19524 do not initialize excess argument slots.
19525 (reserve_phi_args_for_new_edge): Initialize new argument slot
19528 2017-07-20 Bin Cheng <bin.cheng@arm.com>
19530 PR tree-optimization/81388
19532 2016-07-21 Bin Cheng <bin.cheng@arm.com>
19534 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
19535 by removing computation of may_be_zero.
19537 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19538 Tom de Vries <tom@codesourcery.com>
19540 PR middle-end/81030
19541 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
19542 when gimple level profile disagrees with what RTL expander did.
19544 2017-07-20 Richard Biener <rguenther@suse.de>
19546 PR tree-optimization/61171
19547 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
19548 (vect_analyze_stmt): Add slp instance parameter.
19549 (vectorizable_reduction): Likewise.
19550 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
19551 (vect_is_simple_reduction): Deal with chains not detected
19552 as SLP reduction chain, specifically not properly associated
19553 chains containing a mix of plus/minus.
19554 (get_reduction_op): Remove.
19555 (get_initial_defs_for_reduction): Simplify, pass in whether
19556 this is a reduction chain, pass in the SLP node for the PHIs.
19557 (vect_create_epilog_for_reduction): Get the SLP instance as
19559 (vectorizable_reduction): Get the SLP instance as arg.
19560 During analysis remember the SLP node with the PHIs in the
19561 instance. Simplify getting at the vectorized reduction PHIs.
19562 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
19563 through SLP instance.
19564 (vect_slp_analyze_operations): Likewise.
19565 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
19566 (vect_transform_stmt): Likewise.
19568 2017-07-20 Tom de Vries <tom@codesourcery.com>
19570 PR tree-optimization/81489
19571 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
19572 read of phi arg location to before loop that modifies phi.
19574 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19576 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
19579 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19581 PR middle-end/81331
19582 * except.c (execute): Fix ordering issue.
19584 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19586 PR rtl-optimization/81423
19587 * combine.c (make_compound_operation_int): Don't try to optimize
19588 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
19590 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19592 PR rtl-optimization/81423
19593 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
19594 with a constant that is -1 in the truncated to mode.
19596 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19598 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
19599 (determine_unlikely_bbs): ... here.
19600 * predict.h (propagate_unlikely_bbs_forward): Declare.
19601 * cfgexpand.c (pass_expand::execute): Use it.
19602 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
19604 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
19605 propagate_unlikely_bbs_forward.
19607 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19609 PR middle-end/81331
19610 * except.c (maybe_add_nop_after_section_switch): New function.
19613 2017-07-19 Tom de Vries <tom@codesourcery.com>
19615 * gimple.h (gimple_phi_set_arg): Make assert more strict.
19617 2017-07-19 Tom de Vries <tom@codesourcery.com>
19619 * gimple.h (gimple_phi_arg): Make assert more strict.
19621 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
19623 * config.gcc (powerpc*-*-*): Add mmintrin.h.
19624 * config/rs6000/mmintrin.h: New file.
19625 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
19627 2017-07-19 Jakub Jelinek <jakub@redhat.com>
19629 PR tree-optimization/81346
19630 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
19632 2017-07-19 Tom de Vries <tom@codesourcery.com>
19634 * config/nvptx/nvptx.md (VECIM): Add V2DI.
19636 2017-07-19 Tom de Vries <tom@codesourcery.com>
19638 * config/nvptx/nvptx-modes.def: Add V2DImode.
19639 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
19640 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
19641 (nvptx_output_mov_insn): Handle lack of mov.b128.
19642 (nvptx_print_operand): Handle 'H' and 'L' codes.
19643 (nvptx_vector_mode_supported): Allow V2DImode.
19644 (nvptx_preferred_simd_mode): New function.
19645 (nvptx_data_alignment): New function.
19646 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
19647 nvptx_preferred_simd_mode.
19648 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
19650 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
19652 2017-07-19 Tom de Vries <tom@codesourcery.com>
19654 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
19655 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
19656 (nvptx_vector_mode_supported): New function. Allow V2SImode.
19657 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
19658 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
19659 (mov<VECIM>_insn): New define_insn.
19660 (define_expand "mov<VECIM>): New define_expand.
19662 2017-07-19 Tom de Vries <tom@codesourcery.com>
19664 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
19666 2017-07-19 Jakub Jelinek <jakub@redhat.com>
19668 PR tree-optimization/81346
19669 * fold-const.h (fold_div_compare, range_check_type): Declare.
19670 * fold-const.c (range_check_type): New function.
19671 (build_range_check): Use range_check_type.
19672 (fold_div_compare): No longer static, rewritten into
19673 a match.pd helper function.
19674 (fold_comparison): Don't call fold_div_compare here.
19675 * match.pd (X / C1 op C2): New optimization using fold_div_compare
19676 as helper function.
19678 2017-07-19 Nathan Sidwell <nathan@acm.org>
19680 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
19681 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
19682 * tree.c (find_decls_types_r, verify_type): Use
19683 TYPE_{MIN,MAX}_VALUE_RAW.
19684 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
19685 (hash_tree): Likewise.
19686 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19688 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19691 2017-07-18 Tom de Vries <tom@codesourcery.com>
19693 PR middle-end/81464
19694 * omp-expand.c (expand_omp_for_static_chunk): Handle
19695 equal-argument loop exit phi.
19697 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
19700 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
19701 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
19702 operand 2 predicate.
19703 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
19704 operand 2 predicate.
19705 (ror,rol -> rorx splitters): Use const_int_operand as
19706 operand 2 predicate.
19708 2017-06-18 Richard Biener <rguenther@suse.de>
19710 PR tree-optimization/81410
19711 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
19712 the gap in the ! slp_perm SLP case after each group.
19714 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19716 PR middle-end/81463
19717 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
19720 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19722 PR middle-end/81462
19723 * predict.c (set_even_probabilities): Cleanup; do not affect
19724 probabilities that are already known.
19725 (combine_predictions_for_bb): Call even when count is set.
19727 2017-07-18 Nathan Sidwell <nathan@acm.org>
19729 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
19732 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19735 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
19736 optimization for loop niter analysis.
19738 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
19741 * config/avr/avr.c (avr_optimize_casesi): Don't use
19742 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
19744 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
19746 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
19747 body_cost_vec from _vect_peel_extended_info.
19748 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
19749 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
19752 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19754 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
19756 2017-07-18 Richard Biener <rguenther@suse.de>
19758 PR tree-optimization/80620
19759 PR tree-optimization/81403
19760 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
19761 info when re-using a VN table entry.
19763 2017-07-18 Richard Biener <rguenther@suse.de>
19765 PR tree-optimization/81418
19766 * tree-vect-loop.c (vectorizable_reduction): Properly compute
19767 vectype_in. Verify that with lane-reducing reduction operations
19768 we have a single def-use cycle.
19770 2017-07-17 Carl Love <cel@us.ibm.com>
19772 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
19774 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19775 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19776 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19777 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19778 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19779 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19780 VMULOSW): New enum "unspec" values.
19781 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19782 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19783 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19784 altivec_vmulosw): New patterns.
19785 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19786 VMULOSW): Add definitions.
19787 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
19789 * config/alpha/alpha.c: Include predict.h.
19791 2017-07-17 Yury Gribov <tetra2005@gmail.com>
19793 * tree-vrp.c (compare_assert_loc): Fix comparison function
19794 to return predictable results.
19796 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19798 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
19800 (subdi3): Likewise.
19801 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
19802 * doc/invoke.texi (mexpand-adddi): Update text.
19804 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19806 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
19807 that also clobbers the CC register. The old expand code is moved
19809 (*arc_clzsi2): ... here.
19810 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
19811 the CC register. The old expand code is moved to ...
19812 (arc_ctzsi2): ... here.
19814 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19816 * config/arc/arc.opt (mindexed-loads): Use initial value
19817 TARGET_INDEXED_LOADS_DEFAULT.
19818 (mauto-modify-reg): Use initial value
19819 TARGET_AUTO_MODIFY_REG_DEFAULT.
19820 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19821 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19822 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19823 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19825 2017-07-17 Martin Liska <mliska@suse.cz>
19828 * opts.c (finish_options): Do not allow -fgnu-tm
19829 w/ -fsanitize={kernel-,}address. Say sorry.
19831 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19834 * tree-loop-distribution.c (classify_partition): Only assert on
19835 numer of iterations.
19836 (merge_dep_scc_partitions): Delete prameter. Update function call.
19837 (distribute_loop): Remove code handling loop with unknown niters.
19838 (pass_loop_distribution::execute): Skip loop with unknown niters.
19840 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19843 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
19844 function sort_partitions_by_post_order.
19846 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19848 PR tree-optimization/81374
19849 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
19850 the max index of basic blocks, rather than number of basic blocks.
19852 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19854 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
19856 (arc_legitimate_pic_operand_p): Likewise.
19857 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
19859 (arc_needs_pcl_p): Likewise.
19860 (arc_legitimate_pc_offset_p): Likewise.
19861 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
19862 function is also used in constrains.md.
19863 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
19864 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
19865 PLUS. Only return true/false in known cases, otherwise assert.
19866 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
19867 is already called in arc_legitimate_constant_p.
19868 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
19870 (LEGITIMATE_PIC_OPERAND_P): Use
19871 arc_raw_symbolic_reference_mentioned_p function.
19872 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
19877 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19878 Andrew Burgess <andrew.burgess@embecosm.com>
19880 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
19881 (arc_return_address_register): New function.
19882 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
19883 (arc_handle_fndecl_attribute): Add naked attribute.
19884 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19885 (TARGET_WARN_FUNC_RETURN): Likewise.
19886 (arc_allocate_stack_slots_for_args): New function.
19887 (arc_warn_func_return): Likewise.
19888 (machine_function): Change type fn_type.
19889 (arc_compute_function_type): Consider new naked function type,
19890 change function return type.
19891 (arc_must_save_register): Adapt to handle new
19892 arc_compute_function_type's return type.
19893 (arc_expand_prologue): Likewise.
19894 (arc_expand_epilogue): Likewise.
19895 (arc_return_address_regs): Delete.
19896 (arc_return_address_register): New function.
19897 (arc_epilogue_uses): Use above function.
19898 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
19899 (arc_function_type): Change encoding, add naked type.
19900 (ARC_INTERRUPT_P): Change to handle the new encoding.
19901 (ARC_FAST_INTERRUPT_P): Likewise.
19902 (ARC_NORMAL_P): Define.
19903 (ARC_NAKED_P): Likewise.
19904 (arc_compute_function_type): Delete prototype.
19905 * config/arc/arc.md (in_ret_delay_slot): Use
19906 arc_return_address_register function.
19907 (simple_return): Likewise.
19908 (p_return_i): Likewise.
19910 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19912 PR tree-optimization/81428
19913 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
19914 can't be built for those types.
19916 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19918 Remove stuff dead since r239246.
19920 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
19921 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
19922 (avr_inform_devices): Remove dead stuff.
19924 2017-07-17 Tamar Christina <tamar.christina@arm.com>
19926 * config/arm/arm_neon.h: Fix softp typo.
19928 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19930 PR tree-optimization/81365
19931 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
19932 aggregate moves onto bb predecessor edges, make sure there are no
19933 loads that could alias the lhs in between the start of bb and the
19936 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19939 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
19940 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
19941 [LSHIFTRT, outer_code = TRUNCATE]: Same.
19943 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19945 PR tree-optimization/81396
19946 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
19947 (init_symbolic_number): Initialize it to 1.
19948 (perform_symbolic_merge): Add n_ops from both operands into the new
19950 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
19951 without base_addr as useless if they need more than one operation.
19952 (bswap_replace): Handle !bswap case for NULL base_addr.
19954 2017-07-17 Tom de Vries <tom@codesourcery.com>
19957 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
19960 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
19962 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
19963 conditional builtin define __FIX_LEON3FT_B2BST.
19965 2017-07-17 Daniel Cederman <cederman@gaisler.com>
19967 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
19968 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
19971 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19973 PR rtl-optimization/81424
19974 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
19975 to remove potential trapping from operands if -fnon-call-exceptions.
19977 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19979 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
19980 profile_proability for scalling.
19981 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
19983 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19985 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
19987 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19989 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
19990 fixpoint arithmetics.
19992 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19994 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
19995 fixpoint arithmetics.
19997 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19999 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
20000 fixpoint arithmetics.
20002 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
20004 * profile-count.h (profile_probability::from_reg_br_prob_note,
20005 profile_probability::to_reg_br_prob_note): New functions.
20006 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
20007 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
20008 * predict.c (probability_reliable_p): Update.
20009 (edge_probability_reliable_p): Update.
20010 (br_prob_note_reliable_p): Update.
20011 (invert_br_probabilities): Update.
20012 (add_reg_br_prob_note): New function.
20013 (combine_predictions_for_insn): Update.
20014 * asan.c (asan_clear_shadow): Update.
20015 * cfgbuild.c (compute_outgoing_frequencies): Update.
20016 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20017 (update_br_prob_note): Update.
20018 (rtl_verify_edges): Update.
20019 (purge_dead_edges): Update.
20020 (fixup_reorder_chain): Update.
20021 * emit-rtl.c (try_split): Update.
20022 * ifcvt.c (cond_exec_process_insns): Update.
20023 (cond_exec_process_if_block): Update.
20024 (dead_or_predicable): Update.
20025 * internal-fn.c (expand_addsub_overflow): Update.
20026 (expand_neg_overflow): Update.
20027 (expand_mul_overflow): Update.
20028 * loop-doloop.c (doloop_modify): Update.
20029 * loop-unroll.c (compare_and_jump_seq): Update.
20030 * optabs.c (emit_cmp_and_jump_insn_1): Update.
20031 * predict.h: Update.
20032 * reorg.c (mostly_true_jump): Update.
20034 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
20035 * config/alpha/alpha.c (emit_unlikely_jump): Update.
20036 * config/arc/arc.c: (emit_unlikely_jump): Update.
20037 * config/arm/arm.c: (emit_unlikely_jump): Update.
20038 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
20039 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
20040 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
20041 (ix86_print_operand): Update.
20042 (ix86_split_fp_branch): Update.
20043 (predict_jump): Update.
20044 * config/ia64/ia64.c (ia64_print_operand): Update.
20045 * config/mmix/mmix.c (mmix_print_operand): Update.
20046 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
20047 (rs6000_expand_split_stack_prologue): Update.
20048 * config/rs6000/rs6000.c: Update.
20049 * config/s390/s390.c (s390_expand_vec_strlen): Update.
20050 (s390_expand_vec_movstr): Update.
20051 (s390_expand_cs_tdsi): Update.
20052 (s390_expand_split_stack_prologue): Update.
20053 * config/sh/sh.c (sh_print_operand): Update.
20054 (expand_cbranchsi4): Update.
20055 (expand_cbranchdi4): Update.
20056 * config/sparc/sparc.c (output_v9branch): Update.
20057 * config/spu/spu.c (get_branch_target): Update.
20058 (ea_load_store_inline): Update.
20059 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20060 * config/tilepro/tilepro.c: Update.
20062 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20064 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20065 (gimplify_save_expr): Likewise.
20067 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20069 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20071 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20073 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20075 * timevar.def (TV_IPA_FNSUMMARY): Define.
20077 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20079 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20080 to back store errata sensitive sequence from being generated.
20081 (sqrtdf2_fix): Likewise.
20083 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20085 * tree-ssa-threadupdate.c (compute_path_counts,
20086 update_joiner_offpath_counts): Use profile_probability.
20088 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20091 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20093 * config/arm/arm-c.c (arm_cpu_builtins): Define
20094 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20096 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20098 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20099 array entries to represent __ieee128 versions of the
20100 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20101 scalar_extract_sig, and scalar_insert_exp built-in functions.
20102 (altivec_resolve_overloaded_builtin): Add special case handling
20103 for the __builtin_scalar_insert_exp function, as represented by
20104 the P9V_BUILTIN_VEC_VSIEDP constant.
20105 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20106 exponent support for __ieee128 argument.
20107 (VSESQP): Add scalar extract signature support for __ieee128
20109 (VSTDCNQP): Add scalar test negative support for __ieee128
20111 (VSIEQP): Add scalar insert exponent support for __int128 argument
20112 with __ieee128 result.
20113 (VSIEQPF): Add scalar insert exponent support for __ieee128
20114 argument with __ieee128 result.
20115 (VSTDCQP): Add scalar test data class support for __ieee128
20117 (VSTDCNQP): Add overload support for scalar test negative with
20118 __ieee128 argument.
20119 (VSTDCQP): Add overload support for scalar test data class
20120 __ieee128 argument.
20121 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
20122 UNSPEC_VSX_SXSIGDP.
20123 (UNSPEC_VSX_SIEXPQP): New constant.
20124 (xsxexpqp): New insn for VSX scalar extract exponent quad
20126 (xsxsigqp): New insn for VSX scalar extract significand quad
20128 (xsiexpqpf): New insn for VSX scalar insert exponent quad
20129 precision with floating point argument.
20130 (xststdcqp): New expand for VSX scalar test data class quad
20132 (xststdcnegqp): New expand for VSX scalar test negative quad
20134 (xststdcqp): New insn to match expansions for VSX scalar test data
20135 class quad precision and VSX scalar test negative quad precision.
20136 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
20137 special case operand checking to enforce that second operand of
20138 VSX scalar test data class with quad precision argument is a 7-bit
20140 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20141 prototypes and descriptions of __ieee128 versions of
20142 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
20143 scalar_test_data_class, and scalar_test_neg built-in functions.
20145 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20147 PR tree-optimization/81162
20148 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
20149 replace a negate with an add.
20151 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
20153 * doc/invoke.texi (arm/-mcpu): Document +crypto.
20155 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20157 * config/arm/arm-c.c (arm_cpu_builtins): Define
20158 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20160 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20162 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
20163 (armv8-r): Set ARM Cortex-R52 as default CPU.
20164 * config/arm/arm-tables.opt: Regenerate.
20165 * config/arm/arm-tune.md: Regenerate.
20166 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20168 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
20169 extension for -mcpu=cortex-r52.
20171 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20173 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
20174 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
20175 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
20176 (fp-armv8): Define it as FP_ARMv8 only.
20177 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
20178 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
20180 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
20181 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
20182 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
20183 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
20184 than TARGET_FPU_ARMV8.
20185 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
20186 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
20187 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
20189 * config/arm/neon.md (neon_vrint): Likewise.
20190 (neon_vcvt): Likewise.
20191 (neon_<fmaxmin_op><mode>): Likewise.
20192 (<fmaxmin><mode>3): Likewise.
20193 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
20194 * config/arm/predicates.md (arm_cond_move_operator): Check against
20195 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
20197 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
20199 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
20200 to top of function.
20202 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20204 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
20205 loop in comment with memset.
20207 2017-07-14 Martin Liska <mliska@suse.cz>
20209 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
20210 * dwarf2out.c (is_java): Remove the function.
20211 (output_pubname): Remove usage of the function.
20212 (lower_bound_default): Remove usage of DW_LANG_Java.
20213 (gen_compile_unit_die): Likewise.
20214 * gcc.c: Remove compiler defaults for .java and .zip files.
20215 * gimple-expr.c (remove_suffix): Change as there's no longer
20216 extension than 4-letter one.
20217 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
20218 (gimplify_save_expr): Likewise.
20219 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
20220 as it's possible even for other languages than Java.
20221 * langhooks.h (struct lang_hooks): Remove Java from a comment.
20222 * lto-opts.c (lto_write_options): Remove reference to Java.
20223 * opts.c (strip_off_ending): Update file extension handling.
20224 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
20225 * tree-eh.c (lower_resx): Likewise.
20226 * tree.c (free_lang_data_in_type): Remove dead code.
20227 (find_decls_types_r): Likewise.
20228 (build_common_builtin_nodes): Remove Java from a comment.
20229 (verify_type): Remove dead code.
20230 * varasm.c (assemble_external): Remove Java from a comment.
20232 2017-07-14 Martin Liska <mliska@suse.cz>
20234 * opts.c (finish_options): Add quotes.
20235 (common_handle_option): Likewise.
20237 2017-07-14 Martin Liska <mliska@suse.cz>
20239 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
20240 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
20241 Remove N_SO_PASCAL.
20242 * dwarf2out.c (lower_bound_default): Do not handle
20244 (gen_compile_unit_die): Likewise.
20245 * gcc.c: Remove default extension binding for GNU Pascal.
20246 * stmt.c: Remove Pascal language from a comment.
20247 * xcoffout.c: Likewise.
20249 2017-07-13 David Malcolm <dmalcolm@redhat.com>
20252 * diagnostic-show-locus.c (fixit_cmp): New function.
20253 (layout::layout): Sort m_fixit_hints.
20254 (column_range::column_range): Assert that the values are valid.
20255 (struct char_span): New struct.
20256 (correction::overwrite): New method.
20257 (struct source_line): New struct.
20258 (line_corrections::add_hint): Add assertions. Reimplement memcpy
20259 calls in terms of classes source_line and char_span, and
20260 correction::overwrite.
20261 (selftest::test_overlapped_fixit_printing_2): New function.
20262 (selftest::diagnostic_show_locus_c_tests): Call it.
20264 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
20266 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
20267 early if there is no lhs.
20269 2017-07-13 Martin Liska <mliska@suse.cz>
20271 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
20272 (gen_reference_type_die): Likewise.
20273 * stor-layout.c: Remove Pascal-related comment.
20275 2017-07-13 Martin Liska <mliska@suse.cz>
20277 * opts.c (finish_options): Add quotes to error messages.
20278 (parse_sanitizer_options): Likewise.
20280 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
20282 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
20284 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
20286 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
20288 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
20290 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
20292 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
20294 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
20297 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
20298 provides the hardware capability bits, define the macro
20299 __BUILTIN_CPU_SUPPORTS__.
20300 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
20301 if GLIBC does not provide the hardware capability bits. Add a
20302 gcc_unreachable call if the built-in cpu function is neither
20303 __builtin_cpu_is nor __builtin_cpu_supports.
20304 (rs6000_get_function_versions_dispatcher): Change the warning
20305 that an old GLIBC is used which does not export the capability
20306 bits to be an error.
20307 * doc/extend.texi (target_clones attribute): Document the
20308 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
20309 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
20310 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
20311 the macros defined by GCC if the newer GLIBC is available.
20313 2017-07-12 Jeff Law <law@redhat.com>
20315 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
20316 remaining includes slightly.
20317 * config/riscv/riscv-builtins.c: Include profile-count.h.
20319 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20322 * config/avr/avr.c (avr_set_current_function): In diagnostic
20323 messages: Quote keywords and (parts of) identifiers.
20324 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
20327 2017-07-12 Carl Love <cel@us.ibm.com>
20329 * config/rs6000/rs6000-c.c: Add support for built-in functions
20330 vector bool char vec_revb (vector bool char);
20331 vector bool short vec_revb (vector short char);
20332 vector bool int vec_revb (vector bool int);
20333 vector bool long long vec_revb (vector bool long long);
20334 * doc/extend.texi: Update the built-in documentation file for the
20335 new built-in functions.
20337 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20339 * config/s390/s390.md: Remove movcc splitter.
20341 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20343 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
20344 load/store on condition.
20346 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20349 * config/avr/avr.c (avr_encode_section_info)
20350 [progmem && !TREE_READONLY]: Error if progmem object needs
20353 2017-07-11 Michael Collison <michael.collison@arm.com>
20355 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
20358 2017-07-11 Carl Love <cel@us.ibm.com>
20360 * config/rs6000/rs6000-c.c: Add support for builtins
20361 vector unsigned int vec_parity_lsbb (vector signed int);
20362 vector unsigned int vec_parity_lsbb (vector unsigned int);
20363 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
20364 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
20365 vector unsigned long long vec_parity_lsbb (vector signed long long);
20366 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
20367 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
20368 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
20369 * doc/extend.texi: Update the built-in documentation file for the
20370 new built-in functions.
20372 2017-07-11 David Malcolm <dmalcolm@redhat.com>
20374 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
20375 (layout::m_primary_loc): New field.
20376 (layout::layout): Initialize new field. Move location filtering
20377 logic from here to...
20378 (layout::maybe_add_location_range): ...this new method. Add
20379 support for filtering to just the lines already specified by other
20381 (layout::will_show_line_p): New method.
20382 (gcc_rich_location::add_location_if_nearby): New method.
20383 (selftest::test_add_location_if_nearby): New test function.
20384 (selftest::diagnostic_show_locus_c_tests): Call it.
20385 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
20388 2017-07-11 Tom de Vries <tom@codesourcery.com>
20390 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
20391 (bb_first_real_insn): New function.
20392 (nvptx_single): Add extra initialization of broadcasted condition
20395 2017-07-11 Nathan Sidwell <nathan@acm.org>
20397 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
20399 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
20401 * doc/extend.texi (AVR Function Attributes): Remove weblink to
20402 Binutils doc as TEXI will mess them up.
20403 * doc/invoke.texi (AVR Options): Same here.
20405 2017-07-11 Daniel Cederman <cederman@gaisler.com>
20407 * config/sparc/sparc.opt (mfix-ut700): New option.
20408 (mfix-gr712rc): Likewise.
20409 (sparc_fix_b2bst): New variable.
20410 * doc/invoke.texi (SPARC options): Document them.
20411 (ARM options): Fix warnings.
20412 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
20413 instructions to prevent sequences that can trigger the store-store
20414 errata for certain LEON3FT processors.
20415 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
20416 (sparc_option_override): Set sparc_fix_b2bst appropriately.
20417 * config/sparc/sparc.md (fix_b2bst): New attribute.
20418 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
20420 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
20423 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
20425 (*rsqrtsf2_sse): Ditto.
20427 (div<mode>3): Macroize insn from divdf3 and divsf3
20428 using MODEF mode iterator.
20430 2017-07-10 Martin Sebor <msebor@redhat.com>
20432 PR tree-optimization/80397
20433 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
20434 instead of testing for equality to INTEGER_TYPE.
20436 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
20438 * config.gcc: Remove uclibc from arc target spec.
20440 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
20442 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
20444 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20447 * lto-wrapper.c (remove_option): New function.
20448 (merge_and_complain): Merge PIC/PIE options more realistically.
20450 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20452 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
20456 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
20457 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
20458 * config.in: Regenerate.
20459 * configure: Regenerate.
20460 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
20461 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
20462 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
20463 (TARGET_GASISR_PROLOGUES): ...target mask.
20464 * common/config/avr/avr-common.c
20465 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
20466 Set -mgas-isr-prologues.
20467 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
20468 INSERT_PASS_BEFORE for it.
20469 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
20470 * config/avr/avr.c (avr_option_override)
20471 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
20472 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
20473 (avr_attribute_table) <no_gccisr>: Add new function attribute.
20474 (avr_set_current_function) <is_no_gccisr>: Init machine field.
20475 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
20477 (make_avr_pass_pre_proep): New function.
20478 (emit_push_sfr) <treg>: Add argument to function and use it
20479 instead of TMP_REG.
20480 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
20481 and set machine->gasisr.yes.
20482 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
20483 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
20484 __gcc_isr.n_pushed to .L__stack_usage.
20485 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
20486 (avr_asm_final_postscan_insn): ...this new static function.
20487 * config/avr/avr.h (machine_function)
20488 <is_no_gccisr, use_L__stack_usage>: New fields.
20489 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
20490 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
20491 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
20492 (gasisr, *gasisr): New expander and insn.
20493 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20494 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
20495 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
20497 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
20499 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
20502 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20504 Move jump-tables out of .text again.
20507 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
20508 (ASM_OUTPUT_ADDR_VEC): New function.
20509 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
20510 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
20511 INSN_ADDRESSes as asm comment.
20512 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
20513 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
20514 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
20515 * config/avr/avr.md (*tablejump): Adjust comment.
20516 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
20517 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
20519 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
20520 (avr_output_addr_vec): New proto.
20521 (avr_log_t) <insn_addresses>: New field.
20523 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
20526 * config/i386/i386.c (ix86_function_arg_advance): Set
20527 outgoing_args_on_stack to true if there are outgoing arguments
20529 (ix86_function_arg): Likewise.
20530 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
20531 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
20532 * config/i386/i386.h (machine_function): Add
20533 outgoing_args_on_stack.
20535 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
20537 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20538 supporting pthreds.
20539 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
20541 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
20543 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
20544 (REAL_H): Remove $(MACHMODE_H).
20545 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
20547 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
20548 $(MACHMODE_H) and double-int.h.
20549 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
20551 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
20554 2017-07-07 Andrew Pinski <apinski@cavium.com>
20556 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
20557 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
20559 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20561 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
20562 Add warning if GCC was not configured to link against a GLIBC that
20563 exports the hardware capability bits.
20564 (make_resolver_func): Make resolver function private and not a
20565 COMDAT function. Create the name with clone_function_name instead
20566 of make_unique_name.
20569 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
20570 correct operand in doing the split.
20572 2017-07-07 Carl Love <cel@us.ibm.com>
20574 * config/rs6000/rs6000-c: Add support for built-in function
20575 vector unsigned short vec_pack_to_short_fp32 (vector float,
20577 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
20578 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
20579 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
20580 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
20581 (convert_4f32_8i16): Add define_expand.
20582 * doc/extend.texi: Update the built-in documentation file for the
20583 new built-in function.
20585 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20587 * config/sparc/m8.md: New file.
20588 * config/sparc/sparc.md: Include m8.md.
20590 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20592 * config/sparc/sparc.opt: New option -mvis4b.
20593 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
20594 (sparc_option_override): Handle VIS4B.
20595 (enum sparc_builtins): Define
20596 SPARC_BUILTIN_DICTUNPACK{8,16,32},
20597 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
20598 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
20599 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
20600 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
20601 (check_constant_argument): New function.
20602 (sparc_vis_init_builtins): Define builtins
20603 __builtin_vis_dictunpack{8,16,32},
20604 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
20605 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
20606 __builtin_vis_fpcmpde{8,16,32}shl and
20607 __builtin_vis_fpcmpur{8,16,32}shl.
20608 (sparc_expand_builtin): Check that the constant operands to
20609 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
20610 constant and in range.
20611 * config/sparc/sparc-c.c (sparc_target_macros): Handle
20613 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
20614 (SPARC_IMM5_P): Likewise.
20615 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
20616 (enabled): Handle vis4b.
20617 (UNSPEC_DICTUNPACK): New unspec.
20618 (UNSPEC_FPCMPSHL): Likewise.
20619 (UNSPEC_FPUCMPSHL): Likewise.
20620 (UNSPEC_FPCMPDESHL): Likewise.
20621 (UNSPEC_FPCMPURSHL): Likewise.
20622 (cpu_feature): New CPU feature `vis4b'.
20623 (dictunpack{8,16,32}): New insns.
20624 (FPCSMODE): New mode iterator.
20625 (fpcscond): New code iterator.
20626 (fpcsucond): Likewise.
20627 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
20628 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
20629 (fpcmpde{8,16,32}{si,di}shl): Likewise.
20630 (fpcmpur{8,16,32}{si,di}shl): Likewise.
20631 * config/sparc/constraints.md: Define constraints `q' for unsigned
20632 2-bit integer constants and `t' for unsigned 5-bit integer
20634 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
20636 (imm5_operand_dictunpack16): Likewise.
20637 (imm5_operand_dictunpack32): Likewise.
20638 (imm2_operand): Likewise.
20639 * doc/invoke.texi (SPARC Options): Document -mvis4b.
20640 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
20641 ditunpack* and fpcmp*shl builtins.
20643 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20645 * config.gcc: Handle m8 in --with-{cpu,tune} options.
20646 * config.in: Add HAVE_AS_SPARC6 define.
20647 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
20649 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
20651 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
20652 (CPP_CPU_SPEC): Handle m8.
20653 (ASM_CPU_SPEC): Likewise.
20654 * config/sparc/sparc-opts.h (enum processor_type): Add
20656 * config/sparc/sparc.c (m8_costs): New struct.
20657 (sparc_option_override): Handle TARGET_CPU_m8.
20658 (sparc32_initialize_trampoline): Likewise.
20659 (sparc64_initialize_trampoline): Likewise.
20660 (sparc_issue_rate): Likewise.
20661 (sparc_register_move_cost): Likewise.
20662 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
20663 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
20664 (ASM_CPU64_DEFAULT_SPEC): Likewise.
20665 (CPP_CPU_SPEC): Handle M8.
20666 (ASM_CPU_SPEC): Likewise.
20667 (AS_M8_FLAG): Define.
20668 * config/sparc/sparc.md: Add m8 to the cpu attribute.
20669 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
20670 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
20672 * configure: Regenerate.
20673 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
20676 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20678 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
20680 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
20681 ("*movdi_insn_sp32"): Do not set v3pipe.
20682 ("*movsi_insn"): Likewise.
20683 ("*movdi_insn_sp64"): Likewise.
20684 ("*movsf_insn"): Likewise.
20685 ("*movdf_insn_sp32"): Likewise.
20686 ("*movdf_insn_sp64"): Likewise.
20687 ("*zero_extendsidi2_insn_sp64"): Likewise.
20688 ("*sign_extendsidi2_insn"): Likewise.
20689 ("*mov<VM32:mode>_insn"): Likewise.
20690 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20691 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20692 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20693 ("<vlop:code><VL:mode>3"): Likewise.
20694 ("*not_<vlop:code><VL:mode>3"): Likewise.
20695 ("*nand<VL:mode>_vis"): Likewise.
20696 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
20697 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
20698 ("one_cmpl<VL:mode>2"): Likewise.
20699 ("faligndata<VM64:mode>_vis"): Likewise.
20700 ("alignaddrsi_vis"): Likewise.
20701 ("alignaddrdi_vis"): Likweise.
20702 ("alignaddrlsi_vis"): Likewise.
20703 ("alignaddrldi_vis"): Likewise.
20704 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20705 ("bmaskdi_vis"): Likewise.
20706 ("bmasksi_vis"): Likewise.
20707 ("bshuffle<VM64:mode>_vis"): Likewise.
20708 ("cmask8<P:mode>_vis"): Likewise.
20709 ("cmask16<P:mode>_vis"): Likewise.
20710 ("cmask32<P:mode>_vis"): Likewise.
20711 ("pdistn<P:mode>_vis"): Likewise.
20712 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20714 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20716 * config/sparc/sparc.md ("subtype"): New insn attribute.
20717 ("*wrgsr_sp64"): Set insn subtype.
20718 ("*rdgsr_sp64"): Likewise.
20719 ("alignaddrsi_vis"): Likewise.
20720 ("alignaddrdi_vis"): Likewise.
20721 ("alignaddrlsi_vis"): Likewise.
20722 ("alignaddrldi_vis"): Likewise.
20723 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20724 ("fexpand_vis"): Likewise.
20725 ("fpmerge_vis"): Likewise.
20726 ("faligndata<VM64:mode>_vis"): Likewise.
20727 ("bshuffle<VM64:mode>_vis"): Likewise.
20728 ("cmask8<P:mode>_vis"): Likewise.
20729 ("cmask16<P:mode>_vis"): Likewise.
20730 ("cmask32<P:mode>_vis"): Likewise.
20731 ("fchksm16_vis"): Likewise.
20732 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
20733 ("fmean16_vis"): Likewise.
20734 ("fp<plusminus_insn>64_vis"): Likewise.
20735 ("<plusminus_insn>v8qi3"): Likewise.
20736 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20737 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
20738 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
20739 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
20740 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
20741 ("*movqi_insn"): Likewise.
20742 ("*movhi_insn"): Likewise.
20743 ("*movsi_insn"): Likewise.
20744 ("movsi_pic_gotdata_op"): Likewise.
20745 ("*movdi_insn_sp32"): Likewise.
20746 ("*movdi_insn_sp64"): Likewise.
20747 ("movdi_pic_gotdata_op"): Likewise.
20748 ("*movsf_insn"): Likewise.
20749 ("*movdf_insn_sp32"): Likewise.
20750 ("*movdf_insn_sp64"): Likewise.
20751 ("*zero_extendhisi2_insn"): Likewise.
20752 ("*zero_extendqihi2_insn"): Likewise.
20753 ("*zero_extendqisi2_insn"): Likewise.
20754 ("*zero_extendqidi2_insn"): Likewise.
20755 ("*zero_extendhidi2_insn"): Likewise.
20756 ("*zero_extendsidi2_insn_sp64"): Likewise.
20757 ("ldfsr"): Likewise.
20758 ("prefetch_64"): Likewise.
20759 ("prefetch_32"): Likewise.
20760 ("tie_ld32"): Likewise.
20761 ("tie_ld64"): Likewise.
20762 ("*tldo_ldub_sp32"): Likewise.
20763 ("*tldo_ldub1_sp32"): Likewise.
20764 ("*tldo_ldub2_sp32"): Likewise.
20765 ("*tldo_ldub_sp64"): Likewise.
20766 ("*tldo_ldub1_sp64"): Likewise.
20767 ("*tldo_ldub2_sp64"): Likewise.
20768 ("*tldo_ldub3_sp64"): Likewise.
20769 ("*tldo_lduh_sp32"): Likewise.
20770 ("*tldo_lduh1_sp32"): Likewise.
20771 ("*tldo_lduh_sp64"): Likewise.
20772 ("*tldo_lduh1_sp64"): Likewise.
20773 ("*tldo_lduh2_sp64"): Likewise.
20774 ("*tldo_lduw_sp32"): Likewise.
20775 ("*tldo_lduw_sp64"): Likewise.
20776 ("*tldo_lduw1_sp64"): Likewise.
20777 ("*tldo_ldx_sp64"): Likewise.
20778 ("*mov<VM32:mode>_insn"): Likewise.
20779 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20780 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20782 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20784 * config/sparc/sparc.md ("type"): New insn type viscmp.
20785 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
20787 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
20788 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
20789 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20790 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
20792 ("n7_vis_logical_11cycle"): Likewise.
20793 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
20794 * config/sparc/niagara2.md ("niag3_vis": Likewise.
20795 * config/sparc/niagara.md ("niag_vis"): Likewise.
20796 * config/sparc/ultra3.md ("us3_fga"): Likewise.
20797 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
20799 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20801 * config/sparc/sparc.md: New instruction type `bmask'.
20802 (bmaskdi_vis): Use the `bmask' type.
20803 (bmasksi_vis): Likewise.
20804 * config/sparc/ultra3.md (us3_array): Likewise.
20805 * config/sparc/niagara7.md (n7_array): Likewise.
20806 * config/sparc/niagara4.md (n4_array): Likewise.
20807 * config/sparc/niagara2.md (niag2_vis): Likewise.
20808 (niag3_vis): Likewise.
20809 * config/sparc/niagara.md (niag_vis): Likewise.
20811 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20813 * ipa-comdats.c: Remove optimize check from gate.
20814 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
20815 for functions not optimized.
20816 (ipa_fn_summary_read): Skip optimize check.
20817 (ipa_fn_summary_write): Likewise.
20818 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
20820 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
20822 (can_inline_edge_p): Check flag_pcc_struct_return for match.
20823 (check_callers): Give up on caller which is not optimized.
20824 (inline_small_functions): Likewise.
20825 (ipa_inline): Do not give up when not optimizing.
20826 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
20827 away unoptimizes cdtors.
20828 (whole_program_function_and_variable_visibility): Do
20829 ipa_discover_readonly_nonaddressable_vars in LTO mode.
20830 * ipa.c (process_references): Do not check optimize.
20831 (symbol_table::remove_unreachable_nodes): Update optimize check.
20832 (set_writeonly_bit): Update optimize check.
20833 (pass_ipa_cdtor_merge::gate): Do not check optimize.
20834 (pass_ipa_single_use::gate): Remove.
20836 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20838 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
20839 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
20840 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
20841 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
20842 permute_load, permute_store, adjust_extract, adjust_splat,
20843 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
20844 replace_swap_with_copy, dump_swap_insn_table,
20845 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
20846 recombine_lvx_pattern, recombine_stvx_pattern,
20847 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
20848 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
20849 to file rs6000-p8swap.c.
20850 * config/rs6000/rs6000-p8swap.c: New file.
20851 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
20852 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
20853 and rs6000*-*-* targets.
20855 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20857 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
20859 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20861 * lto-wrapper.c (merge_and_complain): Do not merge
20862 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
20863 fsigned_zeros, ftrapping_math, fwrapv.
20864 (append_compiler_options): Do not track these options.
20865 (append_linker_options): Likewie
20867 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20869 * cgraphunit.c (cgraph_node::finalize_function): When
20870 !flag_toplevel_reorde set no_reorder flag.
20871 (varpool_node::finalize_decl): Likewise.
20872 (symbol_table::compile): Drop no toplevel reorder path.
20874 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20876 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
20877 edges; zero probability is not better than uninitialized.
20879 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20881 * asan.h (asan_sanitize_allocas_p): Declare.
20882 * asan.c (asan_sanitize_allocas_p): New function.
20883 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
20884 (handle_builtin_alloca): Likewise.
20885 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
20886 if !asan_sanitize_allocas_p.
20887 * params.def (asan-instrument-allocas): Add new option.
20888 * params.h (ASAN_PROTECT_ALLOCAS): Define.
20889 * opts.c (common_handle_option): Disable allocas sanitization for
20892 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20894 * asan.c: Include gimple-fold.h.
20895 (get_last_alloca_addr): New function.
20896 (handle_builtin_stackrestore): Likewise.
20897 (handle_builtin_alloca): Likewise.
20898 (asan_emit_allocas_unpoison): Likewise.
20899 (get_mem_refs_of_builtin_call): Add new parameter, remove const
20900 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
20901 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
20902 (instrument_builtin_call): Pass gimple iterator to
20903 get_mem_refs_of_builtin_call.
20904 (last_alloca_addr): New global.
20905 * asan.h (asan_emit_allocas_unpoison): Declare.
20906 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
20907 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
20908 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
20909 if function calls alloca.
20910 * gimple-fold.c (replace_call_with_value): Remove static keyword.
20911 * gimple-fold.h (replace_call_with_value): Declare.
20912 * internal-fn.c: Include asan.h.
20913 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
20914 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
20916 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20918 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
20919 (C_SELFTEST_FLAGS): New.
20920 (CPP_SELFTEST_FLAGS): New.
20921 (SELFTEST_DEPS): New, from deps of s-selftest.
20922 (C_SELFTEST_DEPS): New, from deps of s-selftest.
20923 (CPP_SELFTEST_DEPS): New.
20924 (selftest): Add dependency on s-selftest-c++.
20925 (s-selftest): Rename to...
20926 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
20927 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
20928 than SELFTEST_FLAGS.
20929 (selftest-gdb): Rename to...
20930 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
20932 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
20933 (selftest-valgrind): Rename to...
20934 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
20936 (selftest-valgrind): Reintroduce as an alias for
20937 selftest-c-valgrind.
20938 (s-selftest-c++): New.
20939 (selftest-c++-gdb): New.
20940 (selftest-c++-valgrind): New.
20942 2017-07-06 Olivier Hainque <hainque@adacore.com>
20944 * gcc.c (process_command): When deciding if undefined variables
20945 should be ignored when processing specs, accept "gcc -v" as well.
20947 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20949 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
20950 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
20952 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20954 * config/arm/arm-cpus.in (armv8-r): Add new entry.
20955 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
20956 * config/arm/arm-tables.opt: Regenerate.
20957 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
20959 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
20961 2017-07-06 Carl Love <cel@us.ibm.com>
20963 * ChangeLog: Clean up from mid air collision
20965 2017-07-06 Carl Love <cel@us.ibm.com>
20967 * config/rs6000/rs6000-c.c: Add support for built-in functions
20968 vector signed int vec_subc (vector signed int, vector signed int);
20969 vector signed __int128 vec_subc (vector signed __int128,
20970 vector signed __int128);
20971 vector unsigned __int128 vec_subc (vector unsigned __int128,
20972 vector unsigned __int128);
20973 vector signed int vec_sube (vector signed int, vector signed int,
20974 vector signed int);
20975 vector unsigned int vec_sube (vector unsigned int,
20976 vector unsigned int,
20977 vector unsigned int);
20978 vector signed __int128 vec_sube (vector signed __int128,
20979 vector signed __int128,
20980 vector signed__int128);
20981 vector unsigned __int128 vec_sube (vector unsigned __int128,
20982 vector unsigned __int128,
20983 vector unsigned __int128);
20984 vector signed int vec_subec (vector signed int, vector signed int,
20985 vector signed int);
20986 vector unsigned int vec_subec (vector unsigned int,
20987 vector unsigned int,
20988 vector unsigned int);
20989 vector signed __int128 vec_subec (vector signed __int128,
20990 vector signed __int128,
20991 vector signed__int128);
20992 vector unsigned __int128 vec_subec (vector unsigned __int128,
20993 vector unsigned __int128,
20994 vector unsigned __int128);
20995 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
20996 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
20997 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
20998 BU_ALTIVEC_OVERLOAD_X definitions.
20999 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
21000 * doc/extend.texi: Update the built-in documentation file for the new
21001 built-in functions.
21003 2017-07-06 David Malcolm <dmalcolm@redhat.com>
21006 * diagnostic-show-locus.c (layout::layout): Use start and finish
21007 spelling location for the start and finish of each range.
21008 * genmatch.c (linemap_client_expand_location_to_spelling_point):
21009 Add unused aspect param.
21010 * input.c (expand_location_1): Add "aspect" param, and use it
21011 to access the correct part of the location.
21012 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
21014 (expand_location_to_spelling_point): Likewise.
21015 (linemap_client_expand_location_to_spelling_point): Add "aspect"
21016 param, and pass it to expand_location_1.
21018 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
21020 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
21021 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
21022 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
21023 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
21024 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
21025 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
21026 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
21027 _mm_maskz_getmant_ss): New intrinsics.
21028 (__builtin_ia32_getexpss128_mask): Changed to ...
21029 __builtin_ia32_getexpss128_round ... this.
21030 (__builtin_ia32_getexpsd128_mask): Changed to ...
21031 __builtin_ia32_getexpsd128_round ... this.
21032 * config/i386/i386-builtin-types.def
21033 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
21034 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
21035 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
21036 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
21037 __builtin_ia32_getmantss_mask_round): New builtins.
21038 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
21039 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
21040 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
21041 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
21042 * config/i386/sse.md
21043 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
21044 avx512f_sgetexp<mode><mask_scalar_name>
21045 <round_saeonly_scalar_name> ... this.
21046 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
21047 %0, %1, %2<round_saeonly_op3>}): Changed to ...
21048 vgetexp<ssescalarmodesuffix>
21049 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21050 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
21051 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
21052 avx512f_vgetmant<mode><mask_scalar_name>
21053 <round_saeonly_scalar_name> ... this.
21054 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21055 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21056 vgetmant<ssescalarmodesuffix>
21057 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21058 %0<mask_scalar_operand4>, %1, %2
21059 <round_saeonly_scalar_mask_op4>, %3} ... this.
21060 * config/i386/subst.md (mask_scalar_operand4,
21061 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21062 round_saeonly_scalar_nimm_predicate): New subst attributes.
21064 2017-07-06 Julia Koval <julia.koval@intel.com>
21066 * config/i386/i386.c (ix86_erase_embedded_rounding):
21067 Remove code for old rounding pattern.
21069 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21071 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21073 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21075 * doc/sourcebuild.texi (Test Directives, Variants of
21076 dg-require-support): Add documentation for dg-require-stack-check.
21078 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21080 * config/i386/subst.md (mask_scalar, round_scalar,
21081 round_saeonly_scalar): New meta-templates.
21082 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21083 round_scalar_mask_operand3, round_scalar_mask_op3,
21084 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21085 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21086 round_saeonly_scalar_constraint,
21087 round_saeonly_scalar_prefix): New subst attribute.
21088 * config/i386/sse.md
21089 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21090 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21091 <round_scalar_name> ... this.
21092 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21093 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21094 <round_scalar_name> ... this.
21095 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21096 <sse>_vm<code><mode>3<mask_scalar_name>
21097 <round_saeonly_scalar_name> ... this.
21098 (v<plusminus_mnemonic><ssescalarmodesuffix>
21099 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21100 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21101 v<plusminus_mnemonic><ssescalarmodesuffix>
21102 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21103 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21104 (v<multdiv_mnemonic><ssescalarmodesuffix>
21105 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21106 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21107 v<multdiv_mnemonic><ssescalarmodesuffix>
21108 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21109 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21110 (v<maxmin_float><ssescalarmodesuffix>
21111 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21112 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
21113 v<maxmin_float><ssescalarmodesuffix>
21114 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21115 %0<mask_scalar_operand3>, %1, %<iptr>2
21116 <round_saeonly_scalar_mask_op3>} ... this.
21118 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
21120 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
21121 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
21123 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
21124 Alan Hayward <alan.hayward@arm.com>
21125 David Sherwood <david.sherwood@arm.com>
21127 * combine.c (simplify_if_then_else): Remove "enum" before
21129 * compare-elim.c (can_eliminate_compare): Likewise.
21130 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
21132 (aarch64_lookup_simd_builtin_type): Likewise.
21133 (aarch64_simd_builtin_type): Likewise.
21134 (aarch64_init_simd_builtin_types): Likewise.
21135 (aarch64_simd_expand_args): Likewise.
21136 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
21138 (aarch64_reverse_mask): Likewise.
21139 (aarch64_simd_emit_reg_reg_move): Likewise.
21140 (aarch64_gen_adjusted_ldpstp): Likewise.
21141 (aarch64_ccmp_mode_to_code): Likewise.
21142 (aarch64_operands_ok_for_ldpstp): Likewise.
21143 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21144 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
21146 (aarch64_min_divisions_for_recip_mul): Likewise.
21147 (aarch64_reassociation_width): Likewise.
21148 (aarch64_get_condition_code_1): Likewise.
21149 (aarch64_simd_emit_reg_reg_move): Likewise.
21150 (aarch64_simd_attr_length_rglist): Likewise.
21151 (aarch64_reverse_mask): Likewise.
21152 (aarch64_operands_ok_for_ldpstp): Likewise.
21153 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21154 (aarch64_gen_adjusted_ldpstp): Likewise.
21155 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
21157 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
21158 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
21159 (arm_lookup_simd_builtin_type): Likewise.
21160 (arm_simd_builtin_type): Likewise.
21161 (arm_init_simd_builtin_types): Likewise.
21162 (arm_expand_builtin_args): Likewise.
21163 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
21164 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
21165 (ft32_setup_incoming_varargs): Likewise.
21166 (ft32_function_arg): Likewise.
21167 (ft32_function_arg_advance): Likewise.
21168 (ft32_pass_by_reference): Likewise.
21169 (ft32_arg_partial_bytes): Likewise.
21170 (ft32_valid_pointer_mode): Likewise.
21171 (ft32_addr_space_pointer_mode): Likewise.
21172 (ft32_addr_space_legitimate_address_p): Likewise.
21173 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
21175 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
21176 (ix86_emit_outlined_ms2sysv_restore): Likewise.
21177 (iamcu_alignment): Likewise.
21178 (canonicalize_vector_int_perm): Likewise.
21179 (ix86_noce_conversion_profitable_p): Likewise.
21180 (ix86_mpx_bound_mode): Likewise.
21181 (ix86_operands_ok_for_move_multiple): Likewise.
21182 * config/microblaze/microblaze-protos.h
21183 (microblaze_expand_conditional_branch_reg): Likewise.
21184 * config/microblaze/microblaze.c
21185 (microblaze_expand_conditional_branch_reg): Likewise.
21186 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
21188 (rs6000_reassociation_width): Likewise.
21189 (rs6000_invalid_binary_op): Likewise.
21190 (fusion_p9_p): Likewise.
21191 (emit_fusion_p9_load): Likewise.
21192 (emit_fusion_p9_store): Likewise.
21193 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
21195 (riscv_hard_regno_mode_ok_p): Likewise.
21196 (riscv_address_insns): Likewise.
21197 (riscv_split_symbol): Likewise.
21198 (riscv_legitimize_move): Likewise.
21199 (riscv_function_value): Likewise.
21200 (riscv_hard_regno_nregs): Likewise.
21201 (riscv_expand_builtin): Likewise.
21202 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
21203 (riscv_build_integer): Likewise.
21204 (riscv_split_integer): Likewise.
21205 (riscv_legitimate_constant_p): Likewise.
21206 (riscv_cannot_force_const_mem): Likewise.
21207 (riscv_regno_mode_ok_for_base_p): Likewise.
21208 (riscv_valid_base_register_p): Likewise.
21209 (riscv_valid_offset_p): Likewise.
21210 (riscv_valid_lo_sum_p): Likewise.
21211 (riscv_classify_address): Likewise.
21212 (riscv_legitimate_address_p): Likewise.
21213 (riscv_address_insns): Likewise.
21214 (riscv_load_store_insns): Likewise.
21215 (riscv_force_binary): Likewise.
21216 (riscv_split_symbol): Likewise.
21217 (riscv_force_address): Likewise.
21218 (riscv_legitimize_address): Likewise.
21219 (riscv_move_integer): Likewise.
21220 (riscv_legitimize_const_move): Likewise.
21221 (riscv_legitimize_move): Likewise.
21222 (riscv_address_cost): Likewise.
21223 (riscv_subword): Likewise.
21224 (riscv_output_move): Likewise.
21225 (riscv_canonicalize_int_order_test): Likewise.
21226 (riscv_emit_int_order_test): Likewise.
21227 (riscv_function_arg_boundary): Likewise.
21228 (riscv_pass_mode_in_fpr_p): Likewise.
21229 (riscv_pass_fpr_single): Likewise.
21230 (riscv_pass_fpr_pair): Likewise.
21231 (riscv_get_arg_info): Likewise.
21232 (riscv_function_arg): Likewise.
21233 (riscv_function_arg_advance): Likewise.
21234 (riscv_arg_partial_bytes): Likewise.
21235 (riscv_function_value): Likewise.
21236 (riscv_pass_by_reference): Likewise.
21237 (riscv_setup_incoming_varargs): Likewise.
21238 (riscv_print_operand): Likewise.
21239 (riscv_elf_select_rtx_section): Likewise.
21240 (riscv_save_restore_reg): Likewise.
21241 (riscv_for_each_saved_reg): Likewise.
21242 (riscv_register_move_cost): Likewise.
21243 (riscv_hard_regno_mode_ok_p): Likewise.
21244 (riscv_hard_regno_nregs): Likewise.
21245 (riscv_class_max_nregs): Likewise.
21246 (riscv_memory_move_cost): Likewise.
21247 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
21248 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
21249 (rl78_addr_space_address_mode): Likewise.
21250 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21252 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
21253 (rs6000_reassociation_width): Likewise.
21254 (rs6000_invalid_binary_op): Likewise.
21255 (fusion_p9_p): Likewise.
21256 (emit_fusion_p9_load): Likewise.
21257 (emit_fusion_p9_store): Likewise.
21258 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
21259 (ok_for_simple_move_operands): Likewise.
21260 (ok_for_simple_move_strict_operands): Likewise.
21261 (ok_for_simple_arith_logic_operands): Likewise.
21262 (visium_legitimize_reload_address): Likewise.
21263 (visium_select_cc_mode): Likewise.
21264 (output_cbranch): Likewise.
21265 (visium_split_double_move): Likewise.
21266 (visium_expand_copysign): Likewise.
21267 (visium_expand_int_cstore): Likewise.
21268 (visium_expand_fp_cstore): Likewise.
21269 * config/visium/visium.c (visium_pass_by_reference): Likewise.
21270 (visium_function_arg): Likewise.
21271 (visium_function_arg_advance): Likewise.
21272 (visium_libcall_value): Likewise.
21273 (visium_setup_incoming_varargs): Likewise.
21274 (visium_legitimate_constant_p): Likewise.
21275 (visium_legitimate_address_p): Likewise.
21276 (visium_legitimize_address): Likewise.
21277 (visium_secondary_reload): Likewise.
21278 (visium_register_move_cost): Likewise.
21279 (visium_memory_move_cost): Likewise.
21280 (prepare_move_operands): Likewise.
21281 (ok_for_simple_move_operands): Likewise.
21282 (ok_for_simple_move_strict_operands): Likewise.
21283 (ok_for_simple_arith_logic_operands): Likewise.
21284 (visium_function_value_1): Likewise.
21285 (rtx_ok_for_offset_p): Likewise.
21286 (visium_legitimize_reload_address): Likewise.
21287 (visium_split_double_move): Likewise.
21288 (visium_expand_copysign): Likewise.
21289 (visium_expand_int_cstore): Likewise.
21290 (visium_expand_fp_cstore): Likewise.
21291 (visium_split_cstore): Likewise.
21292 (visium_select_cc_mode): Likewise.
21293 (visium_split_cbranch): Likewise.
21294 (output_cbranch): Likewise.
21295 (visium_print_operand_address): Likewise.
21296 * expmed.c (flip_storage_order): Likewise.
21297 * expmed.h (emit_cstore): Likewise.
21298 (flip_storage_order): Likewise.
21299 * genrecog.c (validate_pattern): Likewise.
21300 * hsa-gen.c (gen_hsa_addr): Likewise.
21301 * internal-fn.c (expand_arith_overflow): Likewise.
21302 * ira-color.c (allocno_copy_cost_saving): Likewise.
21303 * lra-assigns.c (find_hard_regno_for_1): Likewise.
21304 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
21305 (process_invariant_for_inheritance): Likewise.
21306 * lra-eliminations.c (move_plus_up): Likewise.
21307 * omp-low.c (lower_oacc_reductions): Likewise.
21308 * simplify-rtx.c (simplify_subreg): Likewise.
21309 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
21310 (TARGET_CHKP_BOUND_MODE): Likewise..
21311 * targhooks.c (default_chkp_bound_mode): Likewise.
21312 (default_setup_incoming_vararg_bounds): Likewise.
21313 * targhooks.h (default_chkp_bound_mode): Likewise.
21314 (default_setup_incoming_vararg_bounds): Likewise.
21315 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
21316 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
21317 (have_whole_vector_shift): Likewise.
21318 * tree-vect-stmts.c (vectorizable_load): Likewise.
21319 * doc/tm.texi: Regenerate.
21321 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21323 Graceful degrade if Binutils PR21472 is not available.
21326 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
21327 .rodata in flash test fails.
21328 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
21329 * confgure: Regenerate.
21330 * config.in: Regenerate.
21331 * config/avr/avr.c (avr_asm_named_section)
21332 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
21333 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
21334 (avr_asm_init_sections): Same.
21336 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21338 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
21339 (fma<VH:mode>4_intrinsic): Likewise.
21340 (*fmsub<VCVTF:mode>4): Likewise.
21341 (*fmsub<VH:mode>4_intrinsic): Likewise.
21343 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21346 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
21347 Don't depend on "optimize > 0".
21348 (out_movhi_r_mr, out_movqi_mr_r): Same.
21349 (out_movhi_mr_r, out_movqi_r_mr): Same.
21350 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
21351 io_address_operand on "optimize > 0".
21353 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21355 * tree-loop-distribution.c: Add general explanantion on the pass.
21356 (generate_loops_for_partition): Mark distributed loop.
21357 (pg_add_dependence_edges): New parameter. Handle alias data
21358 dependence specially and record it in the parameter if asked.
21359 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
21360 (init_partition_graph_vertices, add_partition_graph_edge): New.
21361 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
21362 (free_partition_graph_vdata, build_partition_graph): New.
21363 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
21364 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
21365 (data_ref_segment_size, latch_dominated_by_data_ref): New.
21366 (compute_alias_check_pairs, version_loop_by_alias_check): New.
21367 (version_for_distribution_p, finalize_partitions): New.
21368 (distribute_loop): Handle alias data dependence specially. Factor
21369 out loop fusion code as functions and call these functions.
21371 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21373 * tree-loop-distribution.c (classify_partition): New parameter and
21374 better handle reduction statement.
21375 (rdg_build_partitions): Revise comment.
21376 (distribute_loop): Compute statements in all partitions and pass it
21377 to classify_partition.
21379 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21381 * tree-loop-distribution.c (enum partition_type): New.
21382 (struct partition): New field type.
21383 (partition_merge_into): Add parameter. Update partition type.
21384 (data_dep_in_cycle_p, update_type_for_merge): New functions.
21385 (build_rdg_partition_for_vertex): Compute partition type.
21386 (rdg_build_partitions): Dump partition type.
21387 (distribute_loop): Update calls to partition_merge_into.
21389 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21391 * tree-loop-distribution.c (struct ddr_hasher): New.
21392 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
21394 (classify_partition): Call get_data_dependence.
21395 (pg_add_dependence_edges): Ditto.
21396 (distribute_loop): Release data dependence hash table.
21398 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21400 * tree-loop-distribution.c (ref_base_address): Delete.
21401 (similar_memory_accesses): Rename ...
21402 (share_memory_accesses): ... to this. Check if partitions access
21403 the same memory reference.
21404 (distribute_loop): Call share_memory_accesses.
21406 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21408 * tree-loop-distribution.c (struct partition): New field recording
21409 its data reference.
21410 (partition_alloc, partition_free): Init and release data refs.
21411 (partition_merge_into): Merge data refs.
21412 (build_rdg_partition_for_vertex): Collect data refs for partition.
21413 (pg_add_dependence_edges): Change parameters from vector to bitmap.
21415 (distribute_loop): Remve data refs from vertice data of partition
21418 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21420 * tree-loop-distribution.c (params.h): Include header file.
21421 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
21422 (datarefs_vec): New global var.
21423 (create_rdg_vertices): Use datarefs_vec directly.
21424 (free_rdg): Don't free data references.
21425 (build_rdg): Update use. Don't free data references.
21426 (distribute_loop): Compute global variable for data references.
21427 Bail out if there are too many data references.
21429 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21431 * tree-loop-distribution.c (loop_nest): New global var.
21432 (build_rdg): Use loop directly, rather than loop nest.
21433 (pg_add_dependence_edges): Remove loop nest parameter. Use global
21435 (distribute_loop): Compute global variable loop nest. Update use.
21437 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21439 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
21440 (partition_merge_into): New parameter. Dump reason for fusion.
21441 (distribute_loop): Update use of partition_merge_into.
21443 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21445 * tree-loop-distribution.c (bb_top_order_index): New.
21446 (bb_top_order_index_size, bb_top_order_cmp): New.
21447 (stmts_from_loop): Use topological order.
21448 (pass_loop_distribution::execute): Compute and release topological
21449 order for basic blocks.
21451 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21453 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
21456 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21458 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
21459 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
21460 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
21461 * internal-fn.def (LOOP_DIST_ALIAS): New.
21462 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
21463 (fold_loop_internal_call): ... this.
21464 (vect_loop_dist_alias_call): New function.
21465 (set_uid_loop_bbs): Call fold_loop_internal_call.
21466 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
21469 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21472 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
21473 Require dead FLAGS_REG at the beginning of a peephole.
21475 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21478 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
21479 arguments in the call to __builtin_ia32_sbb_u32.
21480 (_subborrow_u64): Swap _X and _Y arguments in the call to
21481 __builtin_ia32_sbb_u64.
21483 2017-07-04 Jakub Jelinek <jakub@redhat.com>
21486 * tree-vrp.c (compare_assert_loc): Turn into a function template
21487 with stable template parameter. Only test if a->e is NULL,
21488 !a->e == !b->e has been verified already. Use e == NULL or
21489 e != NULL instead of e or ! e tests. If stable is true, don't use
21490 iterative_hash_expr, on the other side allow a or b or both NULL
21491 and sort the NULLs last.
21492 (process_assert_insertions): Sort using compare_assert_loc<false>
21493 instead of compare_assert_loc, later sort using
21494 compare_assert_loc<true> before calling process_assert_insertions_for
21495 in a loop. Use break instead of continue once seen NULL pointer.
21497 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21499 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21500 Cortex-R7 and Cortex-R8 processors.
21502 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21504 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
21505 uninitialized while src is not.
21507 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
21509 * common/config/arm/arm-common.c: Adjust include path for
21511 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
21512 (arm-cpu.h): Create in build directory. Adjust dependency rules.
21513 (arm-cpu-data.h): Likewise.
21514 (arm-cpu-cdata.h): Likewise.
21515 * config/arm/arm-cpu.h: Delete.
21516 * config/arm/arm-cpu-cdata.h: Delete.
21517 * config/arm/arm-cpu-data.h: Delete.
21519 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
21521 * config/arm/arm-cpus.in (cortex-a55): New.
21522 (cortex-a75): Likewise.
21523 (cortex-a75.cortex-a55): Likewise.
21524 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
21526 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
21527 * config/arm/arm-cpu-cdata.h: Regenerate.
21528 * config/arm/arm-cpu-data.h: Regenerate.
21529 * config/arm/arm-cpu.h: Regenerate.
21530 * config/arm/arm-tables.opt: Regenerate.
21531 * config/arm/arm-tune.md: Regenerate.
21533 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21535 * haifa-sched.c (sched_create_recovery_edges): Update profile.
21537 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21539 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
21542 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
21544 PR tree-optimization/81292
21545 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
21546 full_string_p, also call adjust_related_strinfos if the adjustment
21547 is simple, otherwise invalidate related strinfos.
21549 2017-07-04 Martin Liska <mliska@suse.cz>
21552 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
21553 newly created variable as DECL_IGNORED_P.
21555 2017-07-04 Martin Liska <mliska@suse.cz>
21558 * ipa-inline.c (inline_small_functions):
21559 Use xstrdup_for_dump.
21561 2017-07-04 Tom de Vries <tom@codesourcery.com>
21563 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
21565 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21568 * config/darwin.c (darwin_function_switched_text_sections):
21571 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
21573 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
21575 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21577 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
21579 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21581 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
21582 min_profitable_iters, and th as inclusive lower bounds.
21583 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
21584 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
21585 for min_profitable_iters and min_profitable_estimate.
21586 (vect_transform_loop): Treat th as an inclusive lower bound.
21587 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
21589 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21592 * config/darwin.c (darwin_function_switched_text_sections):
21593 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
21594 in two pieces, and suppress the use of buf.
21596 2017-07-03 Nathan Sidwell <nathan@acm.org>
21598 * hash-table.h (hash_table_mod1): Fix indentation.
21600 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21602 PR middle-end/81290
21603 * predict.c (force_edge_cold): Be more careful about propagation
21605 * profile-count.h (profile_probability::guessed,
21606 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
21608 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
21610 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
21612 * doc/invoke.texi (rcpc architecture extension): Document it.
21614 2017-07-03 Richard Biener <rguenther@suse.de>
21616 PR tree-optimization/60510
21617 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
21618 the scalar reduction PHI and use it.
21619 (vectorizable_reduction): Properly guard the single_defuse_cycle
21620 path for non-SLP reduction chains where we cannot use it.
21621 Rework reduc_def/index and vector type deduction. Rework
21622 vector operand gathering during reduction op code-gen.
21623 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
21624 chains dissolve the chain and leave it to non-SLP reduction
21627 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21629 * tree-data-ref.h (dr_alignment): Declare.
21630 * tree-data-ref.c (dr_alignment): New function.
21631 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
21632 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
21634 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
21636 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21638 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
21639 and base_misalignment fields.
21640 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
21641 * tree-data-ref.c: Include builtins.h.
21642 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
21643 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
21644 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
21645 * tree-vect-data-refs.c: Include tree-cfg.h.
21646 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
21647 fields instead of calculating an alignment here.
21648 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
21649 innermost_loop_behavior fields.
21651 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21653 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
21655 (DR_STEP_ALIGNMENT): New macro.
21656 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
21657 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
21658 (create_data_ref): Print it.
21659 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
21660 to tell whether the step preserves vector (mis)alignment.
21661 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
21662 Move the check for an integer step and generalise to all INTEGER_CST.
21663 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
21664 Print the outer step alignment.
21666 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21668 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
21669 with offset_alignment.
21670 (DR_ALIGNED_TO): Delete.
21671 (DR_OFFSET_ALIGNMENT): New macro.
21672 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
21673 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
21674 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
21675 (create_data_ref): Likewise.
21676 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
21677 (vect_analyze_data_refs): Likewise.
21678 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
21679 creating dummy innermost behavior.
21681 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21683 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
21684 with a "innermost_loop_behavior *" and refeence tree.
21685 * tree-data-ref.c (dr_analyze_innermost): Likewise.
21686 (create_data_ref): Update call accordingly.
21687 * tree-predcom.c (find_looparound_phi): Likewise.
21689 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21691 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
21692 fields with dr_wrt_vec_loop.
21693 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
21694 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
21695 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
21696 (vect_dr_behavior): New function.
21697 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21698 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
21699 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
21700 track whether the step preserves the misalignment.
21701 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21702 Use vect_dr_behavior.
21703 (vect_setup_realignment): Update call accordingly.
21704 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
21705 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
21706 call to vect_create_addr_base_for_vector_ref.
21707 (vect_create_cond_for_align_checks): Likewise.
21708 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
21709 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
21710 (vect_recog_mask_conversion_pattern): Likewise.
21711 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
21712 (new_stmt_vec_info): Remove redundant zeroing.
21714 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21716 * common/config/arm/arm-common.c (arm_be8_option): New function.
21717 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
21718 (ISA_ARMv6): Add isa_bit_be8.
21719 * config/arm/arm.h (arm_be8_option): Add prototype.
21720 (BE8_SPEC_FUNCTION): New define.
21721 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
21722 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
21723 (mlittle-endian): Similarly.
21724 (mbe8, mbe32): New options.
21725 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
21726 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
21728 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21730 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
21732 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21734 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
21735 (cleanup_tree_cfg_bb): Use it.
21736 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
21738 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
21740 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21743 * loop-doloop.c (add_test): Update profile.
21745 2017-07-03 Martin Liska <mliska@suse.cz>
21748 * sanopt.c (rewrite_usage_of_param): New function.
21749 (sanitize_rewrite_addressable_params): Likewise.
21750 (pass_sanopt::execute): Call rewrite_usage_of_param.
21752 2017-07-03 Richard Biener <rguenther@suse.de>
21754 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
21755 back to using VIEW_CONVERT_EXPR.
21757 2017-07-03 Martin Liska <mliska@suse.cz>
21760 * doc/extend.texi: Document when a resolver function is
21761 generated for target_clones.
21763 2017-07-03 Martin Liska <mliska@suse.cz>
21765 * asan.c (asan_emit_stack_protection): Unpoison just red zones
21766 and shadow memory of auto variables which are subject of
21767 use-after-scope sanitization.
21768 (asan_expand_mark_ifn): Add do set only when is_poison.
21770 2016-07-03 Richard Biener <rguenther@suse.de>
21772 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
21774 (vect_force_simple_reduction): Record reduction def -> phi mapping.
21775 (vectorizable_reduction): Perform reduction PHI creation when
21776 visiting a reduction PHI and adjust and simplify code generation
21777 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
21778 (vect_transform_loop): Visit reduction PHIs.
21779 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
21780 defs into the SLP tree.
21781 (vect_build_slp_tree): Reduction defs terminate the recursion.
21782 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
21784 (vect_get_vec_defs_for_stmt_copy): Export.
21785 (vect_get_vec_defs): Likewise.
21786 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
21788 (vect_get_vec_defs_for_stmt_copy): Declare.
21789 (vect_get_vec_defs): Likewise.
21791 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21793 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
21794 parameter with a "loop" parameter and use it instead of the
21795 loop containing DR_STMT. Don't check simple_iv when doing
21796 BB analysis. Describe the two analysis modes in the comment.
21798 2017-07-03 Tom de Vries <tom@codesourcery.com>
21800 PR tree-optimization/69468
21801 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
21802 (find_same_succ_bb): Handle ignore_edge_flags.
21804 2017-07-03 Tom de Vries <tom@codesourcery.com>
21806 PR tree-optimization/81192
21807 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
21809 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
21811 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
21813 2017-07-03 Tom de Vries <tom@codesourcery.com>
21815 PR tree-optimization/81192
21816 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
21817 BB_SAME_SUCC (bb) == NULL.
21819 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21821 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
21824 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21826 * dumpfile.c: Include profile-count.h
21827 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
21829 (insert_cond_bb): Update profile.
21830 * tree-cfg.h (insert_cond_bb): Update prototype.
21831 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
21832 * tree-dump.c: Do not include tree-cfg.
21834 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21836 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
21838 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21840 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
21843 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21845 * tree-complex.c (expand_complex_div_wide): update profile.
21847 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21848 Alan Hayward <alan.hayward@arm.com>
21849 David Sherwood <david.sherwood@arm.com>
21851 * Makefile.in (MACHMODE_H): Remove insn-modes.h
21852 (CORETYPES_H): New define.
21853 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
21854 (insn-modes-inline.h, s-modes-inline-h): New rules.
21855 (generated_files): Add insn-modes-inline.h.
21856 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
21857 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
21858 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
21859 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
21860 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
21861 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
21862 (build/gencodes.o, build/genconditions.o): Likewise.
21863 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
21864 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
21865 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
21866 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
21867 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
21868 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
21869 * coretypes.h: Include everything up to real.h for generators.
21870 Include insn-modes.h first. Include wide-int-print.h after
21871 wide-int.h. Include insn-modes-inline.h and then machmode.h.
21872 * machmode.h: Don't include insn-modes.h here.
21873 * function-tests.c: Remove includes of signop.h, machmode.h,
21874 double-int.h and wide-int.h.
21876 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
21878 * optc-save-gen.awk: Likewise.
21879 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
21880 * godump.c: Remove include of wide-int-print.h.
21881 * pretty-print.h: Likewise.
21882 * wide-int-print.cc: Likewise.
21883 * wide-int.cc: Likewise.
21884 * hash-map-tests.c: Remove include of signop.h.
21885 * hash-set-tests.c: Likewise.
21886 * rtl-tests.c: Likewise.
21887 * mkconfig.sh: Remove include of machmode.h.
21888 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
21890 (emit_insn_modes_inline_h): ...this new function. Emit the code
21891 into an insn-modes-inline.h header file, adding appropriate
21892 include guards and end comments.
21893 (emit_insn_modes_c_header): Remove include of machmode.h.
21894 (emit_min_insn_modes_c_header): Include coretypes.h rather than
21896 (main): Handle -i flag and call emit_insn_modes_inline_h when
21899 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21901 * tree-ssa-strlen.c (strinfo): Rename the length field to
21902 nonzero_chars. Add a full_string_p field.
21903 (compare_nonzero_chars, zero_length_string_p): New functions.
21904 (get_addr_stridx): Add an offset_out parameter.
21905 Use compare_nonzero_chars.
21906 (get_stridx): Update accordingly. Use compare_nonzero_chars.
21907 (new_strinfo): Update after above changes to strinfo.
21908 (set_endptr_and_length): Set full_string_p.
21909 (get_string_length): Update after above changes to strinfo.
21910 (unshare_strinfo): Update call to new_strinfo.
21911 (maybe_invalidate): Likewise.
21912 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
21913 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
21914 as a uhwi instead of an shwi. Update after above changes to
21915 strinfo and new_strinfo.
21916 (zero_length_string): Assert that chainsi contains full strings.
21917 Use zero_length_string_p. Update call to new_strinfo.
21918 (adjust_related_strinfos): Update after above changes to strinfo.
21919 Copy full_string_p from origsi.
21920 (adjust_last_stmt): Use zero_length_string_p.
21921 (handle_builtin_strlen): Update after above changes to strinfo and
21922 new_strinfo. Install the lhs as the string length if the previous
21923 entry didn't describe a full string.
21924 (handle_builtin_strchr): Update after above changes to strinfo
21926 (handle_builtin_strcpy): Likewise.
21927 (handle_builtin_strcat): Likewise.
21928 (handle_builtin_malloc): Likewise.
21929 (handle_pointer_plus): Likewise.
21930 (handle_builtin_memcpy): Likewise. Track nonzero characters
21931 that aren't necessarily followed by a nul terminator.
21932 (handle_char_store): Likewise.
21934 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21936 PR tree-optimization/80769
21937 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
21938 for malloc and calloc. Document the new invariant that all related
21939 strinfos have delayed lengths or none do.
21940 (verify_related_strinfos): Move earlier in file.
21941 (set_endptr_and_length): New function, split out from...
21942 (get_string_length): ...here. Also set the lengths of related
21944 (zero_length_string): Assert that chainsi has known (rather than
21946 (adjust_related_strinfos): Likewise.
21948 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21950 PR tree-optimization/81136
21951 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
21952 assert that two references with the same misalignment have the same
21953 compile-time misalignment if those compile-time misalignments
21956 2017-07-01 Andi Kleen <ak@linux.intel.com>
21958 * print-tree.c (print_node): Print all attributes.
21960 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21962 * cfg.c (scale_bbs_frequencies): New function.
21963 * cfg.h (scale_bbs_frequencies): Declare it.
21964 * cfgloopanal.c (single_likely_exit): Cleanup.
21965 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
21967 (scale_loop_profile): Likewise.
21968 (loop_version): Likewise.
21969 (create_empty_loop_on_edge): Update.
21970 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
21971 scale_loop_frequencies, scale_loop_profile, loopify,
21972 loop_version): Update prototypes.
21973 * modulo-sched.c (sms_schedule): Update.
21974 * predict.c (unlikely_executed_edge_p): Also check probability.
21975 (probably_never_executed_edge_p): Fix typo.
21976 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21977 * tree-parloops.c (gen_parallel_loop): Update.
21978 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
21979 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21980 * tree-ssa-loop-split.c (split_loop): Update.
21981 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21982 * tree-vect-loop-manip.c (vect_do_peeling): Update.
21983 (vect_loop_versioning): Update.
21984 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21986 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21988 * trans-mem.c (split_bb_make_tm_edge): Update profile.
21990 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21992 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
21993 to keep profile consistent.
21995 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21997 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
21998 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
21999 * profile-count.h (max_safe_multiplier): Make unsigned.
22000 (profile_count::guessed_zero): New.
22002 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22004 * bb-reorder.c (fix_up_crossing_landing_pad,
22005 fix_crossing_conditional_branches): Use make_single_succ_edge
22006 to keep profile consistent.
22008 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
22010 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
22013 2017-07-01 Jakub Jelinek <jakub@redhat.com>
22016 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
22017 the right scopes, make sure cond_jump isn't preserved between multiple
22018 iterations. Search for fallthru edge whenever there are 3+ edges and
22019 use find_fallthru_edge for it.
22021 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22023 Patch by Alexander Monakov <amonakov@ispras.ru>
22024 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
22025 probabilities consistently.
22027 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22029 * pa.c (pa_expand_compare_and_swap_loop): Update call of
22030 emit_cmp_and_jump_insns.
22032 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22035 * tree-inline.c (expand_call_inline): Combine profile statuses.
22037 2017-06-30 Andrew Pinski <apinski@cavium.com>
22039 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
22040 fold_stmt returned true.
22042 2017-06-30 Nathan Sidwell <nathan@acm.org>
22044 * ggc.h (empty_string): Delete.
22045 * cfgexpand.c (expand_asm_stmt): Use plain "".
22046 * optabs.c (expand_asm_memory_barrier): Likewise.
22047 * stringpool.c (empty_string): Delete.
22048 (digit_vector, digit_string): Delete.
22049 (ggc_alloc_string): Use plain "", don't optimize single digit
22050 strings. Use ggc_alloc_atomic.
22052 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22054 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22055 comparison set and one other set, use the cost of the non-comparison
22058 2017-06-30 Nathan Sidwell <nathan@acm.org>
22060 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22063 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22065 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22066 loops. Remove now unneeded calls to gimple_switch_set_label() that
22067 just set removed labels to NULL_TREE.
22069 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22071 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22072 (set_range_info): ...here. Only call set_range_info_raw if domain
22074 (set_nonzero_bits): Call set_range_info_raw.
22075 * tree-ssanames.h (set_range_info_raw): New.
22077 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22080 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22081 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22082 of nonimmediate_operand and <store_mask_constraint> instead of m
22083 for the input operand. For V8FI iterator, always split if input
22084 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22085 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22086 <store_mask_predicate> instead of register_operand and
22087 <store_mask_constraint> instead of v for the input operand. Make
22088 sure both operands aren't MEMs for if not <mask_applied>.
22090 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22092 * lto-wrapper.c (copy_file) Close both file descriptors before
22095 2017-06-30 Martin Liska <mliska@suse.cz>
22098 * multiple_target.c (create_dispatcher_calls): Make ifunc
22099 also for function that don't have calls or are not referenced.
22101 2017-06-30 Richard Biener <rguenther@suse.de>
22103 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22104 analyze the first scalar stmt. Move vector type computation
22105 for the BB case here from ...
22106 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22107 live operation processing in the SLP case properly.
22109 2017-06-30 Richard Biener <rguenther@suse.de>
22111 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
22113 2017-06-30 Martin Liska <mliska@suse.cz>
22116 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
22117 before BUILT_IN_UNWIND_RESUME when ASAN is used.
22119 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
22121 * doc/invoke.texi (AArch64): Add missing options and remove redundant
22124 2017-06-30 Richard Biener <rguenther@suse.de>
22126 PR tree-optimization/81249
22127 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
22128 condition reduction result to original scalar type.
22130 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22132 * profile-count.h (enum profile_quality): Fix typos and whitespace
22135 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22137 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
22138 type for branch probabilities.
22140 2017-06-29 Julian Brown <julian@codesourcery.com>
22141 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22143 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
22144 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
22145 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
22146 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
22148 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22150 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
22151 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
22152 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
22153 CC usage from generic code to here.
22154 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
22155 CC usage into the target macros.
22157 2017-06-29 Maya Rashish <coypu@sdf.org>
22159 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
22162 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22164 * arm/arm-builtins.c: Include profile-count.h
22165 * except.c (sjlj_emit_function_enter): Use
22166 profile_probability::unlikely.
22168 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22170 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
22171 and tocrel_offset be pointer args rather than implicitly using
22173 (legitimate_constant_pool_address_p, rs6000_emit_move,
22174 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
22175 tocrel_offset and use in toc_relative_expr_p call.
22176 (print_operand, print_operand_address): Use static tocrel_base_oac
22177 and tocrel_offset_oac.
22178 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
22181 2017-06-29 Maya Rashish <coypu@sdf.org>
22183 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
22185 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
22187 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
22188 objects, take into account only the alignment of 'op0' and 'mode1' if
22191 2017-06-29 Steve Ellcey <sellcey@cavium.com>
22193 * ccmp.c (ccmp_tree_comparison_p): New function.
22194 (ccmp_candidate_p): Update to use above function.
22195 (get_compare_parts): New function.
22196 (expand_ccmp_next): Update to use new functions.
22197 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
22199 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
22200 take mode as argument.
22201 * ccmp.h (expand_ccmp_expr): Add mode as argument.
22202 * expr.c (expand_expr_real_1): Pass mode as argument.
22204 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
22206 * combine.c (combine_instructions): Print insns to dump_file, together
22209 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22211 * asan.c (asan_emit_stack_protection): Update.
22212 (create_cond_insert_point): Update.
22213 * auto-profile.c (afdo_propagate_circuit): Update.
22214 * basic-block.h (struct edge_def): Turn probability to
22215 profile_probability.
22216 (EDGE_FREQUENCY): Update.
22217 * bb-reorder.c (find_traces_1_round): Update.
22218 (better_edge_p): Update.
22219 (sanitize_hot_paths): Update.
22220 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
22221 (make_single_succ_edge): Update.
22222 (check_bb_profile): Update.
22223 (dump_edge_info): Update.
22224 (update_bb_profile_for_threading): Update.
22225 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
22226 probabilitycount to 0.
22227 * cfgbuild.c (compute_outgoing_frequencies): Update.
22228 * cfgcleanup.c (try_forward_edges): Update.
22229 (outgoing_edges_match): Update.
22230 (try_crossjump_to_edge): Update.
22231 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
22232 (expand_gimple_tailcall): Update.
22233 (construct_init_block): Use make_single_succ_edge.
22234 (construct_exit_block): Use make_single_succ_edge.
22235 * cfghooks.c (verify_flow_info): Update.
22236 (redirect_edge_succ_nodup): Update.
22237 (split_edge): Update.
22238 (account_profile_record): Update.
22239 * cfgloopanal.c (single_likely_exit): Update.
22240 * cfgloopmanip.c (scale_loop_profile): Update.
22241 (set_zero_probability): Remove.
22242 (duplicate_loop_to_header_edge): Update.
22243 * cfgloopmanip.h (loop_version): Update prototype.
22244 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
22245 (force_nonfallthru_and_redirect): Update.
22246 (update_br_prob_note): Update.
22247 (rtl_verify_edges): Update.
22248 (purge_dead_edges): Update.
22249 (rtl_lv_add_condition_to_bb): Update.
22250 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
22251 * cgraphunit.c (init_lowered_empty_function): Update.
22252 (cgraph_node::expand_thunk): Update.
22253 * cilk-common.c: Include profile-count.h
22254 * dojump.c (inv): Remove.
22255 (jumpifnot): Update.
22256 (jumpifnot_1): Update.
22257 (do_jump_1): Update.
22259 (do_jump_by_parts_greater_rtx): Update.
22260 (do_compare_rtx_and_jump): Update.
22261 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
22262 do_jump_1. do_compare_rtx_and_jump): Update prototype.
22263 * dwarf2cfi.c: Include profile-count.h
22264 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
22265 (sjlj_emit_dispatch_table): Likewise.
22266 * explow.c: Include profile-count.h
22267 * expmed.c (emit_store_flag_force): Update.
22268 (do_cmp_and_jump): Update.
22269 * expr.c (compare_by_pieces_d::generate): Update.
22270 (compare_by_pieces_d::finish_mode): Update.
22271 (emit_block_move_via_loop): Update.
22272 (store_expr_with_bounds): Update.
22273 (store_constructor): Update.
22274 (expand_expr_real_2): Update.
22275 (expand_expr_real_1): Update.
22276 * expr.h (try_casesi, try_tablejump): Update prototypes.
22277 * gimple-pretty-print.c (dump_probability): Update.
22278 (dump_profile): New.
22279 (dump_gimple_label): Update.
22280 (dump_gimple_bb_header): Update.
22281 * graph.c (draw_cfg_node_succ_edges): Update.
22282 * hsa-gen.c (convert_switch_statements): Update.
22283 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
22284 (find_if_case_1): Update.
22285 (find_if_case_2): Update.
22286 * internal-fn.c (expand_arith_overflow_result_store): Update.
22287 (expand_addsub_overflow): Update.
22288 (expand_neg_overflow): Update.
22289 (expand_mul_overflow): Update.
22290 (expand_vector_ubsan_overflow): Update.
22291 * ipa-cp.c (good_cloning_opportunity_p): Update.
22292 * ipa-split.c (split_function): Use make_single_succ_edge.
22293 * ipa-utils.c (ipa_merge_profiles): Update.
22294 * loop-doloop.c (add_test): Update.
22295 (doloop_modify): Update.
22296 * loop-unroll.c (compare_and_jump_seq): Update.
22297 (unroll_loop_runtime_iterations): Update.
22298 * lra-constraints.c (lra_inheritance): Update.
22299 * lto-streamer-in.c (input_cfg): Update.
22300 * lto-streamer-out.c (output_cfg): Update.
22301 * mcf.c (adjust_cfg_counts): Update.
22302 * modulo-sched.c (sms_schedule): Update.
22303 * omp-expand.c (expand_omp_for_init_counts): Update.
22304 (extract_omp_for_update_vars): Update.
22305 (expand_omp_ordered_sink): Update.
22306 (expand_omp_for_ordered_loops): Update.
22307 (expand_omp_for_generic): Update.
22308 (expand_omp_for_static_nochunk): Update.
22309 (expand_omp_for_static_chunk): Update.
22310 (expand_cilk_for): Update.
22311 (expand_omp_simd): Update.
22312 (expand_omp_taskloop_for_outer): Update.
22313 (expand_omp_taskloop_for_inner): Update.
22314 * omp-simd-clone.c (simd_clone_adjust): Update.
22315 * optabs.c (expand_doubleword_shift): Update.
22316 (expand_abs): Update.
22317 (emit_cmp_and_jump_insn_1): Update.
22318 (expand_compare_and_swap_loop): Update.
22319 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
22320 * predict.c (predictable_edge_p): Update.
22321 (edge_probability_reliable_p): Update.
22322 (set_even_probabilities): Update.
22323 (combine_predictions_for_insn): Update.
22324 (combine_predictions_for_bb): Update.
22325 (propagate_freq): Update.
22326 (estimate_bb_frequencies): Update.
22327 (force_edge_cold): Update.
22328 * profile-count.c (profile_count::dump): Add missing space into dump.
22329 (profile_count::debug): Add newline.
22330 (profile_count::differs_from_p): Explicitly convert to unsigned.
22331 (profile_count::stream_in): Update.
22332 (profile_probability::dump): New member function.
22333 (profile_probability::debug): New member function.
22334 (profile_probability::differs_from_p): New member function.
22335 (profile_probability::differs_lot_from_p): New member function.
22336 (profile_probability::stream_in): New member function.
22337 (profile_probability::stream_out): New member function.
22338 * profile-count.h (profile_count_quality): Rename to ...
22339 (profile_quality): ... this one.
22340 (profile_probability): New.
22341 (profile_count): Update.
22342 * profile.c (compute_branch_probabilities): Update.
22343 * recog.c (peep2_attempt): Update.
22344 * sched-ebb.c (schedule_ebbs): Update.
22345 * sched-rgn.c (find_single_block_region): Update.
22346 (compute_dom_prob_ps): Update.
22347 (schedule_region): Update.
22348 * sel-sched-ir.c (compute_succs_info): Update.
22349 * stmt.c (struct case_node): Update.
22350 (do_jump_if_equal): Update.
22351 (get_outgoing_edge_probs): Update.
22352 (conditional_probability): Update.
22353 (emit_case_dispatch_table): Update.
22354 (expand_case): Update.
22355 (expand_sjlj_dispatch_table): Update.
22356 (emit_case_nodes): Update.
22357 * targhooks.c: Update.
22358 * tracer.c (better_p): Update.
22359 (find_best_successor): Update.
22360 * trans-mem.c (expand_transaction): Update.
22361 * tree-call-cdce.c: Update.
22362 * tree-cfg.c (gimple_split_edge): Upate.
22363 (move_sese_region_to_fn): Upate.
22364 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
22365 * tree-eh.c (lower_resx): Upate.
22366 (cleanup_empty_eh_move_lp): Upate.
22367 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22368 * tree-inline.c (copy_edges_for_bb): Update.
22369 (copy_cfg_body): Update.
22370 * tree-parloops.c (gen_parallel_loop): Update.
22371 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
22372 (gimple_gen_time_profiler): Update.
22373 * tree-ssa-dce.c (remove_dead_stmt): Update.
22374 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22375 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
22376 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
22377 (unloop_loops): Update.
22378 (try_peel_loop): Update.
22379 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22380 * tree-ssa-loop-split.c (connect_loops): Update.
22381 (split_loop): Update.
22382 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22383 (hoist_guard): Update.
22384 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
22385 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
22386 (value_replacement): Update.
22387 * tree-ssa-reassoc.c (branch_fixup): Update.
22388 * tree-ssa-tail-merge.c (replace_block_by): Update.
22389 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
22390 (create_edge_and_update_destination_phis): Update.
22391 (compute_path_counts): Update.
22392 (recompute_probabilities): Update.
22393 (update_joiner_offpath_counts): Update.
22394 (freqs_to_counts_path): Update.
22395 (duplicate_thread_path): Update.
22396 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
22397 (struct switch_conv_info): Update.
22398 (gen_inbound_check): Update.
22399 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22400 (vect_do_peeling): Update.
22401 (vect_loop_versioning): Update.
22402 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22403 (optimize_mask_stores): Update.
22404 * ubsan.c (ubsan_expand_null_ifn): Update.
22405 * value-prof.c (gimple_divmod_fixed_value): Update.
22406 (gimple_divmod_fixed_value_transform): Update.
22407 (gimple_mod_pow2): Update.
22408 (gimple_mod_pow2_value_transform): Update.
22409 (gimple_mod_subtract): Update.
22410 (gimple_mod_subtract_transform): Update.
22411 (gimple_ic): Update.
22412 (gimple_stringop_fixed_value): Update.
22413 (gimple_stringops_transform): Update.
22414 * value-prof.h: Update.
22416 2017-06-29 Carl Love <cel@us.ibm.com>
22418 * config/rs6000/rs6000-c.c: Add support for built-in functions
22419 vector signed int vec_signed (vector float);
22420 vector signed long long vec_signed (vector double);
22421 vector signed int vec_signed2 (vector double, vector double);
22422 vector signed int vec_signede (vector double);
22423 vector signed int vec_signedo (vector double);
22424 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
22425 instruction generator.
22426 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22427 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
22428 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
22430 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
22431 vunsignede_v2df): Add define_expands.
22432 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
22433 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
22434 VEC_UNSIGNEDO): Add definitions.
22435 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22436 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
22437 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
22438 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
22439 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
22440 * config/rs6000/altivec.h (vec_signed, vec_signed2,
22441 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
22442 vec_unsignede, vec_unsignedo): Add builtin defines.
22443 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
22445 * doc/extend.texi: Update the built-in documentation file for the
22446 new built-in functions.
22448 2017-06-29 Richard Biener <rguenther@suse.de>
22450 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
22451 reduction chains to LOOP_VINFO_REDUCTIONS.
22452 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
22453 SLP reductions after processing reduction chains.
22455 2017-06-29 Nathan Sidwell <nathan@acm.org>
22457 * builtins.c (fold_builtin_FUNCTION): Use
22458 lang_hooks.decl_printable_name.
22460 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
22462 PR middle-end/81194
22463 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
22464 with only one label.
22465 * stmt.c (expand_case): Assert NCASES is greater than one.
22467 2017-06-29 Richard Biener <rguenther@suse.de>
22469 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
22471 (group_case_labels): Likewise.
22472 (find_taken_edge): Push sanity checking on val to workers...
22473 (find_taken_edge_cond_expr): ... here
22474 (find_taken_edge_switch_expr): ... and here, handle cases
22475 with just a default label.
22476 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
22477 (group_case_labels): Likewise.
22478 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
22479 group_case_labels does anything cleanup the CFG again.
22481 2017-06-29 Bin Cheng <bin.cheng@arm.com>
22483 PR tree-optimization/81196
22484 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
22485 exit condition comparing two IVs.
22487 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
22489 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
22490 profile to the dummy entry at the end of the list of architectures.
22491 * config/arm/arm-cpu-cdata.h: Regenerated.
22493 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22494 Michael Collison <michael.collison@arm.com>
22497 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
22499 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
22500 (*aarch64_reg_<mode>3_minus_mask): New pattern.
22501 (*aarch64_<optab>_reg_di3_mask2): New pattern.
22502 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
22503 of shift when the shift amount is masked with constant equal to
22504 the size of the mode.
22505 * config/aarch64/predicates.md (subreg_lowpart_operator): New
22508 2017-06-29 Martin Liska <mliska@suse.cz>
22510 * config/i386/i386.opt: Change range from [1,5] to [0,5].
22512 2017-06-29 Yury Gribov <tetra2005@gmail.com>
22515 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
22518 (edge_growth_cache_entry::edge_growth_cache_entry): New
22520 (reset_edge_growth_cache): Update to use constructor.
22522 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22524 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
22525 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22526 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22528 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
22530 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
22531 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
22533 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
22535 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
22536 (*-linux-uclibc*): Add t-uclibc tmake_file.
22537 * config/t-musl: New.
22538 * config/t-uclibc: New.
22540 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
22542 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
22544 (gen_comm_data): Emit architectural setting of arch_prof.
22545 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
22547 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
22548 (armv8-m.base, armv8-m.main): Likewise.
22549 * arm-protos.h (arm_build_target): Add profile field.
22550 (arch_option): Likewise.
22551 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
22553 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
22554 arm_active_target.profile.
22556 2017-06-28 Richard Biener <rguenther@suse.de>
22558 PR middle-end/81227
22559 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
22560 TYPE_OVERFLOW_WRAPS.
22561 * match.pd (negate_expr_p): Likewise.
22562 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
22563 fold_build2, not fold_binary.
22565 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22567 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
22568 Convert memory address to Pmode.
22569 (aarch64_print_operand): Assert MEM operands are always Pmode.
22571 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22574 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
22575 Remove redundant if.
22576 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
22577 * config/arm/aarch-common-protos.h
22578 (aarch_forward_to_shift_is_not_shifted_re): Remove.
22579 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
22581 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
22584 * multiple_target.c (create_dispatcher_calls): Set the default
22585 clone to be static, not public.
22587 2017-06-28 Richard Biener <rguenther@suse.de>
22589 * tree-vect-loop.c (vectorizable_reduction): Move special
22590 cond reduction IV var creation ...
22591 (vect_create_epilog_for_reduction): ... here. Remove induction_index
22592 parameter. Use STMT_VINFO_VECTYPE.
22593 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
22596 2017-06-28 Martin Liska <mliska@suse.cz>
22599 * ipa-visibility.c (non_local_p): Handle visibility.
22601 2017-06-28 Martin Liska <mliska@suse.cz>
22604 * common.opt: Add IntegerRange to various options.
22605 * opt-functions.awk (integer_range_info): New function.
22606 * optc-gen.awk: Add integer_range_info to cl_options struct.
22607 * opts-common.c (decode_cmdline_option): Handle
22608 CL_ERR_INT_RANGE_ARG.
22609 (cmdline_handle_error): Likewise.
22610 * opts.c (print_filtered_help): Show valid interval in
22611 when --help is provided.
22612 * opts.h (struct cl_option): Add range_min and range_max fields.
22613 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
22615 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
22617 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
22618 (x * C EQ/NE y * C): New transformation.
22620 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
22622 * genmultilib (combination_space): Accept '+' in option names.
22624 2017-06-28 Martin Liska <mliska@suse.cz>
22627 * asan.c (instrument_derefs): Bail out inner references
22628 that are hard register variables.
22630 2017-06-28 Jakub Jelinek <jakub@redhat.com>
22633 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
22634 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
22636 2017-06-28 Richard Biener <rguenther@suse.de>
22638 * tree-vectorizer.h (vect_get_vec_defs): Remove.
22639 (vect_get_slp_defs): Adjust.
22640 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
22642 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
22644 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
22645 get_initial_defs_for_reduction instead of vect_get_vec_defs.
22646 (vectorizable_reduction): Adjust.
22647 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
22649 (vect_get_slp_defs): Likewise.
22650 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
22651 (vectorizable_bswap): Adjust.
22652 (vectorizable_call): Likewise.
22653 (vectorizable_conversion): Likewise.
22654 (vectorizable_assignment): Likewise.
22655 (vectorizable_shift): Likewise.
22656 (vectorizable_operation): Likewise.
22657 (vectorizable_store): Likewise.
22658 (vectorizable_condition): Likewise.
22659 (vectorizable_comparison): Likewise.
22661 2017-06-28 Michael Collison <michael.collison@arm.com>
22664 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
22666 (arm_gen_movmemqi): Removed unused variable 'i'.
22667 Convert 'for' loop into 'while' loop.
22668 (arm_expand_prologue): Remove last unnecessary set of insn.
22669 (thumb_pop): Remove unused variable 'pushed_words'.
22670 (thumb_exit): Remove last unnecessary set of regs_to_pop.
22672 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22674 * config/s390/predicates.md: Use s390_rel_address_ok_p.
22675 * config/s390/s390-protos.h: Add prototype of
22676 s390_rel_address_ok_p.
22677 * config/s390/s390.c (s390_got_symbol): New function.
22678 (s390_rel_address_ok_p): New function.
22679 (legitimize_pic_address): Use s390_rel_address_ok_p.
22680 (s390_load_got): Use s390_got_symbol.
22681 (s390_option_override): Issue error if
22682 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
22683 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
22685 * config/s390/s390.opt: New option mpic-data-is-text-relative.
22687 2017-06-27 Andrew Pinski <apinski@cavium.com>
22689 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
22690 (X * copysign (1.0, X)): New pattern.
22691 (X * copysign (1.0, -X)): New pattern.
22692 (copysign (-1.0, CST)): New pattern.
22694 2017-06-27 Joseph Myers <joseph@codesourcery.com>
22696 * genmultilib (combination_space): Remove variable.
22697 Validate reuse rules against regular expression for any sequence
22698 of multilib options in any order.
22700 2017-06-27 Michael Collison <michael.collison@arm.com>
22702 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
22703 call aarch64_split_simd_combine.
22704 * (aarch64_combine_internal<mode>): Delete pattern.
22705 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
22706 Allow register and subreg operands.
22708 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22710 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
22711 specific need, just fallback on defaults.
22712 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
22714 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22715 Olivier Hainque <hainque@adacore.com>
22717 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
22719 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
22720 targets. Pick a default if no particular attempt applied.
22721 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
22724 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22726 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
22727 (x86_64-wrs-vxworks7): Likewise.
22729 2017-06-27 Marek Polacek <polacek@redhat.com>
22732 * ubsan.c (instrument_null): Check get_base_address's result for null.
22734 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22736 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
22738 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22740 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
22741 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
22742 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
22743 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
22744 New function types.
22745 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
22746 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
22747 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
22748 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
22749 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
22750 BUILT_IN_FEUPDATEENV): New builtins.
22751 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
22752 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
22753 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
22754 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
22756 (builtin_structptr_types): Adjust size.
22757 * tree.c (builtin_structptr_types): Add four entries.
22759 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22760 Olivier Hainque <hainque@adacore.com>
22762 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
22763 (TLS_SYM): New local macro, forcing reference to __tls__ on
22764 link command lines for VxWorks 7 RTPs, triggering initialization
22766 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
22767 OS features TLS support, true for RTPs on VxWorks 7.
22768 * config/vxworks.c (vxworks_override_options): Setup emutls
22771 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22773 * predict.c (test_prediction_value_range): Use -1U instead of -1
22774 to avoid narrowing conversion warning.
22775 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
22776 to avoid narrowing conversion warning.
22777 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
22779 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
22781 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22783 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
22784 64bit configurations.
22785 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
22786 (SIZE_TYPE): Likewise.
22787 * config/vxworks.c (vxworks_emutls_var_fields): Use
22788 long_unsigned_type_node instead of unsigned_type_node as the offset
22789 field type, which is "pointer" mode in emutls.c.
22791 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22794 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
22796 PR middle-end/81207
22797 * gimple-fold.c (replace_call_with_call_and_fold): Handle
22798 gimple_vuse copying separately from gimple_vdef copying.
22800 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22802 * value-prof.c (free_hist): Remove call to memset and the enclosing if
22805 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22806 Olivier Hainque <hainque@adacore.com>
22808 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
22809 for all vxworks7 targets.
22810 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
22811 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
22812 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
22813 variations for VX6/VX7 and 32/64bits later on in ...
22814 (VXWORKS_LIB_SPEC): Leverage new macros.
22815 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
22816 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
22818 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22820 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
22821 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
22823 2017-06-26 Carl Love <cel@us.ibm.com>
22825 * config/rs6000/rs6000-c.c: Add support for built-in functions
22826 vector bool char vec_reve (vector bool char);
22827 vector signed char vec_reve (vector signed char);
22828 vector unsigned char vec_reve (vector unsigned char);
22829 vector bool int vec_reve (vector bool int);
22830 vector signed int vec_reve (vector signed int);
22831 vector unsigned int vec_reve (vector unsigned int);
22832 vector bool long long vec_reve (vector bool long long);
22833 vector signed long long vec_reve (vector signed long long);
22834 vector unsigned long long vec_reve (vector unsigned long long);
22835 vector bool short vec_reve (vector bool short);
22836 vector signed short vec_reve (vector signed short);
22837 vector double vec_reve (vector double);
22838 vector float vec_reve (vector float);
22839 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
22840 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
22841 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
22842 (altivec_vreve): New pattern.
22843 * config/rs6000/altivec.h (vec_reve): New define.
22844 * doc/extend.texi (vec_rev): Update the built-in documentation file
22845 for the new built-in functions.
22847 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22849 PR tree-optimization/71815
22850 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
22852 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
22854 (slsr_process_phi): Likewise.
22855 (replace_uncond_cands_and_profitable_phis): Don't replace a
22856 multiply candidate with a stride of 1 (copy or cast).
22857 (phi_incr_cost): Call uses_consumed_by_stmt rather than
22859 (lowest_cost_path): Likewise.
22860 (total_savings): Likewise.
22862 2017-06-26 Richard Biener <rguenther@suse.de>
22865 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
22866 Use def_builtin_pure for all gather builtins.
22868 2017-06-26 Richard Biener <rguenther@suse.de>
22870 PR tree-optimization/81203
22871 * tree-tailcall.c (find_tail_calls): Do not move stmts into
22872 non-dominating BBs.
22874 2017-06-26 Marek Polacek <polacek@redhat.com>
22877 * doc/invoke.texi: Document -Wmultistatement-macros.
22879 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
22881 * doc/sourcebuild.texi (ARM-specific attributes): Document new
22882 arm_neon_ok_no_float_abi effective target.
22884 2017-06-26 Richard Biener <rguenther@suse.de>
22886 PR tree-optimization/80928
22887 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
22888 (copy_bbs): Set BB_DUPLICATED flag early.
22889 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
22891 (execute_on_shrinking_pred): Likewise.
22892 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
22893 BB_DUPLICATED blocks.
22894 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
22895 iterate over all PHIs considering removal of *gsi.
22897 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
22899 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
22902 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22904 * config/rs6000/rs6000-string.c: (expand_block_clear,
22905 do_load_for_compare, select_block_compare_mode,
22906 compute_current_alignment, expand_block_compare,
22907 expand_strncmp_align_check, expand_strn_compare,
22908 expand_block_move, rs6000_output_load_multiple)
22909 Move functions related to string/block move/compare
22910 to a separate file.
22911 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
22912 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
22913 for this function which is now used in two files.
22914 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
22915 * config.gcc: Add rs6000-string.o to extra_objs for
22916 targets powerpc*-*-* and rs6000*-*-*.
22918 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
22921 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
22922 32-bit, since indexed is not valid for DImode.
22923 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
22924 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
22925 (define_peephole2 for Altivec d-form load): Add 32-bit support.
22926 (define_peephole2 for Altivec d-form store): Likewise.
22929 * multiple_target.c (create_dispatcher_calls): Only create the
22930 dispatcher call if the function is the default clone of a
22931 versioned function.
22933 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
22935 PR middle-end/80902
22936 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
22937 a call, force the call to not be a tail call.
22939 2017-06-23 Jeff Law <law@redhat.com>
22941 * doc/contrib.texi: Add entry for Steven Pemberton's work on
22944 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
22946 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
22947 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
22948 handling for early expansion of vector shifts (sl,sr,sra,rl).
22949 (builtin_function_type): Add vector shift right instructions
22950 to the unsigned argument list.
22952 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
22954 rtl-optimizatoin/79286
22955 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
22956 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
22957 trap. PIC register plus a const unspec without offset can never trap.
22959 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
22961 * tree.h (builtin_structptr_type): New type.
22962 (builtin_structptr_types): Declare new array.
22963 * tree.c (builtin_structptr_types): New array.
22964 (free_lang_data, build_common_tree_nodes): Use it.
22966 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
22969 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
22972 2017-06-22 Matt Turner <mattst88@gmail.com>
22974 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
22975 Lake models to skylake case. Assume skylake for unknown
22976 models with clflushopt.
22978 2017-06-22 Jeff Law <law@redhat.com>
22980 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
22981 frame sizes that do not satisfy aarch64_uimm12_shift.
22983 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
22985 * profile-count.h (apply_probability,
22986 apply_scale, probability_in): Fix checks for zero.
22988 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22990 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
22991 * doc/cppdiropts.texi (-I @var{dir}): Document it.
22993 2016-06-22 Richard Biener <rguenther@suse.de>
22995 * tree-vect-loop.c (vect_model_reduction_cost): Handle
22996 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
22997 REDUC_MAX_EXPR support.
22998 (vectorizable_reduction): Likewise.
22999 (vect_create_epilog_for_reduction): Likewise.
23001 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
23003 * match.pd (A / (1 << B) -> A >> B): New.
23004 * generic-match-head.c: Include optabs-tree.h.
23005 * gimple-match-head.c: Likewise.
23006 * optabs-tree.h (target_supports_op_p): New.
23007 * optabs-tree.c (target_supports_op_p): New.
23009 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23011 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
23012 $gcc_cv_ld --help output.
23013 (gcc_cv_ld_demangle): Likewise.
23014 (gcc_cv_ld_eh_frame_hdr): Likewise.
23015 (gcc_cv_ld_pie): Likewise.
23016 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
23017 (gcc_cv_ld_buildid): Likewise.
23018 (gcc_cv_ld_sysroot): Likewise.
23019 (ld_bndplt_support): Likewise.
23020 (ld_pushpopstate_support): Likewise.
23021 * configure: Regenerate.
23022 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
23024 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23027 * config/i386/sse.md (round<mode>2): Renumber match_dup and
23028 operands indexes to avoid gap between operands and match_dups.
23030 2017-06-21 Andrew Pinski <apinski@cavium.com>
23032 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
23033 Increment Arith_shift and Arith_shift_reg by 1.
23034 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
23036 * config/aarch64/aarch64.c (thunderx_tunings): Enable
23037 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
23038 (aarch64_strip_extend): Add new argument and test for it.
23039 (aarch64_cheap_mult_shift_p): New function.
23040 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
23041 add a cost if it is true.
23042 Update calls to aarch64_strip_extend.
23043 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
23045 2017-06-21 Andrew Pinski <apinski@cavium.com>
23047 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
23049 (thunderxt88): Likewise.
23050 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
23051 (thunderx_prefetch_tune): New variable.
23052 (thunderx2t99_prefetch_tune): Update for the correct values.
23053 (thunderxt88_tunings): New variable.
23054 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23055 generic_prefetch_tune.
23056 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23058 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23060 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23061 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23062 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23063 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23064 (aarch64_atomic_cas<mode>, GPI): Likewise.
23066 2017-06-21 Martin Liska <mliska@suse.cz>
23068 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23069 statements on cold and hot labels.
23070 * predict.c (tree_estimate_probability_bb): Remove the
23071 prediction from this place.
23073 2017-06-21 Martin Liska <mliska@suse.cz>
23075 PR tree-optimization/79489
23076 * gimplify.c (maybe_add_early_return_predict_stmt): New
23078 (gimplify_return_expr): Call the function.
23079 * predict.c (tree_estimate_probability_bb): Remove handling
23081 * predict.def: Update comment about early return predictor.
23082 * gimple-predict.h (is_gimple_predict): New function.
23083 * predict.def: Change default value of early return to 66.
23084 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23086 * passes.def: Put pass_strip_predict_hints to the beginning of
23089 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23091 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23092 FUNCTION_DECL declarations.
23093 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23095 (dwaf2out_decl): Likewise.
23096 * godump.c (go_early_global_decl): Skip call to the real debug hook
23097 for FUNCTION_DECL declarations.
23098 * passes.c (rest_of_decl_compilation): Skip call to the
23099 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23100 -fdump-go-spec is passed.
23102 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23104 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23105 Reorder for compactness.
23106 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23107 (def_builtin_pure, def_builtin_pure2): New functions.
23108 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23110 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23112 * match.pd (nop_convert): New predicate.
23113 ((A +- CST1) +- CST2): Allow some NOP conversions.
23115 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23118 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
23119 with ctors/dtors if GOVD_SHARED is set.
23121 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23123 * config/aarch64/aarch64.md (movti_aarch64):
23124 Emit mov rather than orr.
23125 (movtf_aarch64): Likewise.
23126 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
23127 Emit mov rather than orr.
23129 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23131 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
23132 Swap alternatives, make integer dup more expensive.
23134 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23136 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23137 Return true for non-tls symbols.
23139 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
23141 * config/aarch64/aarch64-cores.def (cortex-a55): New.
23142 (cortex-a75): Likewise.
23143 (cortex-a75.cortex-a55): Likewise.
23144 * config/aarch64/aarch64-tune.md: Regenerate.
23145 * doc/invoke.texi (-mtune): Document new values for -mtune.
23147 2017-06-21 Tom de Vries <tom@codesourcery.com>
23149 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
23150 stack_size feature.
23151 (Effective-Target Keywords, Other attributes): Suggest using
23152 dg-add-options stack_size feature to get stack limit in stack_size
23153 effective target documentation.
23155 2017-06-21 Julian Brown <julian@codesourcery.com>
23156 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23158 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
23159 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
23160 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
23162 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
23163 attribute type list for neon_multiply.
23164 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
23165 attribute type list for neon_multiply.
23166 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
23167 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
23168 attribute type list for neon_multiply.
23169 * config/arm/types.md (crypto_pmull): Add.
23170 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
23171 attribute type list.
23173 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
23175 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
23178 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23180 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
23181 to make sure not to dereference a NULL cost_classes_ptr pointer.
23183 2017-06-20 Carl Love <cel@us.ibm.com>
23185 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23186 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23187 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23188 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23189 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23190 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23191 VMULOSW): New enum "unspec" values.
23192 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23193 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23194 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23195 altivec_vmulosw): New patterns.
23196 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23197 VMULOSW): Add definitions.
23199 2017-06-20 Julia Koval <julia.koval@intel.com>
23201 * config/i386/i386.c: Fix rounding expand for new pattern.
23202 * config/i386/subst.md: Fix pattern (parallel -> unspec).
23204 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23206 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
23207 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
23209 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23211 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
23214 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23216 * config/aarch64/aarch64-cores.def: Rearrange to sort by
23217 architecture, then by implementer ID.
23218 * config/aarch64/aarch64-tune.md: Regenerate.
23220 2017-06-20 Richard Biener <rguenther@suse.de>
23222 PR middle-end/81097
23223 * fold-const.c (split_tree): Fold to type before negating.
23225 2017-06-20 David Malcolm <dmalcolm@redhat.com>
23227 * diagnostic-show-locus.c
23228 (selftest::test_fixit_deletion_affecting_newline): New function.
23229 (selftest::diagnostic_show_locus_c_tests): Call it.
23231 2017-06-20 Andreas Schwab <schwab@suse.de>
23234 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
23237 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23239 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
23240 __ARM_FEATURE_COPROC according to support.
23242 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23244 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23245 Rewritten to avoid overflow for > 32-bit pointers.
23248 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
23249 by removing enum keyword.
23250 (ubsan_type_descriptor): Likewise. Formatting fix.
23253 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
23254 splitter): Require TARGET_SSE2 in the condition.
23256 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
23259 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
23260 for doing vector set of SFmode on ISA 3.0.
23261 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
23262 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
23264 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
23265 SFmode value into a V4SF variable that was extracted from another
23266 V4SF variable without converting the element to double precision
23267 and back to single precision vector format.
23268 (vsx_insert_extract_v4sf_p9_2): Likewise.
23270 2017-06-19 Jakub Jelinek <jakub@redhat.com>
23272 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
23273 in UWHI to avoid undefined overflow.
23276 * ubsan.h (enum ubsan_encode_value_phase): New.
23277 (ubsan_encode_value): Change second argument to
23278 enum ubsan_encode_value_phase with default value of
23279 UBSAN_ENCODE_VALUE_GENERIC.
23280 * ubsan.c (ubsan_encode_value): Change second argument to
23281 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
23282 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
23283 create_tmp_var_raw instead of create_tmp_var and use a
23285 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
23286 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
23287 ubsan_encode_value callers.
23290 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
23291 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
23292 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
23294 2017-06-19 Richard Biener <rguenther@suse.de>
23296 PR middle-end/81118
23297 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
23298 estimates if we changed anything.
23300 2017-06-19 Richard Biener <rguenther@suse.de>
23302 PR tree-optimization/80887
23303 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
23304 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
23305 simplified lookups, then reset mprts_hook.
23306 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
23308 (try_to_simplify): Likewise.
23310 2017-06-19 Martin Liska <mliska@suse.cz>
23313 * gimplify.c (gimplify_switch_expr):
23314 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
23316 2017-06-19 Martin Liska <mliska@suse.cz>
23318 * doc/install.texi: Document that PGO runs in 4 stages.
23320 2017-06-19 Martin Liska <mliska@suse.cz>
23323 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
23324 to dispatcher function name.
23325 * multiple_target.c (replace_function_decl): New function.
23326 (create_dispatcher_calls): Redirect both edges and references.
23328 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
23330 * profile-count.c (profile_count::dump): Dump quality.
23331 (profile_count::differs_from_p): Update for unsigned val.
23332 * profile-count.h (profile_count_quality): New enum.
23333 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
23335 2017-06-19 Richard Biener <rguenther@suse.de>
23337 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
23338 struct function as arg.
23339 (estimate_numbers_of_iterations): Export overload with loop arg.
23340 (free_numbers_of_iterations_estimates_loop): Use an overload of
23341 free_numbers_of_iterations_estimates instead.
23342 * tree-cfg.c (remove_bb): Adjust.
23343 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
23344 * tree-parloops.c (gen_parallel_loop): Likewise.
23345 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
23347 (tree_unroll_loops_completely): Likewise.
23348 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
23349 Use an overload instead and export.
23350 (estimated_loop_iterations): Adjust.
23351 (max_loop_iterations): Likewise.
23352 (likely_max_loop_iterations): Likewise.
23353 (estimate_numbers_of_iterations): Take struct function as arg
23355 (loop_exits_before_overflow): Adjust.
23356 (free_numbers_of_iterations_estimates_loop): Use an overload.
23357 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
23358 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
23360 2017-06-19 Richard Biener <rguenther@suse.de>
23363 * ipa-prop.c (find_constructor_constant_at_offset): Handle
23364 RANGE_EXPR conservatively.
23366 2017-06-16 Carl Love <cel@us.ibm.com>
23368 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23369 definitions for vec_float, vec_float2, vec_floato,
23370 vec_floate built-ins.
23371 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
23372 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
23374 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
23375 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
23376 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
23377 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
23378 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
23379 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
23380 vec_floato): Add builtin defines.
23381 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
23382 Update the built-in documentation file for the new built-in
23385 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23387 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
23388 (mthumb): Mark as the negative of -marm.
23390 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23392 * doc/invoke.texi (ARM Options, -mcpu): Document supported
23394 (ARM Options, -mtune): Document that this accepts the same
23395 extension options as -mcpu.
23396 (ARM Options, -mfpu): Document addition of -mfpu=auto.
23398 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23400 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
23401 permitted extensions.
23403 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23405 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
23406 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
23407 (armv8-m.main): Add option +nodsp.
23408 * config/arm/arm-cpu-cdata.h: Regenerated.
23410 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23412 * config/arm/t-fuchsia: New file.
23413 * config.gcc (arm*-*-fuchsia*): Use it.
23415 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23417 * config/arm/t-symbian: Rewrite for new option infrastructure.
23419 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23421 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
23422 (MULTILIB_REQUIRED): Likewise.
23424 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23426 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
23427 (MULTILIB_RESUE): Likewise.
23428 (MULTILIB_MATCHES): Likewise.
23429 (MULTLIB_REQUIRED): Likewise.
23431 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23433 * config/arm/t-rtems: Rewrite for new option framework.
23435 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23437 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
23438 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
23439 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
23440 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
23441 * config/arm/t-multilib: ... here.
23442 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
23443 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
23444 armv7-a and armv8*-a when A-profile libraries have not been built.
23445 * config/arm/t-rmprofile: Rewrite.
23447 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23449 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
23450 with a backslash. Remove the backslash after substituting unescaped
23452 * doc/fragments.texi (MULTILIB_REUSE): Document it.
23454 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23456 * config.gcc: (arm*-*-*): When building a-profile libraries, force
23457 the driver to pass through the default setting of -mfloat-abi.
23458 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
23460 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
23461 (all_feat_combs): New rule.
23462 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
23464 * config/arm/t-aprofile: Rewrite.
23466 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23468 * config/arm/arm.h (FPUTYPE_AUTO): Define.
23469 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
23470 fpu is not specified by the user/command-line.
23471 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
23472 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
23473 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
23474 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
23475 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
23476 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
23478 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23480 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
23481 * config/arm/t-arm-elf: Rewritten.
23483 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23485 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
23486 have some floating-point instructions.
23487 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
23488 (TARGET_MAYBE_HARD_FLOAT): New macro.
23489 * config/arm/arm-builtins.c (arm_init_builtins): Use
23490 TARGET_MAYBE_HARD_FLOAT.
23491 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
23493 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23495 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
23496 (configargs.h): Include it.
23497 (arm_print_hint_for_fpu_option): New function.
23498 (arm_parse_fpu_option): New function.
23499 (candidate_extension): New class.
23500 (arm_canon_for_multilib): New function.
23501 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
23502 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
23503 (ARCH_CANONICAL_SPECS): New macro.
23504 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
23506 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23508 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
23509 are set after handling multilib fragments. Set target_cpu_default2
23512 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23514 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
23516 (arm*-*-*): Set target_cpu_default2 to a quoted string.
23517 * config/arm/parsecpu.awk (check_cpu): Validate any extension
23519 (check_arch): Likewise.
23520 * config/arm/arm.c (arm_configure_build_target): Handle
23521 TARGET_CPU_DEFAULT being a string constant. Scan any feature
23522 options in the default.
23524 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23526 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
23527 when an option is an alias of another.
23528 * config/arm/parsecpu.awk (optalias): New parser token.
23529 (gen_comm_data): Mark non-alias options as such. Emit entries
23530 for extension aliases.
23531 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
23532 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
23533 (armv6kz, armv6zk, armv6t2): Likewise.
23534 (armv7): Make vfpv3-d16 an alias.
23535 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
23537 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
23538 Sort in canonical order.
23539 (armv8-a): Sort in canonical order.
23540 (armv8.1-a, armv8.2-a): Likewise.
23541 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
23543 (cortex-a9): Sort in canonical order.
23544 * config/arm/arm.c (selftests.h): Include it.
23545 (arm_test_cpu_arch_data): New function.
23546 (arm_run_self_tests): New function.
23547 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
23548 (targetm): Move declaration to the end of the file.
23549 * arm-cpu-cdata.h: Regenerated.
23551 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23553 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
23554 call to target_mode_check describing the type of option passed.
23555 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
23556 (arm_target_thumb_only): Use arm_parse_arch_option_name or
23557 arm_parse_cpu_option_name to match parameters against list of
23559 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
23560 arm_arch_core_flags data structure.
23561 * config/arm/arm-cpu_cdata.h: Regenerated.
23563 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23565 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
23567 (arm_print_hint_for_cpu_option): Likewise.
23568 (arm_print_hint_for_arch_option): Likewise.
23569 (arm_parse_cpu_option_name): Likewise.
23570 (arm_parse_arch_option_name): Likewise.
23571 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
23572 of entries in the all_fpus list.
23573 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
23574 (arm_parse_cpu_option_name): Declare.
23575 (arm_parse_arch_option_name): Declare.
23576 (arm_parse_option_features): Declare.
23577 (arm_intialize_isa): Declare.
23578 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
23580 (gen_comm_data): ... here. Make definitions non-static.
23581 * config/arm/arm-cpu-data.h: Regenerated.
23582 * config/arm/arm-cpu-cdata.h: Regenerated.
23584 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23586 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
23587 (cpu_arch_extension): New structure.
23588 (cpu_arch_option, arch_option, cpu_option): New structures.
23589 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
23590 architecture types.
23591 (gen_data): Generate new format data tables.
23592 * config/arm/arm.c (cpu_tune): New structure.
23593 (cpu_option, processors): Delete.
23594 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
23595 (arm_print_hint_for_cpu_option): ... this and ...
23596 (arm_print_hint_for_arch_option): ... this.
23597 (arm_parse_arch_cpu_name): Delete. Replace with ...
23598 (arm_parse_cpu_option_name): ... this and ...
23599 (arm_parse_arch_option_name): ... this.
23600 (arm_unrecognized_feature): Change type of target parameter to
23602 (arm_parse_arch_cpu_features): Delete. Replace with ...
23603 (arm_parse_option_features): ... this.
23604 (arm_configure_build_target): Rework to use new configuration data
23606 (arm_print_tune_info): Rework for new configuration data tables.
23607 * config/arm/arm-cpu-data.h: Regenerated.
23608 * config/arm/arm-cpu.h: Regenerated.
23610 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23612 * Makefile.in (OBJS): Move sbitmap.o from here ...
23613 (OBJS-libcommon): ... to here.
23615 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23617 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
23618 (ISA_ALL_CRYPTO): New macro.
23619 (ISA_ALL_SIMD): New macro
23620 (ISA_ALL_FP): New macro.
23621 * config/arm/arm.c (fpu_bitlist): Update initializer.
23622 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
23624 (arm9e): Add fpu. Add option for nofp
23625 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
23626 (arm926ej-s, arm1026ej-s): Likewise.
23627 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
23628 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
23629 neon-fp16, neon-vfpv4, nofp and nosimd.
23630 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
23631 (cortex-a8): Add fpu. Add option for nofp.
23632 (cortex-a9): Add fpu. Add options for nosimd and nofp.
23633 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
23634 (cortex-r4f): Add fpu.
23635 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
23636 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
23638 (cortex-r8): Likewise.
23639 (cortex-m4): Add fpu. Add option for nofp.
23640 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
23641 (cortex-a17.cortex-a7): Likewise.
23642 (cortex-a32): Add fpu. Add options for crypto and nofp.
23643 (cortex-a35, cortex-a53): Likewise.
23644 (cortex-a57): Add fpu. Add option for crypto.
23645 (cortex-a72, cortex-a73): Likewise.
23646 (exynos-m1): Likewise.
23647 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
23648 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
23649 (cortex-m33): Add fpu. Add option for nofp.
23650 * config/arm/arm-cpu-cdata.h: Regenerated
23651 * config/arm/arm-cpu-data.h: Regenerated.
23653 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23655 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
23656 (armv5te, armv5tej): Likewise.
23657 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
23658 (armv7): Add options fp and vfpv3-d16.
23659 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
23660 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
23662 (armv7ve): Likewise.
23663 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
23664 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
23665 (armv8-a): Add nocrypto option.
23666 (armv8.1-a, armv8.2-a): Likewise.
23667 (armv8-m.main): add options fp, fp.dp and nofp.
23669 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23671 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
23673 (armv8-a+crc): Delete.
23674 (armv8.1-a): Add options simd, crypto and nofp.
23675 (armv8.2-a): Add options fp16, simd, crypto and nofp.
23676 (armv8.2-a+fp16): Delete.
23677 (armv8-m.main): Add option dsp.
23678 (armv8-m.main+dsp): Delete.
23679 (cortex-a8): Add fpu. Add nofp option.
23680 (cortex-a9): Add fpu. Add nofp and nosimd options.
23681 * config/arm/parsecpu.awk (gen_data): Generate option tables and
23682 link to main cpu and architecture data structures.
23683 (gen_comm_data): Only put isa attributes from the main architecture
23685 (option): New statement for architecture and CPU entries.
23686 * arm.c (struct cpu_option): New structure.
23687 (struct processors): Add entry for options.
23688 (arm_unrecognized_feature): New function.
23689 (arm_parse_arch_cpu_name): Ignore any characters after the first
23691 (arm_parse_arch_cpu_feature): New function.
23692 (arm_configure_build_target): Separate out any CPU and architecture
23693 features and parse separately. Don't error out if -mfpu=auto is
23694 used with only an architecture string.
23695 (arm_print_asm_arch_directives): New function.
23696 (arm_file_start): Call it.
23697 * config/arm/arm-cpu-cdata.h: Regenerated.
23698 * config/arm/arm-cpu-data.h: Likewise.
23699 * config/arm/arm-tables.opt: Likewise.
23701 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23703 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
23704 assembler when it is not -mfpu=auto.
23706 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23708 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
23709 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
23710 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
23711 (ASM_CPU_SPEC): Rewrite.
23712 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
23713 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
23714 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
23715 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
23716 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
23717 copied string is NUL-terminated. Also strip any characters prefixed
23719 (arm_rewrite_selected_arch): New function.
23720 (arm_rewrite_march): New function.
23722 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23724 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
23725 (x_arm_cpu_string, x_arm_tune_string): Likewise.
23726 (march, mcpu, mtune): Convert to string-based options.
23727 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
23728 (arm_parse_arch_cpu_name): New function.
23729 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
23730 identify selected architecture or CPU.
23731 (arm_option_save): New function.
23732 (TARGET_OPTION_SAVE): Redefine.
23733 (arm_option_restore): Restore string options.
23734 (arm_option_print): Print string options.
23736 2017-06-16 Martin Sebor <msebor@redhat.com>
23738 PR tree-optimization/80933
23739 PR tree-optimization/80934
23740 * builtins.c (fold_builtin_3): Do not handle bcmp here.
23741 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
23742 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
23743 (gimple_fold_builtin): Call them.
23745 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23747 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
23748 as unlikely; update profile.
23750 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23752 * predict.c (force_edge_cold): Handle declaring edges impossible
23755 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23757 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
23759 (try_unroll_loop_completely): Fix reporting.
23761 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23763 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
23765 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
23768 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
23769 if given a non-constant argument for an intrinsic which requires a
23772 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23774 * profile.c (compare_freqs): New function.
23775 (branch_prob): Sort edge list.
23776 (find_spanning_tree): Assume that the list is priority sorted.
23778 2017-06-16 Richard Biener <rguenther@suse.de>
23780 PR tree-optimization/81090
23781 * passes.def (pass_record_bounds): Remove.
23782 * tree-pass.h (make_pass_record_bounds): Likewise.
23783 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
23784 make_pass_record_bounds): Likewise.
23785 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
23786 not free niter estimates at the beginning but at the end.
23787 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
23789 2017-06-16 Richard Biener <rguenther@suse.de>
23791 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
23792 initializer to workaround ICE in host GCC 4.8.
23794 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23796 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
23798 (clone_inlined_nodes): Update.
23800 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23802 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
23803 prefetch settings, and enable prefetching by default at -O3.
23805 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23807 * config/aarch64/aarch64.c (aarch64_override_options_internal):
23808 Set flag_prefetch_loop_arrays according to tuning data.
23810 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23812 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
23813 New tune structure.
23814 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
23815 [Unrelated to main purpose of the patch] Place the pointer field last
23816 to enable type checking errors when tune structure are wrongly merged.
23817 * config/aarch64/aarch64.c (generic_prefetch_tune,)
23818 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
23819 (thunderx2t99_prefetch_tune): New tune constants.
23820 (tune_params *_tunings): Update all tunings (no functional change).
23821 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
23822 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
23823 from tunings structures.
23825 2017-06-16 Jakub Jelinek <jakub@redhat.com>
23828 * ubsan.c (instrument_null): Add T argument, use it instead
23829 of computing it based on IS_LHS.
23830 (instrument_object_size): Likewise.
23831 (pass_ubsan::execute): Adjust instrument_null and
23832 instrument_object_size callers to pass gimple_get_lhs or
23833 gimple_assign_rhs1 result to it. Use instrument_null instead of
23834 calling get_base_address and instrument_mem_ref. Handle
23835 aggregate call arguments for object-size sanitization.
23837 2017-06-16 Yury Gribov <tetra2005@gmail.com>
23839 PR tree-optimization/81089
23840 * tree-vrp.c (is_masked_range_test): Validate operands of
23843 2017-06-15 Martin Sebor <msebor@redhat.com>
23846 * dumpfile.c (dump_register): Avoid calling memset to initialize
23847 a class with a default ctor.
23848 * gcc.c (struct compiler): Remove const qualification.
23849 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
23850 * hash-table.h: Ditto.
23851 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
23853 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
23854 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
23856 * params.h (struct param_info): Make struct members non-const.
23857 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
23858 with default initialization.
23859 * vec.h (vec_copy_construct, vec_default_construct): New helper
23861 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
23862 with vec_copy_construct.
23863 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
23864 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
23865 * doc/invoke.texi (-Wclass-memaccess): Document.
23867 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23869 * emit-rtl.h (is_leaf): Update comment about local
23870 register allocator.
23872 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23875 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
23876 for a variable to have a section before checking if the section has a
23878 Set section to.persistent if persistent attribute is set.
23879 Warn if .persistent attribute is used on an automatic variable.
23881 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
23883 PR rtl-optimization/80474
23884 * reorg.c (update_block): Do not ignore instructions in a delay slot.
23886 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
23888 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
23891 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
23893 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
23894 (casesi): Emit bounds checking as RTL.
23895 (casesi_internal_mips16_<mode>): Remove bounds checking.
23897 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
23899 * config/xtensa/xtensa.c (xtensa_option_override): Append
23900 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
23901 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
23902 xtensa_doloop_hooks): Define unconditionally.
23903 (xtensa_reorg_loops): Only call reorg_loops in the presence of
23905 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
23906 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
23907 for it in xtensa_option_override.
23908 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
23909 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
23911 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
23913 * doc/cppopts.texi: Document '-' special value to -MF.
23915 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
23917 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
23918 (cortex_a53_fconst): Likewise.
23919 (cortex_a53_fpmul): Likewise.
23920 (cortex_a53_f_load_64): Likewise.
23921 (cortex_a53_f_load_many): Likewise.
23922 (cortex_a53_advsimd_alu): Likewise.
23923 (cortex_a53_advsimd_alu_q): Likewise.
23924 (cortex_a53_advsimd_mul): Likewise.
23925 (cortex_a53_advsimd_mul_q): Likewise.
23926 (fpmac bypass): Add new bypass for fpmac-fpmac case.
23927 Add missing fmul, r2f_cvt and fconst cases.
23929 2017-06-14 Richard Biener <rguenther@suse.de>
23931 PR middle-end/81088
23932 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
23934 (fold_binary_loc): When associating do not treat pre-existing
23935 TREE_OVERFLOW on literal constants as a reason to allow
23936 TREE_OVERFLOW on associated literal constants.
23938 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23940 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
23941 (MASK_FEATURES): New macro.
23942 * config/sparc/sparc.c (sparc_option_override): Remove the special
23943 handling of -mfpu and generalize it to all MASK_FEATURES switches.
23945 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23947 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
23948 a division of 0 if non-call exceptions are enabled.
23950 2017-06-14 Andrew Pinski <apinski@cavium.com>
23951 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23954 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
23955 Improve vector initialization code gen for only variable case.
23957 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23959 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
23961 2017-06-14 Richard Biener <rguenther@suse.de>
23963 PR tree-optimization/81083
23964 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
23967 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23969 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
23970 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
23971 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
23972 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
23973 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
23974 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
23976 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23978 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
23979 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
23981 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23983 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
23985 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23987 * config/rs6000/t-rtems: Don't handle SPE.
23989 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23991 * config/rs6000/t-linux: Don't handle SPE.
23993 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23995 * config/rs6000/eabispe.h: Delete file.
23997 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23999 * config/rs6000/t-spe: Delete file.
24001 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24003 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
24004 (rs6000_legitimate_offset_address_p): Return false for anything in
24005 V2SImode or V2SFmode.
24007 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
24009 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
24010 except V2SF and V2SI. Rearrange the vector modes, and add comments.
24011 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
24013 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
24014 (rs6000_legitimate_offset_address_p): Ditto.
24015 (rs6000_emit_move): Ditto.
24016 (rs6000_init_builtins): Remove V4HI_type_node.
24018 2017-06-13 Martin Liska <mliska@suse.cz>
24021 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
24022 (gate_asan): Likewise.
24023 * asan.h (asan_no_sanitize_address_p): Remove the function.
24024 (sanitize_flags_p): New function.
24025 * builtins.def: Fix coding style.
24026 * common.opt: Use renamed enum value.
24027 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
24028 * doc/extend.texi: Document no_sanitize attribute.
24029 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
24030 to SANITIZE_UNDEFINED_NONDEFAULT.
24031 * gcc.c (sanitize_spec_function): Use the renamed enum value.
24032 * gimple-fold.c (optimize_atomic_compare_exchange_p):
24033 Use sanitize_flags_p.
24034 * gimplify.c (gimplify_function_tree): Likewise.
24035 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
24036 * opts.c (parse_no_sanitize_attribute): New function.
24037 (common_handle_option): Use renamed enum value.
24038 * opts.h (parse_no_sanitize_attribute): Declare.
24039 * tree.c (sanitize_flags_p): New function.
24040 * tree.h: Declared here.
24041 * tsan.c: Use sanitize_flags_p.
24042 * ubsan.c (ubsan_expand_null_ifn): Likewise.
24043 (instrument_mem_ref): Likewise.
24044 (instrument_bool_enum_load): Likewise.
24045 (do_ubsan_in_current_function): Remove the function.
24046 (pass_ubsan::execute): Use sanitize_flags_p.
24047 * ubsan.h: Remove do_ubsan_in_current_function
24048 * tree-cfg.c (print_no_sanitize_attr_value): New function.
24049 (dump_function_to_file): Use it here.
24051 2017-06-13 Martin Jambor <mjambor@suse.cz>
24053 PR tree-optimization/80803
24054 PR tree-optimization/81063
24055 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24056 (propagate_subaccesses_across_link): Enqueue subtree whenever
24057 necessary instead of relying on the caller.
24059 2017-06-13 Martin Jambor <mjambor@suse.cz>
24061 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24062 that have a first_link.
24063 (sort_and_splice_var_accesses): Do not check first_link before
24065 (subtree_mark_written_and_enqueue): Likewise.
24066 (propagate_all_subaccesses): Likewise and do not stop at first
24067 parent with a first_link.
24069 2017-06-13 Martin Jambor <mjambor@suse.cz>
24071 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24074 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24076 * match.pd: New pattern.
24078 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24080 * tree-vrp.c (is_masked_range_test): New function.
24081 (register_edge_assert_for): Determine ranges for
24084 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24086 PR tree-optimization/67328
24087 * fold-const.c (maskable_range_p): New function.
24088 (build_range_check): Generate bittests if possible.
24090 2017-06-13 Martin Liska <mliska@suse.cz>
24092 * gimple-pretty-print.c (dump_probability): Add new argument.
24093 (dump_edge_probability): Dump both probability and count.
24094 (dump_gimple_label): Likewise.
24095 (dump_gimple_bb_header): Likewise.
24097 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24100 * config/avr/avr-devices.c: Fix indentation.
24101 * config/avr/gen-avr-mmcu-specs.c: Dito.
24103 2017-06-13 Richard Biener <rguenther@suse.de>
24105 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24106 instead get vector type from stmt_info.
24107 (vectorizable_reduction): Adjust. Remove dead code.
24109 2017-06-13 Richard Biener <rguenther@suse.de>
24111 PR middle-end/81065
24112 * fold-const.c (extract_muldiv_1): Remove bogus distribution
24113 case of C * (x * C2 + C3).
24114 (fold_addr_of_array_ref_difference): Properly fold index difference.
24116 2017-06-12 David S. Miller <davem@davemloft.net>
24119 * config/sparc/sparc.md (return expander): Emit frame blockage if
24120 function uses alloca.
24122 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
24124 * combine.c (make_field_assignment): Check len rather than the mode
24125 precision when calling force_to_mode.
24127 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
24129 Support multilibs and devices that see flash in RAM address range.
24132 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
24133 (avr_mcu_t) <flash_pm_offset>: New field.
24134 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
24135 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
24136 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
24137 (AVR_TINY_PM_OFFSET): Remove macro.
24138 * config/avr/avr.opt (-mshort-calls): New option.
24139 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24140 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
24141 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
24142 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
24143 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
24144 instead of avr_arch->have_jmp_call.
24145 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
24146 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
24147 avr_arch->flash_pm_offset to define.
24148 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
24149 new field flash_pm_offset. Add entry for avrxmega3.
24150 (avr_texinfo): Add entry for avrxmega3.
24151 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
24152 attiny212, attiny214,
24153 attiny412, attiny414, attiny416, attiny417,
24154 attiny814, attiny816, attiny817,
24155 attiny1614, attiny1616, attiny1617,
24156 attiny3214, attiny3216, attiny3217.
24157 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
24158 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
24159 (avr_print_operand_address) [AVR_TINY]: Same.
24160 (avr_asm_init_sections) <readonly_data_section>: Only patch
24161 callback if avr_arch->flash_pm_offset = 0.
24162 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
24163 for rodata if avr_arch->flash_pm_offset != 0.
24164 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
24165 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
24166 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
24167 (m_options): Append opt_rcall.
24168 (m_dirnames): Append dir_rcall.
24169 * config/avr/t-multilib: Regenerate.
24171 * configure.ac [target=avr]: Check whether avrxmega3 default
24172 linker description file works as needed.
24173 * configure: Regenerate.
24174 * doc/avr-mmcu.texi: Regenerate.
24175 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
24176 <__AVR_ARCH__>: Document avrxmega3 and 103.
24177 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
24178 <__AVR_SHORT_CALLS__>: Document it.
24179 <__AVR_PM_BASE_ADDRESS__>: Document it.
24180 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
24181 (AVR Variable Attributes) <progmem>: Document this is
24182 not needed for avrxmega3.
24183 (AVR Named Address Spaces) <__flash>: Dito.
24185 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
24187 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
24189 2017-06-12 Doug Rupp <rupp@adacore.com>
24191 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
24192 Append vxworks-stdint.h to the tm_file list.
24193 * config/vxworks-stdint.h: New file.
24195 2017-06-12 Martin Liska <mliska@suse.cz>
24197 PR tree-optimization/81041
24198 * tree-profile.c (gimple_gen_ic_func_profiler):
24199 Create an extra BB in profile-generate
24200 (gimple_gen_time_profiler): Likewise.
24202 2017-06-12 Jakub Jelinek <jakub@redhat.com>
24204 PR tree-optimization/81003
24205 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
24206 (update_range_test): Use it instead of force_gimple_operand_gsi.
24208 2017-06-12 Richard Biener <rguenther@suse.de>
24210 PR tree-optimization/81053
24211 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
24212 with backedge value not defined in loop. Simplify def stmt
24215 2017-06-11 Tom de Vries <tom@codesourcery.com>
24218 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
24220 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
24221 nvptx_cannot_force_const_mem.
24223 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24225 * opts.c (finish_options): Move test for flag_split_stack after
24226 it has been initialized.
24228 2017-06-11 Jason Merrill <jason@redhat.com>
24230 * tree.h (id_equal): New.
24231 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
24232 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
24233 instead of strcmp of IDENTIFIER_POINTER.
24235 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24237 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
24238 (mark_all_inlined_calls_cdtor): Fix formating.
24239 (inline_transform): Rescale profile before inlining.
24241 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24243 * cgraph.h (cgraph_edge::clone): Update prototype.
24244 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
24245 (cgraph_node::create_clone): Update.
24246 (cgraph_node::create_version_clone): Update.
24247 * tree-inline.c (copy_bb): Update.
24248 (expand_call_inline): Update.
24250 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24252 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
24253 factored out from ...
24254 (rs6000_emit_prologue): ... here.
24256 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24258 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
24259 factored out from ...
24260 (rs6000_emit_prologue): ... here.
24262 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24264 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
24266 (handle_missing_profiles): Fix computation of tp_first_run.
24267 (counts_to_freqs): Do not touch freqs when count is 0.
24269 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24271 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
24274 2017-06-10 Tom de Vries <tom@codesourcery.com>
24276 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24277 attributes): Document signal effective target.
24279 2017-06-10 Tom de Vries <tom@codesourcery.com>
24281 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24282 Document effective target stack_size.
24284 2017-06-09 David Malcolm <dmalcolm@redhat.com>
24286 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
24287 to the edit_context if they can be auto-applied.
24289 2017-06-9 Ian Lance Taylor <iant@golang.org>
24291 * opts.c (finish_options): If -fsplit-stack, disable implicit
24292 -forder-blocks-and-partition.
24293 * doc/invoke.texi (Optimize Options): Document that when using
24294 -fsplit-stack -forder-blocks-and-partition is not implicitly
24297 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24299 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
24300 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
24301 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
24302 * builtins.def (abort, trap, unreachable): Declare cold.
24303 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
24304 * tree-core.h (ECF_COLD): New.
24305 * tree.c (set_call_expr_flags): Handle ECF_COLD.
24306 (build_common_builtin_nodes): Mark unreachable and abort as cold.
24308 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24310 * predict.c (unlikely_executed_stmt_p): Cleanup.
24312 2017-06-09 Richard Biener <rguenther@suse.de>
24314 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
24315 model if the ref is always written to.
24317 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24319 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
24321 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24323 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
24326 2017-06-09 Tom de Vries <tom@codesourcery.com>
24329 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
24330 "target cannot support label values" when encountering LABEL_REF.
24332 2017-06-09 Martin Liska <mliska@suse.cz>
24334 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
24335 (gimple_gen_ic_func_profiler): Emit direct comparison
24336 of __gcov_indirect_call_callee with NULL.
24337 (gimple_gen_time_profiler): Change probability from
24338 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
24340 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24342 * profile.c (edge_gcov_counts): Turn to pointer.
24343 (compute_branch_probabilities, compute_branch_probabilities): Update.
24344 (branch_prob): Do not clear edge_gcov_count.
24345 * profile.h (edge_gcov_counts): Turn to pointer.
24346 (edge_gcov_count): Update.
24348 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24350 * gimple.h (gimple_check_failed): Mark cold.
24352 2017-06-09 Richard Biener <rguenther@suse.de>
24354 PR tree-optimization/66623
24355 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
24356 refactor check_reduction into two parts, properly computing
24357 whether we have to check reduction validity for outer loop
24360 2017-06-09 Richard Biener <rguenther@suse.de>
24362 PR tree-optimization/79483
24363 * graphite-scop-detection.c (order): New global.
24364 (get_order): Compute bb to order mapping that satisfies code
24365 generation constraints.
24366 (cmp_pbbs): New helper.
24367 (build_scops): Start domwalk at entry block, sort generated
24370 2017-06-09 Richard Biener <rguenther@suse.de>
24372 PR middle-end/81007
24373 * ipa-polymorphic-call.c
24374 (ipa_polymorphic_call_context::restrict_to_inner_class):
24375 Skip FIELD_DECLs with error_mark_node type.
24376 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
24379 2017-06-09 Martin Liska <mliska@suse.cz>
24381 * predict.c (struct branch_predictor): New struct.
24382 (test_prediction_value_range): New test.
24383 (predict_c_tests): New function.
24384 * selftest-run-tests.c (selftest::run_tests): Run the function.
24385 * selftest.h: Declare new tests.
24387 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
24390 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
24391 gen_add3_insn did not fail.
24392 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
24393 r0, construct that number in a temporary reg and add that reg to r0.
24394 If asked to put the result in r0 as well, fail.
24396 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
24398 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24399 for early expansion of vec_eqv.
24401 2017-06-08 Jakub Jelinek <jakub@redhat.com>
24403 PR middle-end/81005
24404 * ubsan.c (instrument_null): Avoid pointless code temporary.
24405 (pass_ubsan::execute): Instrument aggregate arguments of calls.
24407 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
24411 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
24414 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
24417 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24419 * predict.c (unlikely_executed_edge_p): Move ahead.
24420 (probably_never_executed_edge_p): Use it.
24422 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24424 PR middle-end/79988
24425 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
24426 gimple_call_builtin_p call.
24428 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24430 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
24431 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
24432 rtl_check_failed_type2, rtl_check_failed_code1,
24433 rtl_check_failed_code2, rtl_check_failed_code_mode,
24434 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
24435 rtvec_check_failed_bounds, rtl_check_failed_flag,
24436 _fatal_insn_not_found, _fatal_insn): Likewise.
24437 * tree.h (tree_contains_struct_check_failed,
24438 tree_check_failed, tree_not_check_failed,
24439 tree_class_check_failed, tree_range_check_failed,
24440 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
24441 tree_vec_elt_check_failed, phi_node_elt_check_failed,
24442 tree_operand_check_failed, omp_clause_check_failed,
24443 omp_clause_operand_check_failed, omp_clause_range_check_failed):
24446 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24448 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
24449 flag_branch_probabilities.
24450 * ipa-inline.c (edge_badness): Likewise.
24451 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
24452 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
24453 * predict.c (maybe_hot_frequency_p): Likewise.
24454 (probably_never_executed): Likewise.
24455 * sched-ebb.c (schedule_ebbs): Likewise.
24456 * sched-rgn.c (find_single_block_region): Likewise.
24457 * tracer.c (tail_duplicate): Likewise.
24459 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24461 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
24462 longer requires x_flag_profile_use.
24464 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24466 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
24467 instead of flag_reorder_blocks_and_partition.
24468 * dbxout.c (dbxout_function_end): Likewise.
24469 * dwarf2out.c (gen_subprogram_die): Likewise.
24470 * haifa-sched.c (sched_create_recovery_edges): Likewise.
24471 * hw-doloop.c (reorg_loops): Likewise.
24472 * varasm.c (assemble_start_function,
24473 assemble_end_function): Likewise.
24474 (decide_function_section): Do not check for
24475 flag_reorder_blocks_and_partition.
24477 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24479 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24481 (chkp_get_hard_register_fake_addr_expr): Ditto.
24482 (chkp_build_addr_expr): Add check for hard reg case.
24483 (chkp_parse_array_and_component_ref): Ditto.
24484 (chkp_find_bounds_1): Ditto.
24485 (chkp_process_stmt): Don't generate bounds store for
24488 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24490 * predict.c (maybe_hot_bb_p): Do not check profile status.
24491 (maybe_hot_edge_p): Likewise.
24492 (probably_never_executed): Check for zero counts even if profile
24494 (unlikely_executed_edge_p): New function.
24495 (unlikely_executed_stmt_p): New function.
24496 (unlikely_executed_bb_p): New function.
24497 (set_even_probabilities): Use unlikely predicates.
24498 (combine_predictions_for_bb): Likewise.
24499 (predict_paths_for_bb): Likewise.
24500 (predict_paths_leading_to_edge): Likewise.
24501 (determine_unlikely_bbs): New function.
24502 (estimate_bb_frequencies): Use it.
24503 (compute_function_frequency): Use zero counts even if profile is
24505 * profile-count.h: Fix typo.
24507 2017-08-08 Julia Koval <julia.koval@intel.com>
24509 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
24510 _mm512_mask_cvtsepi16_storeu_epi8,
24511 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
24512 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
24513 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
24514 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
24515 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
24516 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
24517 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
24518 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
24519 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
24520 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
24521 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
24522 __builtin_ia32_pmovuswb256mem_mask,
24523 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
24524 __builtin_ia32_pmovwb512mem_mask): New builtins.
24526 2017-08-08 Julia Koval <julia.koval@intel.com>
24528 PR target/73350,80862
24529 * config/i386/subst.md (round): Fix round pattern.
24530 * config/i386/i386.c (ix86_erase_embedded_rounding):
24531 Fix erasing rounding for the fixed pattern.
24533 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24535 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
24537 2017-06-08 Martin Liska <mliska@suse.cz>
24539 PR gcov-profile/80911
24540 * gcov.c (block_info::block_info): New constructor.
24542 2017-06-07 Carl Love <cel@us.ibm.com>
24544 * config/rs6000/rs6000-c: The return type of the following
24545 built-in functions was implemented as int not long long. Fix sign
24546 of return value for the unsigned version of vec_mulo and vec_mule.
24547 vector unsigned long long vec_bperm (vector unsigned long long,
24548 vector unsigned char)
24549 vector signed long long vec_mule (vector signed int,
24551 vector unsigned long long vec_mule (vector unsigned int,
24552 vector unsigned int)
24553 vector signed long long vec_mulo (vector signed int,
24555 vector unsigned long long vec_mulo (vector unsigned int,
24556 vector unsigned int)
24557 * doc/extend.texi: Fix the documentation for the built-in
24560 2017-06-07 Carl Love <cel@us.ibm.com>
24563 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
24566 2017-06-07 Carl Love <cel@us.ibm.com>
24568 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
24569 support, Generate doublehv for signed int/float for BE case only.
24571 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
24573 * doc/invoke.texi (mcx16): Rewrite.
24575 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24577 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
24578 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
24579 *mov<mode>_softfloat, and an anonymous splitter): Use
24580 nonimmediate_operand instead of rs6000_nonimmediate_operand.
24582 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24584 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
24586 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24587 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
24588 (rs6000_debug_reg_global): Adjust.
24589 (rs6000_init_hard_regno_mode_ok): Adjust.
24590 (rs6000_dbx_register_number): Adjust.
24591 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
24592 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24593 Remove SPE_ACC and SPEFSCR.
24594 (REG_ALLOC_ORDER): Ditto.
24595 (FRAME_POINTER_REGNUM): Change to 111.
24596 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
24597 (REG_CLASS_NAMES): Ditto.
24598 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
24599 (REGISTER_NAMES): Ditto.
24600 (ADDITIONAL_REG_NAMES): Ditto.
24601 (rs6000_reg_names): Ditto.
24602 * config/rs6000/rs6000.md: Renumber some register number
24605 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24607 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
24609 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24610 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
24612 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
24613 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24614 Delete the SPE high registers.
24615 (REG_ALLOC_ORDER): Ditto.
24616 (enum reg_class): Remove SPE_HIGH_REGS.
24617 (REG_CLASS_NAMES): Ditto.
24618 (REG_CLASS_CONTENTS): Delete the SPE high registers.
24619 (REGISTER_NAMES): Ditto.
24620 (rs6000_reg_names): Ditto.
24621 * doc/tm.texi.in: Remove SPE as example.
24622 * doc/tm.texi: Regenerate.
24624 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24626 * config/rs6000/8540.md (ppc8540_brinc): Delete.
24627 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
24628 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
24629 * config/rs6000/rs6000.md (type): Remove "brinc".
24631 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24633 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
24634 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
24635 * config/rs6000/linuxspe.h: Delete file.
24636 * config/rs6000/rs6000.md: Don't include spe.md.
24637 * config/rs6000/spe.h: Delete file.
24638 * config/rs6000/spe.md: Delete file.
24639 * config/rs6000/t-rs6000: Remove spe.md.
24641 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24643 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
24644 (reg_or_none500mem_operand): Delete.
24645 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
24646 instead of reg_or_none500mem_operand.
24648 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24650 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24651 handling of SPE flags.
24652 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
24654 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24656 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
24658 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
24659 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
24660 paired_divv2sf3): Similar.
24661 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
24662 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
24663 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
24665 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
24666 Rename the paired_* instruction patterns.
24667 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
24669 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
24670 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
24671 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
24672 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
24673 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
24674 PAIRED_VECTOR_MODE.
24675 (struct machine_function): Delete field spe_insn_chain_scanned_p.
24676 (spe_func_has_64bit_regs_p): Delete.
24677 (spe_expand_predicate_builtin): Delete.
24678 (spe_expand_evsel_builtin): Delete.
24679 (TARGET_DWARF_REGISTER_SPAN): Do not define.
24680 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
24681 (invalid_e500_subreg): Delete.
24682 (rs6000_legitimize_address): Always force_reg op2 as well, for
24683 paired single memory accesses.
24684 (rs6000_member_type_forces_blk): Delete.
24685 (rs6000_spe_function_arg): Delete.
24686 (rs6000_expand_unop_builtin): Delete SPE handling.
24687 (rs6000_expand_binop_builtin): Ditto.
24688 (spe_expand_stv_builtin): Delete.
24689 (bdesc_2arg_spe): Delete.
24690 (spe_expand_builtin): Delete.
24691 (spe_expand_predicate_builtin): Delete.
24692 (spe_expand_evsel_builtin): Delete.
24693 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
24694 (spe_init_builtins): Delete.
24695 (spe_func_has_64bit_regs_p): Delete.
24696 (savres_routine_name): Delete "info" parameter. Adjust callers.
24697 (rs6000_emit_stack_reset): Ditto.
24698 (rs6000_dwarf_register_span): Delete.
24699 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
24700 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
24701 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
24703 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
24705 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
24706 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
24707 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
24708 mulv2sf3, divv2sf3): Delete expanders.
24710 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24712 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
24714 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24716 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
24717 * config/rs6000/rs6000.c: Ditto.
24719 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24721 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
24722 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
24723 comparison_operator.
24725 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24727 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
24728 * config/rs6000/rs6000.opt: Ditto.
24729 * config/rs6000/t-rtems: Ditto.
24731 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24733 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
24734 TARGET_E500_SINGLE by 0, simplify.
24735 * config/rs6000/rs6000.c: Ditto.
24736 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
24737 (spe_build_register_parallel): Delete.
24738 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
24739 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
24740 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
24741 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
24742 (E500_CONVERT): Delete.
24743 * config/rs6000/spe.md: Remove many patterns and all define_constants.
24745 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24747 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
24748 * config/rs6000/dfp.md: Ditto.
24749 (negdd2, *negdd2_fpr): Merge.
24750 (absdd2, *absdd2_fpr): Merge.
24751 (negtd2, *negtd2_fpr): Merge.
24752 (abstd2, *abstd2_fpr): Merge.
24753 * config/rs6000/e500.h: Delete file.
24754 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
24755 TARGET_FPRS by 1 and simplify.
24756 * config/rs6000/rs6000-c.c: Ditto.
24757 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
24758 TARGET_DF_SPE by 0.
24759 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
24761 * config/rs6000/rs6000.md: Ditto.
24762 (floatdidf2, *floatdidf2_fpr): Merge.
24763 (move_from_CR_gt_bit): Delete.
24764 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
24765 (E500_CR_IOR_COMPARE): Delete.
24766 (All patterns that require !TARGET_FPRS): Delete.
24767 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
24769 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24771 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
24773 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24775 * graphds.c (add_edge): Intitialize edge's attached data.
24776 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
24777 pointer parameter. Call pointed function on each edge during
24778 graph traversing. Skip traversing the edge when the function
24780 (graphds_dfs, graphds_scc): Ditto.
24781 (for_each_edge): New parameter. Pass the new parameter to callback
24783 * graphds.h (skip_edge_callback): New function pointer type.
24784 (graphds_dfs, graphds_scc): New function pointer parameter.
24785 (graphds_edge_callback, for_each_edge): New parameter.
24787 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24789 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
24790 out code checking if runtime alias check is possible to below ...
24791 Call the new function.
24792 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
24793 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
24795 2017-06-07 Marek Polacek <polacek@redhat.com>
24798 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
24799 TYPE_OVERFLOW_WRAPS check.
24801 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24803 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
24804 if versioning is required.
24805 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
24806 peeling with the check for versioning.
24808 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24810 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
24811 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
24812 Set true to new parameter if new ssa variable is defined.
24813 (vect_gen_vector_loop_niters): Refactor. Set range information
24814 for the new vector loop bound variable.
24815 (vect_do_peeling): Ditto.
24817 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24819 * tree-affine.c (ssa.h): Include header file.
24820 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
24821 has wrapping overflow behavior.
24823 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24825 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
24827 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24829 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
24830 (tree_to_aff_combination): ... here.
24832 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24834 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
24835 reg_pressure model function.
24836 (ivopts_global_cost_for_size): Delete.
24837 (determine_set_costs, iv_ca_recount_cost): Call new model function
24838 ivopts_estimate_reg_pressure.
24840 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24842 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
24843 expensive than udiv. Remove floating point cases from mod.
24845 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24847 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
24848 Increase idiv cost.
24850 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24852 * config/aarch64/aarch64.md
24853 (copysignsf3): Fix mask generation.
24855 2017-06-07 Jakub Jelinek <jakub@redhat.com>
24857 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
24859 (class dump_manager): Add register_dumps method.
24860 * dumpfile.c: Include langhooks.h.
24861 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
24862 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
24863 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
24864 (dump_manager::dump_register): Start with 512 entries instead of 32.
24865 (dump_manager::register_dumps): New method.
24866 * toplev.c (general_init): Instead of invoking register_dumps
24867 langhook, invoke register_dumps method on the dump manager.
24868 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
24871 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
24873 * doc/md.texi: Clarify the restrictions on a define_insn condition.
24874 Say that # requires an associated define_split to exist, and that
24875 the define_split must be suitable for use after register allocation.
24877 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24879 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
24880 (compute_outgoing_frequencies): Also initialize zero counts.
24881 (find_many_sub_basic_blocks): Do not produce uninitialized profile
24882 around loops; preserve more of profile when nothing changes.
24884 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
24886 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
24888 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
24889 * config/arm/arm-cpu-cdata.h: Regenerate.
24890 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
24891 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
24892 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
24893 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
24894 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
24896 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
24898 * config/arm/t-rmprofile: Likewise.
24899 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
24901 2017-06-06 David S. Miller <davem@davemloft.net>
24904 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
24905 blockage if function uses alloca.
24907 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24909 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
24910 New "uid" fields to hold pretty-print IDs of group and ref.
24911 Memory references are now identified as <group_id>:<ref_id>
24912 instead of using [random] addresses.
24913 (dump_mem_details): Simplify, no functional change.
24914 (dump_mem_ref): Simplify and make output more concise.
24915 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
24916 (find_or_create_group): Initialize group uid.
24917 (record_ref): Initialize ref uid. Improve debug output.
24918 (prune_group_by_reuse, should_issue_prefetch_p,)
24919 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
24920 (mark_nontemporal_store, determine_loop_nest_reuse):
24921 Improve debug output.
24923 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24925 * dbgcnt.def (prefetch): New debug counter.
24926 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
24927 (schedule_prefetches): Stop issueing prefetches if debug counter
24930 2017-06-06 Tom de Vries <tom@codesourcery.com>
24932 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
24933 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
24935 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24937 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
24938 Use aarch64_reg_or_zero predicate for operand 4.
24939 (aarch64_compare_and_swap<mode> define_insn_and_split):
24940 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
24941 (aarch64_store_exclusive<mode>): Likewise for operand 2.
24943 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24945 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
24946 (arm_compute_save_core_reg_mask): This.
24947 (thumb1_compute_save_reg_mask): Rename into ...
24948 (thumb1_compute_save_core_reg_mask): This.
24949 (arm_compute_save_reg0_reg12_mask): Adapt comment.
24950 (arm_compute_frame_layout): Likewise.
24952 2017-06-06 Richard Biener <rguenther@suse.de>
24954 PR tree-optimization/80974
24955 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
24956 keep or clear leaders SSA info.
24958 2017-06-06 Tom de Vries <tom@codesourcery.com>
24960 * config/nvptx/nvptx.c (split_mode_p): New function.
24961 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
24963 2017-06-06 Tom de Vries <tom@codesourcery.com>
24965 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
24967 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24970 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
24973 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24975 * shrink-wrap.c (handle_simple_exit): Update profile.
24976 (try_shrink_wrapping): Upate profile.
24978 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24980 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
24981 (tree_guess_outgoing_edge_probabilities): New.
24982 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
24983 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
24985 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24987 * ipa-split.c (split_function): Initialize return bb profile.
24989 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24991 * profile.c (compute_branch_probabilities): Also initialize
24992 EXIT_BLOCK profile.
24994 2017-06-06 Richard Biener <rguenther@suse.de>
24996 PR tree-optimization/80928
24997 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
24998 (vect_analyze_loop_operations): Properly guard analysis for
25000 (vect_transform_loop): Likewise.
25001 (vect_analyze_loop_2): Also reset SLP type on PHIs.
25002 (vect_model_induction_cost): Do not cost for pure SLP.
25003 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
25004 of induction in inner loop vectorization.
25005 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
25006 (vect_get_and_check_slp_defs): Handle vect_induction_def.
25007 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
25009 (vect_analyze_slp_cost_1): Cost induction.
25010 (vect_detect_hybrid_slp_stmts): Handle PHIs.
25011 (vect_get_slp_vect_defs): Likewise.
25012 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
25013 (vect_transform_stmt): Handle SLP reductions.
25014 * tree-vectorizer.h (vectorizable_induction): Adjust.
25016 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25018 * config/rs6000/rs6000.c (make_resolver_func): Update
25019 init_lowered_empty_function call.
25021 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
25023 * doc/invoke.texi: Document the -fprofile-abs-path option.
25024 * common.opt (fprofile-abs-path): New option.
25025 * gcov-io.h (gcov_write_filename): Declare.
25026 * gcov-io.c (gcov_write_filename): New function.
25027 * coverage.c (coverage_begin_function): Use gcov_write_filename.
25028 * profile.c (output_location): Likewise.
25030 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25032 * shring-wrap.c: Revert accidental commit.
25034 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
25036 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
25038 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25040 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
25042 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
25043 profile in callgraph edge.
25044 * profile-count.h (apply_probability): If THIS is 0, then result is 0
25045 (apply_scale): Likewise.
25046 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
25047 Also scale profile when inlining function with zero profile.
25048 (initialize_cfun): Update exit block profile even when it is zero.
25049 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
25050 when profile is read.
25052 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25054 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25055 (CLONE_*): New constants to define the processors we can generate
25056 code for with the target_clone attribute.
25057 (rs6000_clone_map): New array to identify which clone processors
25058 the current program is running on.
25059 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25060 target_clone attribute.
25061 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25062 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25063 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25064 (cpu_expand_builtin): Add support for target_clone attribute.
25065 (rs6000_valid_attribute_p): Allow "default" attribute.
25066 (get_decl_name): New debug function to simplify printing the
25067 current function name in debugging statements.
25068 (rs6000_clone_priority): New functions to support the target_clone
25069 attribute, and be able to generate code to switch between ISA 2.05
25070 through ISA 3.0 (power6 through power9).
25071 (rs6000_compare_version_priority): Likewise.
25072 (rs6000_get_function_versions_dispatcher): Likewise.
25073 (make_resolver_func): Likewise.
25074 (add_condition_to_bb): Likewise.
25075 (dispatch_function_versions): Likewise.
25076 (rs6000_generate_version_dispatcher_body): Likewise.
25077 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25078 (fusion_gpr_load_p): Fix a spacing issue.
25079 * doc/extend.texi (Common Function Attributes): Document that the
25080 PowerPC supports the target_clone attribute.
25082 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25084 * config/arm/arm.h: explain F symbol found in description of ARM
25085 register allocation in its legend.
25087 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25089 * config/mips/frame-header-opt.c: Include profile-count.h.
25090 * config/riscv/riscv.c: Include profile-count.h
25092 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25094 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25096 (sm_set_flag_if_changed): Add bbs field.
25097 (execute_sm_if_changed_flag_set): Pass BBS.
25098 (execute_sm): Update.
25100 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25102 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25105 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25107 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25108 (peephole2): New peephole2 to emit the above.
25109 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25111 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25113 * config/aarch64/aarch64.c (define_peephole2 above
25114 *sub_<shift>_<mode>): New peephole.
25116 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25118 * config/i386/i386.c (make_resolver_func): Update.
25119 * Makefile.in: Add profile-count.h and profile-count.o
25120 * auto-profile.c (afdo_indirect_call): Update to new API.
25121 (afdo_set_bb_count): Update.
25122 (afdo_propagate_edge): Update.
25123 (afdo_propagate_circuit): Update.
25124 (afdo_calculate_branch_prob): Update.
25125 (afdo_annotate_cfg): Update.
25126 * basic-block.h: Include profile-count.h
25127 (struct edge_def): Turn count to profile_count.
25128 (struct basic_block_def): Likewie.
25129 (REG_BR_PROB_BASE): Move to profile-count.h
25130 (RDIV): Move to profile-count.h
25131 * bb-reorder.c (max_entry_count): Turn to profile_count.
25132 (find_traces): Update.
25133 (rotate_loop):Update.
25134 (connect_traces):Update.
25135 (sanitize_hot_paths):Update.
25136 * bt-load.c (migrate_btr_defs): Update.
25137 * cfg.c (RDIV): Remove.
25138 (init_flow): Use alloc_block.
25139 (alloc_block): Uninitialize count.
25140 (unchecked_make_edge): Uninitialize count.
25141 (check_bb_profile): Update.
25142 (dump_edge_info): Update.
25143 (dump_bb_info): Update.
25144 (update_bb_profile_for_threading): Update.
25145 (scale_bbs_frequencies_int): Update.
25146 (scale_bbs_frequencies_gcov_type): Update.
25147 (scale_bbs_frequencies_profile_count): New.
25148 * cfg.h (update_bb_profile_for_threading): Update.
25149 (scale_bbs_frequencies_profile_count): Declare.
25150 * cfgbuild.c (compute_outgoing_frequencies): Update.
25151 (find_many_sub_basic_blocks): Update.
25152 * cfgcleanup.c (try_forward_edges): Update.
25153 (try_crossjump_to_edge): Update.
25154 * cfgexpand.c (expand_gimple_tailcall): Update.
25155 (construct_exit_block): Update.
25156 * cfghooks.c (verify_flow_info): Update.
25157 (dump_bb_for_graph): Update.
25158 (split_edge): Update.
25159 (make_forwarder_block): Update.
25160 (duplicate_block): Update.
25161 (account_profile_record): Update.
25162 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
25163 (get_estimated_loop_iterations): Update.
25164 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
25165 (single_likely_exit): Update.
25166 * cfgloopmanip.c (scale_loop_profile): Update.
25168 (set_zero_probability): Update.
25169 (lv_adjust_loop_entry_edge): Update.
25170 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
25171 (purge_dead_edges): Update.
25172 (rtl_account_profile_record): Update.
25173 * cgraph.c (cgraph_node::create): Uninitialize count.
25174 (symbol_table::create_edge): Uninitialize count.
25175 (cgraph_update_edges_for_call_stmt_node): Update.
25176 (cgraph_edge::dump_edge_flags): Update.
25177 (cgraph_node::dump): Update.
25178 (cgraph_edge::maybe_hot_p): Update.
25179 * cgraph.h: Include profile-count.h
25180 (create_clone), create_edge, create_indirect_edge): Update.
25181 (cgraph_node): Turn count to profile_count.
25182 (cgraph_edge0: Likewise.
25183 (make_speculative, clone): Update.
25184 (create_edge): Update.
25185 (init_lowered_empty_function): Update.
25186 * cgraphclones.c (cgraph_edge::clone): Update.
25187 (duplicate_thunk_for_node): Update.
25188 (cgraph_node::create_clone): Update.
25189 * cgraphunit.c (cgraph_node::analyze): Update.
25190 (cgraph_node::expand_thunk): Update.
25191 * final.c (dump_basic_block_info): Update.
25192 * gimple-streamer-in.c (input_bb): Update.
25193 * gimple-streamer-out.c (output_bb): Update.
25194 * graphite.c (print_global_statistics): Update.
25195 (print_graphite_scop_statistics): Update.
25196 * hsa-brig.c: Include basic-block.h.
25197 * hsa-dump.c: Include basic-block.h.
25198 * hsa-gen.c (T sum_slice): Update.
25199 (convert_switch_statements):Update.
25200 * hsa-regalloc.c: Include basic-block.h.
25201 * ipa-chkp.c (chkp_produce_thunks): Update.
25202 * ipa-cp.c (struct caller_statistics): Update.
25203 (init_caller_stats): Update.
25204 (gather_caller_stats): Update.
25205 (ipcp_cloning_candidate_p): Update.
25206 (good_cloning_opportunity_p): Update.
25207 (get_info_about_necessary_edges): Update.
25208 (dump_profile_updates): Update.
25209 (update_profiling_info): Update.
25210 (update_specialized_profile): Update.
25211 (perhaps_add_new_callers): Update.
25212 (decide_about_value): Update.
25213 (ipa_cp_c_finalize): Update.
25214 * ipa-devirt.c (struct odr_type_warn_count): Update.
25215 (struct decl_warn_count): Update.
25216 (struct final_warning_record): Update.
25217 (possible_polymorphic_call_targets): Update.
25218 (ipa_devirt): Update.
25219 * ipa-fnsummary.c (redirect_to_unreachable): Update.
25220 * ipa-icf.c (sem_function::merge): Update.
25221 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
25222 * ipa-inline.c (compute_uninlined_call_time): Update.
25223 (compute_inlined_call_time): Update.
25224 (want_inline_small_function_p): Update.
25225 (want_inline_self_recursive_call_p): Update.
25226 (edge_badness): Update.
25227 (lookup_recursive_calls): Update.
25228 (recursive_inlining): Update.
25229 (inline_small_functions): Update.
25230 (dump_overall_stats): Update.
25231 (dump_inline_stats): Update.
25232 * ipa-profile.c (ipa_profile_generate_summary): Update.
25233 (ipa_propagate_frequency): Update.
25234 (ipa_profile): Update.
25235 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25236 * ipa-utils.c (ipa_merge_profiles): Update.
25237 * loop-doloop.c (doloop_modify): Update.
25238 * loop-unroll.c (report_unroll): Update.
25239 (unroll_loop_runtime_iterations): Update.
25240 * lto-cgraph.c (lto_output_edge): Update.
25241 (lto_output_node): Update.
25242 (input_node): Update.
25243 (input_edge): Update.
25244 (merge_profile_summaries): Update.
25245 * lto-streamer-in.c (input_cfg): Update.
25246 * lto-streamer-out.c (output_cfg): Update.
25247 * mcf.c (create_fixup_graph): Update.
25248 (adjust_cfg_counts): Update.
25249 (sum_edge_counts): Update.
25250 * modulo-sched.c (sms_schedule): Update.
25251 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
25252 * predict.c (maybe_hot_count_p): Update.
25253 (probably_never_executed): Update.
25254 (dump_prediction): Update.
25255 (combine_predictions_for_bb): Update.
25256 (propagate_freq): Update.
25257 (handle_missing_profiles): Update.
25258 (counts_to_freqs): Update.
25259 (rebuild_frequencies): Update.
25260 (force_edge_cold): Update.
25261 * predict.h: Include profile-count.h
25262 (maybe_hot_count_p, counts_to_freqs): UPdate.
25263 * print-rtl-function.c: Do not include cfg.h
25264 * print-rtl.c: Include basic-block.h
25265 * profile-count.c: New file.
25266 * profile-count.h: New file.
25267 * profile.c (is_edge_inconsistent): Update.
25268 (correct_negative_edge_counts): Update.
25269 (is_inconsistent): Update.
25270 (set_bb_counts): Update.
25271 (read_profile_edge_counts): Update.
25272 (compute_frequency_overlap): Update.
25273 (compute_branch_probabilities): Update; Initialize and deinitialize
25275 (branch_prob): Update.
25276 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
25277 (edge_gcov_count): New.
25278 (bb_gcov_count): New.
25279 * shrink-wrap.c (try_shrink_wrapping): Update.
25280 * tracer.c (better_p): Update.
25281 * trans-mem.c (expand_transaction): Update.
25282 (ipa_tm_insert_irr_call): Update.
25283 (ipa_tm_insert_gettmclone_call): Update.
25284 * tree-call-cdce.c: Update.
25285 * tree-cfg.c (gimple_duplicate_sese_region): Update.
25286 (gimple_duplicate_sese_tail): Update.
25287 (gimple_account_profile_record): Update.
25288 (execute_fixup_cfg): Update.
25289 * tree-inline.c (copy_bb): Update.
25290 (copy_edges_for_bb): Update.
25291 (initialize_cfun): Update.
25292 (freqs_to_counts): Update.
25293 (copy_cfg_body): Update.
25294 (expand_call_inline): Update.
25295 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25296 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
25297 (try_unroll_loop_completely): Update.
25298 (try_peel_loop): Update.
25299 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25300 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
25301 * tree-ssa-loop-split.c (connect_loops): Update.
25302 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
25303 * tree-ssa-reassoc.c (branch_fixup): Update.
25304 * tree-ssa-tail-merge.c (replace_block_by): Update.
25305 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
25306 (compute_path_counts): Update.
25307 (update_profile): Update.
25308 (recompute_probabilities): Update.
25309 (update_joiner_offpath_counts): Update.
25310 (estimated_freqs_path): Update.
25311 (freqs_to_counts_path): Update.
25312 (clear_counts_path): Update.
25313 (ssa_fix_duplicate_block_edges): Update.
25314 (duplicate_thread_path): Update.
25315 * tree-switch-conversion.c (case_bit_test_cmp): Update.
25316 (struct switch_conv_info): Update.
25317 * tree-tailcall.c (decrease_profile): Update.
25318 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25319 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25320 * value-prof.c (check_counter): Update.
25321 (gimple_divmod_fixed_value): Update.
25322 (gimple_mod_pow2): Update.
25323 (gimple_mod_subtract): Update.
25324 (gimple_ic_transform): Update.
25325 (gimple_stringop_fixed_value): Update.
25326 * value-prof.h (gimple_ic): Update.
25328 2017-06-02 Carl Love <cel@us.ibm.com>
25330 * config/rs6000/rs6000-c: Add support for built-in functions
25331 vector double vec_doublee (vector signed int);
25332 vector double vec_doublee (vector unsigned int);
25333 vector double vec_doublee (vector float);
25334 vector double vec_doubleh (vector signed int);
25335 vector double vec_doubleh (vector unsigned int);
25336 vector double vec_doubleh (vector float);
25337 vector double vec_doublel (vector signed int);
25338 vector double vec_doublel (vector unsigned int);
25339 vector double vec_doublel (vector float);
25340 vector double vec_doubleo (vector signed int);
25341 vector double vec_doubleo (vector unsigned int);
25342 vector double vec_doubleo (vector float);.
25343 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
25344 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
25346 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
25347 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
25348 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
25350 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
25351 vec_doublel, vec_doubleh.
25352 * doc/extend.texi: Update the built-in documentation file for the
25353 new built-in functions.
25355 2017-06-02 David Malcolm <dmalcolm@redhat.com>
25358 * ipa-inline-analysis.c (free_growth_caches): Set
25359 edge_removal_hook_holder to NULL after removing it.
25361 2017-06-02 Sudakshina Das <sudi.das@arm.com>
25363 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
25364 comparision with zero.
25366 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25367 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25368 for early expansion of vec_min and vec_max builtins.
25369 (builtin_function_type): Add min/max unsigned variants to those
25370 identified as having unsigned arguments.
25372 2017-06-02 Olivier Hainque <hainque@adacore.com>
25374 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
25376 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25378 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
25379 Use VALL_F16 iterator rather than VALL.
25381 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25383 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
25384 Emit CBNZ inside loop when doing a strong exchange and comparing
25385 against zero. Generate the CC flags after the loop.
25387 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
25389 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
25390 (dl_section_ref): New.
25391 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
25392 On AIX, append an expression to subtract the size of the
25393 section length to dl_section_ref.
25395 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25397 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25398 for early expansion of vector absolute builtins.
25400 2017-06-02 Richard Biener <rguenther@suse.de>
25402 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
25403 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
25405 2017-06-02 Richard Biener <rguenther@suse.de>
25407 PR tree-optimization/80948
25408 * tree-tailcall.c (find_tail_calls): Track stmts to move in
25409 stmt order as well.
25411 2017-06-02 Richard Biener <rguenther@suse.de>
25413 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
25415 * tree-vect-stmts.c (process_use): Do not mark backedge defs
25416 for inductions as relevant.
25418 2017-06-02 Richard Biener <rguenther@suse.de>
25420 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
25421 (vectorizable_induction): ... this. Remove dead code.
25423 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
25425 * builtins. (expand_builtin_alloca): Remove second parameter and
25426 infer its value from the first parameter instead.
25427 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
25429 2017-06-02 Jakub Jelinek <jakub@redhat.com>
25431 PR rtl-optimization/80903
25432 * loop-doloop.c (add_test): Unshare sequence.
25434 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
25436 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
25438 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
25440 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
25442 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
25443 xlogue_layout::get_instance, logue_layout::xlogue_layout,
25444 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
25445 (xlogue_layout::get_stub_rtx): Make static.
25446 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
25447 (xlogue_layout::compute_stub_managed_regs): Rename to...
25448 (xlogue_layout::count_stub_managed_regs): ...this.
25449 (xlogue_layout::is_stub_managed_reg): New function.
25450 (xlogue_layout::m_stub_names): Rename to...
25451 (xlogue_layout::s_stub_names): ...this, make static.
25452 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
25453 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
25454 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
25455 xlogue_layout::s_stub_names): Instantiate statics.
25456 (stub_managed_regs): Remove.
25457 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
25458 (disable_call_ms2sysv_xlogues): Rename to...
25459 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
25460 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
25462 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
25463 change after reload_completed.
25464 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
25466 (ix86_expand_prologue): Likewise.
25467 (ix86_expand_epilogue): Likewise.
25468 (ix86_expand_split_stack_prologue): Likewise.
25469 (ix86_compute_frame_layout): Remove frame parameter ...
25470 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
25471 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
25473 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
25474 (ix86_frame): Move from here ...
25475 * config/i386/i386.h (ix86_frame): ... to here.
25476 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
25477 complete ix86_frame data structure instead. Remove some_ld_name.
25479 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25481 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
25482 symbols that hold a DECL_VALUE_EXPR.
25484 2017-06-01 Martin Jambor <mjambor@suse.cz>
25486 PR tree-optimization/80898
25487 * tree-sra.c (process_subtree_disqualification): Removed.
25488 (disqualify_candidate): Do not acll
25489 process_subtree_disqualification.
25490 (subtree_mark_written_and_enqueue): New function.
25491 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
25492 RHS has been disqualified and re-queue LHS if necessary. Apart
25493 from that, ignore disqualified RHS.
25495 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25497 * config/s390/s390.c (s390_emit_epilogue): Disable early return
25498 address fetch for z10 or later.
25500 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25502 * config/arc/arc.md (tst_movb): Add guard when splitting.
25504 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25506 * config/arc/arc.c (arc_can_eliminate): Test against
25507 arc_frame_pointer_needed.
25509 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25511 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
25512 to prevent store reordering.
25513 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
25514 (type): Add block type.
25515 (stack_tie): Define special instruction to be used in
25518 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25520 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
25521 constraint. It is not valid for the pattern.
25522 (noncommutative_binary_comparison): Likewise.
25524 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25526 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
25529 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25531 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
25532 be used by the reg-alloc.
25534 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25536 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
25537 reg-alloc when having mul64 or mul32x16 instructions.
25538 (mulsidi3): Likewise.
25539 (umulsidi3): Likewise.
25540 (mulsi32x16): New pattern.
25541 (mulsi64): Likewise.
25542 (mulsidi64): Likewise.
25543 (umulsidi64): Likewise.
25544 (MUL32x16_REG): Define.
25545 (mul64_600): Use MUL32x16_REG.
25546 (mac64_600): Likewise.
25547 (umul64_600): Likewise.
25548 (umac64_600): Likewise.
25550 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25552 * config/arc/arc.md (mulsi3_700): Make it commutative.
25554 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
25556 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
25558 (*sign_extendsidi2_insn): Likewise for movstosw.
25560 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25562 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
25563 the type of the input discriminant value. Convert the
25564 discriminant value of signedness vary.
25566 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
25568 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
25569 Add to -Wall section.
25571 2017-06-01 Richard Biener <rguenther@suse.de>
25573 PR middle-end/66313
25574 * fold-const.c (fold_plusminus_mult_expr): If the factored
25575 factor may be zero use a wrapping type for the inner operation.
25576 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
25577 and handle moved defs.
25578 (process_assignment): Properly guard the unary op case. Return a
25579 tri-state indicating that moving the stmt before the call may allow
25580 to continue. Pass through to_move.
25581 (find_tail_calls): Handle moving unrelated defs before
25584 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
25587 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
25588 splitter result in the canonical way.
25590 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
25592 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
25593 also for 32bit target. Update insn attributes.
25594 (zero-extendsidi2 splitter): Allow all registers for operand 1.
25596 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
25598 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
25599 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
25600 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
25601 (_mm_maskz_min_ss): New intrinsics.
25603 2017-05-31 Martin Liska <mliska@suse.cz>
25605 * tree-vect-loop.c (vect_create_epilog_for_reduction):
25606 Change comment style to one we normally use.
25607 (vectorizable_reduction): Likewise.
25608 (vectorizable_induction): Likewise.
25609 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
25610 (vectorizable_call): Likewise.
25611 (vectorizable_simd_clone_call): Likewise.
25612 (vectorizable_conversion): Likewise.
25613 (vectorizable_assignment): Likewise.
25614 (vectorizable_shift): Likewise.
25615 (vectorizable_operation): Likewise.
25616 (vectorizable_store): Likewise.
25617 (vectorizable_load): Likewise.
25618 * tree-vectorizer.h: Likewise.
25620 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
25622 * passes.c (emergency_dump_function): New.
25623 * tree-pass.h (emergency_dump_function): Declare.
25624 * plugin.c (plugins_internal_error_function): Remove.
25625 * plugin.h (plugins_internal_error_function): Remove declaration.
25626 * toplev.c (internal_error_function): New static function. Use it...
25627 (general_init): ...here.
25629 2017-05-31 Graham Markall <graham.markall@embecosm.com>
25631 * config/arc/arc.c (arc_print_operand): Handle constant operands.
25632 (arc_rtx_costs): Add costs for new patterns.
25633 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
25634 * config/arc/predicates.md: Add _1_2_3_operand predicate.
25636 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
25638 * tree-ssa-strlen.c (get_next_strinfo): New function.
25639 (get_stridx_plus_constant): Use it.
25640 (zero_length_string): Likewise.
25641 (adjust_related_strinfos): Likewise.
25642 (adjust_last_stmt): Likewise.
25644 2017-05-31 Richard Biener <rguenther@suse.de>
25647 * config/i386/i386.c (ix86_expand_builtin): Remove assert
25648 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
25650 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
25652 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
25653 loop_vinfo argument and use of dependence distance vectors.
25654 Check instead whether the two references differ only in their
25655 initial value and assume that they have the same alignment if the
25656 difference is a multiple of the vector alignment.
25657 (vect_analyze_data_refs_alignment): Update call accordingly.
25659 2017-05-31 Martin Liska <mliska@suse.cz>
25662 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
25664 2017-05-31 Bin Cheng <bin.cheng@arm.com>
25666 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
25667 (create_intersect_range_checks): Move from ...
25668 * tree-data-ref.c (create_intersect_range_checks_index)
25669 (create_intersect_range_checks): ... to here.
25670 (create_runtime_alias_checks): New function factored from ...
25671 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
25672 here. Call above function.
25673 * tree-data-ref.h (create_runtime_alias_checks): New function.
25675 2017-05-31 Bin Cheng <bin.cheng@arm.com>
25677 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
25678 segment length for dr_b and compute it in wide_int.
25680 2017-05-31 Richard Biener <rguenther@suse.de>
25682 PR tree-optimization/80906
25683 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
25684 and pass through iv_map.
25685 (copy_bb_and_scalar_dependences): Adjust.
25686 (translate_pending_phi_nodes): Likewise.
25687 (copy_loop_close_phi_args): Handle code-generating IVs instead
25690 2017-05-30 David Malcolm <dmalcolm@redhat.com>
25692 * diagnostic-color.c (color_dict): Add "type-diff".
25693 (parse_gcc_colors): Update comment.
25694 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
25695 -fdiagnostics-show-template-tree and -fno-elide-type.
25696 (GCC_COLORS): Add type-diff to example.
25698 (-fdiagnostics-show-template-tree): New.
25699 (-fno-elide-type): New.
25700 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
25701 the pp_format_decoder callback. Call any m_format_postprocessor's
25703 (pretty_printer::pretty_printer): Initialize
25704 m_format_postprocessor.
25705 (pretty_printer::~pretty_printer): Delete any
25706 m_format_postprocessor.
25707 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
25708 (class format_postprocessor): New class.
25709 (struct pretty_printer::format_decoder): Document the new parameters.
25710 (struct pretty_printer::m_format_postprocessor): New field.
25711 * tree-diagnostic.c (default_tree_printer): Update for new
25712 bool and const char ** params.
25713 * tree-diagnostic.h (default_tree_printer): Likewise.
25715 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
25717 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
25718 (lwa_operand): Delete rs6000_gen_cell_microcode test.
25719 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25720 rs6000_gen_cell_microcode code.
25721 (rs6000_final_prescan_insn): Delete.
25722 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
25723 "warn-cell-microcode" entries.
25724 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
25725 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
25726 throughout. Change cc_reg_not_micro_cr0_operand to
25727 cc_reg_not_cr0_operand throughout.
25728 (*extendhi<mode>2_noload): Delete.
25729 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
25730 (mwarn-cell-microcode): Delete.
25731 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
25732 -mgen-cell-microcode and -mwarn-cell-microcode.
25734 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
25737 * config/i386/constraints.md (Yd): New constraint.
25739 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
25740 and (?Yd, r) alternatives. Update insn attributes.
25741 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
25742 and (?*Yd, r) alternatives. Update insn attributes.
25743 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
25745 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
25747 * gimplify.c (gimplify_modify_expr): Don't create a
25748 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
25751 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
25753 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
25755 2017-05-30 Richard Biener <rguenther@suse.de>
25757 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
25758 and reduc_def fields.
25759 (STMT_VINFO_REDUC_TYPE): New define.
25760 (STMT_VINFO_REDUC_DEF): Likewise.
25761 (vect_force_simple_reduction): Adjust prototype.
25762 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
25763 (vect_is_simple_reduction): Remove check_reduction argument.
25764 (vect_force_simple_reduction): Adjust and set
25765 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25766 (vectorizable_reduction): Do not re-do reduction analysis
25767 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25768 * tree-parloops.c (gather_scalar_reductions): Adjust.
25770 2017-05-30 Richard Biener <rguenther@suse.de>
25772 PR middle-end/80901
25773 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
25776 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
25778 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
25779 Introduce unknown_misalignment parameter and remove vf.
25780 (vect_peeling_hash_get_lowest_cost):
25781 Pass unknown_misalignment parameter.
25782 (vect_enhance_data_refs_alignment):
25783 Fix unsupportable data ref treatment.
25785 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25787 * tree-vect-data-refs.c (vect_get_data_access_cost):
25788 Workaround for SLP handling.
25789 (vect_enhance_data_refs_alignment):
25790 Compute costs for doing no peeling at all, compare to the best
25791 peeling costs so far and avoid peeling if cheaper.
25793 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25795 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
25796 Return peeling info and set costs to zero for unlimited cost
25798 (vect_enhance_data_refs_alignment): Also inspect all datarefs
25799 with unknown misalignment. Compute and costs for unknown
25800 misalignment, compare them to the costs for known misalignment
25801 and choose the cheapest for peeling.
25803 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25805 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
25806 (vect_get_peeling_costs_all_drs): Create function.
25807 (vect_peeling_hash_get_lowest_cost):
25808 Use vect_get_peeling_costs_all_drs.
25809 (vect_peeling_supportable): Create function.
25810 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
25812 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25814 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
25815 DR_HAS_NEGATIVE_STEP.
25816 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
25817 (vect_enhance_data_refs_alignment): Use.
25818 (vect_duplicate_ssa_name_ptr_info): Use.
25819 * tree-vectorizer.h (dr_misalignment): Use.
25820 (known_alignment_for_access_p): Use.
25822 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25825 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
25827 (has_section_name): New function.
25829 2017-05-30 Martin Liska <mliska@suse.cz>
25832 * auto-profile.c (get_function_decl_from_block): Fix
25835 2017-05-30 Richard Biener <rguenther@suse.de>
25837 PR middle-end/80876
25838 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
25840 2017-05-30 Martin Liska <mliska@suse.cz>
25842 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
25843 * dumpfile.h (struct dump_file_info): Remove ctors.
25845 2017-05-30 Martin Liska <mliska@suse.cz>
25847 * predict.def: Fix GNU coding style.
25849 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25851 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25852 Mark 'to' argument with ATTRIBUTE_UNUSED.
25854 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25856 * config/xtensa/xtensa.c (xtensa_emit_call): Use
25857 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
25858 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
25861 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
25863 * doc/install.texi (Options specification): Restore entry of
25864 --enable-sjlj-exceptions.
25866 2017-05-27 Michael Eager <eager@eagercon.com>
25869 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
25871 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
25873 * config/microblaze/microblaze.h
25874 (FIXED_REGISTERS): Update in macro.
25875 (CALL_USED_REGISTERS): Update in macro.
25877 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
25879 * doc/install.texi: Add links to macOS binary distributions.
25881 2017-05-27 Jakub Jelinek <jakub@redhat.com>
25885 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25887 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25889 2017-05-26 Martin Liska <mliska@suse.cz>
25891 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
25893 2017-05-26 Martin Liska <mliska@suse.cz>
25895 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
25896 always leading ';; '.
25897 (dump_bb_info): Likewise.
25898 (brief_dump_cfg): Likewise.
25899 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
25900 * dumpfile.c: Remove usage of TDF_VERBOSE.
25901 * dumpfile.h (enum dump_kind): Likewise.
25902 (dump_gimple_bb_header): Do not use TDF_COMMENT.
25903 * print-tree.c (debug_verbose): Remove.
25904 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
25905 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
25906 * tree-diagnostic.c (default_tree_printer): Replace
25907 TDF_DIAGNOSTIC with TDF_SLIM.
25909 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25911 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
25912 in parameter loop, rather than loop_vinfo.
25913 (create_intersect_range_checks): Ditto.
25914 (vect_create_cond_for_alias_checks): Update call to above functions.
25916 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25918 PR tree-optimization/80815
25919 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
25920 for merging runtime alias checks. Handle negative DR_STEPs.
25922 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25924 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
25926 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
25927 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
25928 out code pruning runtime alias checks.
25929 * tree-data-ref.c (prune_runtime_alias_test_list): New function
25930 factored out from above.
25931 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25933 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25935 (prune_runtime_alias_test_list): New decalaration.
25937 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25939 * tree-vect-data-refs.c (compare_tree): Rename and move ...
25940 * tree-data-ref.c (data_ref_compare_tree): ... to here.
25941 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
25942 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
25943 (operator==, comp_dr_with_seg_len_pair): Ditto.
25944 (vect_prune_runtime_alias_test_list): Ditto.
25946 2017-05-26 Martin Liska <mliska@suse.cz>
25949 * params.def: Bound partial-inlining-entry-probability param.
25951 2017-05-26 Marek Polacek <polacek@redhat.com>
25954 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
25957 2017-05-26 Richard Biener <rguenther@suse.de>
25959 PR tree-optimization/80842
25960 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
25963 2017-05-26 Richard Biener <rguenther@suse.de>
25965 PR tree-optimization/80844
25966 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
25968 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
25970 * doc/md.texi (Machine Constraints): Update x86 family
25971 machine constraints section to match 'config/i386/constraints.md'.
25973 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
25975 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
25977 2017-05-25 Nathan Sidwell <nathan@acm.org>
25979 * doc/invoke.texi (--enable-languages): Update documentation.
25981 2017-05-25 Martin Liska <mliska@suse.cz>
25983 * dumpfile.c: Add TDF_FOLDING.
25984 * dumpfile.h (enum dump_kind): Likewise.
25985 * genmatch.c (dt_simplify::gen_1): Use it.
25987 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25989 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
25991 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25993 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25994 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
25996 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25998 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
25999 * match.pd (X == C): Rewrite it here.
26000 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
26001 with_certain_nonzero_bits2): New predicates.
26002 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
26004 2017-05-24 Nathan Sidwell <nathan@acm.org>
26006 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
26009 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
26012 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26014 * config/powerpcspe: New port. Files are copied from the rs6000
26015 port, with "rs6000" in filenames replaced by "powerpcspe".
26017 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
26019 PR rtl-optimization/80754
26020 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
26022 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
26024 * config/sparc/sparc.md (length): Return the correct value for -mflat
26025 sibcalls to match output_sibcall.
26027 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26031 * config/rs6000/rs6000.c (struct machine_function): Add new field
26033 (rs6000_get_separate_components): Init that field, use it.
26034 (rs6000_components_for_bb): Use the field.
26036 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
26038 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
26040 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
26042 PR middle-end/80823
26043 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
26045 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26048 * config/s390/s390.c (s390_check_qrst_address): Check incoming
26049 address against address_operand predicate.
26050 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
26052 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26054 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26055 parameters passed indirectly.
26057 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26059 * config/i386/i386.md (*movdi_internal): Remove SSE4
26060 alternative 18 (?r, *v). Update insn attributes.
26061 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26062 Update insn attributes.
26063 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26064 Update insn attributes.
26065 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26066 alternative 1 (r, v). Remove isa attribute.
26067 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26068 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26069 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26071 2017-05-23 Tom de Vries <tom@codesourcery.com>
26073 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26076 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26078 * cgraphunit.c (symbol_table::process_new_functions): Update.
26079 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26080 (inline_generate_summary): Rename to ...
26081 (ipa_fn_summary_generate): ... this one.
26082 (inline_read_summary): Rename to ...
26083 (ipa_fn_summary_read): ... this one.
26084 (inline_write_summary): Rename to ...
26085 (ipa_fn_summary_write): ... this one.
26086 (inline_free_summary): Rename to ...
26087 (ipa_free_fn_summary): ... this one.
26088 (pass_data_local_fn_summary, pass_local_fn_summary,
26089 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26090 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26091 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26092 make_pass_ipa_fn_summary): New.
26093 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26094 inline_write_summary, inline_free_summary): Remove.
26095 (ipa_free_fn_summary) : New.
26096 * ipa-inline.c (ipa_inline): Update.
26097 (pass_ipa_inline): Do not generate summaries.
26098 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26100 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26101 and add pass_ipa_fn_summary.
26102 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26104 (make_pass_inline_parameters): Remove.
26106 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26108 * omp-low.c (struct omp_context): Remove "default_kind" member.
26111 * omp-offload.c (execute_oacc_device_lower): Remove the
26112 parallelism dimensions function attributes for unparallelized
26113 OpenACC kernels constructs.
26115 2017-05-23 Martin Liska <mliska@suse.cz>
26117 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
26119 (cgraph_edge::make_speculative): Likewise.
26120 (cgraph_edge::resolve_speculation): Likewise.
26121 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26122 (cgraph_node::dump): Likewise.
26123 * cgraph.h: Likewise.
26124 * cgraphunit.c (analyze_functions): Likewise.
26125 (symbol_table::compile): Likewise.
26126 * ipa-cp.c (print_all_lattices): Likewise.
26127 (determine_versionability): Likewise.
26128 (initialize_node_lattices): Likewise.
26129 (ipcp_verify_propagated_values): Likewise.
26130 (estimate_local_effects): Likewise.
26131 (update_profiling_info): Likewise.
26132 (create_specialized_node): Likewise.
26133 (perhaps_add_new_callers): Likewise.
26134 (decide_about_value): Likewise.
26135 (decide_whether_version_node): Likewise.
26136 (identify_dead_nodes): Likewise.
26137 (ipcp_store_bits_results): Likewise.
26138 * ipa-devirt.c (dump_targets): Likewise.
26139 (ipa_devirt): Likewise.
26140 * ipa-icf.c (sem_item::dump): Likewise.
26141 (sem_function::equals): Likewise.
26142 (sem_variable::equals): Likewise.
26143 (sem_item_optimizer::read_section): Likewise.
26144 (sem_item_optimizer::execute): Likewise.
26145 (congruence_class::dump): Likewise.
26146 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
26147 (dump_inline_summary): Likewise.
26148 (estimate_node_size_and_time): Likewise.
26149 (inline_analyze_function): Likewise.
26150 * ipa-inline-transform.c (inline_call): Likewise.
26151 * ipa-inline.c (report_inline_failed_reason): Likewise.
26152 (want_early_inline_function_p): Likewise.
26153 (edge_badness): Likewise.
26154 (update_edge_key): Likewise.
26155 (inline_small_functions): Likewise.
26156 * ipa-profile.c (ipa_profile): Likewise.
26157 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
26158 (ipa_make_edge_direct_to_target): Likewise.
26159 (remove_described_reference): Likewise.
26160 (ipa_impossible_devirt_target): Likewise.
26161 (propagate_controlled_uses): Likewise.
26162 (ipa_print_node_params): Likewise.
26163 (ipcp_transform_function): Likewise.
26164 * ipa-pure-const.c (pure_const_read_summary): Likewise.
26165 (propagate_pure_const): Likewise.
26166 * ipa-reference.c (generate_summary): Likewise.
26167 (read_write_all_from_decl): Likewise.
26168 (propagate): Likewise.
26169 (ipa_reference_read_optimization_summary): Likewise.
26170 * ipa-utils.c (ipa_merge_profiles): Likewise.
26171 * ipa.c (walk_polymorphic_call_targets): Likewise.
26172 (symbol_table::remove_unreachable_nodes): Likewise.
26173 (ipa_single_use): Likewise.
26174 * passes.c (execute_todo): Likewise.
26175 * predict.c (drop_profile): Likewise.
26176 * symtab.c (symtab_node::get_dump_name): New function.
26177 (symtab_node::dump_name): Likewise.
26178 (symtab_node::dump_asm_name): Likewise.
26179 (symtab_node::dump_references): Likewise.
26180 (symtab_node::dump_referring): Likewise.
26181 (symtab_node::dump_base): Likewise.
26182 (symtab_node::debug_symtab): Likewise.
26183 * tree-sra.c (convert_callers_for_node): Likewise.
26184 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
26185 * value-prof.c (init_node_map): Likewise.
26187 2017-05-23 Martin Liska <mliska@suse.cz>
26189 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
26190 and symtab_node::debug_symtab to symbol_table::debug.
26191 * cgraphunit.c (analyze_functions): Use the renamed function.
26192 (symbol_table::compile): Likewise.
26193 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
26194 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
26195 * passes.c (execute_todo): Likewise.
26196 * symtab.c (symbol_table::dump): New function.
26197 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
26199 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26201 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
26202 that nonconst implies exec.
26204 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26206 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26207 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26208 (inline_edge_summary_vec): Turn into ...
26209 (ipa_call_summaries): ... this one.
26210 (redirect_to_unreachable, edge_set_predicate,
26211 evaluate_properties_for_edge, inline_summary_alloc,
26212 reset_ipa_call_summary, reset_inline_summary,
26213 inline_summary_t::duplicate): Update.
26214 (inline_edge_duplication_hook): Turn to ...
26215 (ipa_call_summary_t::duplicate): ... this one.
26216 (inline_edge_removal_hook): Turn to ...
26217 (ipa_call_summary_t::remove): ... this one.
26218 (dump_inline_edge_summary): Turn to ...
26219 (dump_ipa_call_summary): ... this one.
26220 (estimate_function_body_sizes): Update.
26221 (inline_update_callee_summaries): Update.
26222 (remap_edge_change_prob): Update.
26223 (remap_edge_summaries): Update.
26224 (inline_merge_summary): Update.
26225 (do_estimate_edge_time): Update.
26226 (inline_generate_summary): Update.
26227 (inline_read_section): Update.
26228 (inline_read_summary): Update.
26229 (inline_free_summary): Update.
26230 * ipa-inline.c (can_inline_edge_p): Update.
26231 (compute_inlined_call_time): Update.
26232 (want_inline_small_function_p): Update.
26233 (edge_badness): Update.
26234 (early_inliner): Update.
26235 * ipa-inline.h (inline_edge_summary): Turn to ...
26236 (ipa_call_summary): ... this one.
26237 (ipa_call_summary_t): New class.
26238 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26239 (ipa_call_summaries): New.
26240 (inline_edge_summary): Remove.
26241 (estimate_edge_growth): Update.
26242 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26243 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26244 * ipa-split.c (execute_split_functions): Update.
26245 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26247 2017-05-23 Tom de Vries <tom@codesourcery.com>
26249 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26250 attributes): Document rdrand effective target.
26252 2017-05-23 Tom de Vries <tom@codesourcery.com>
26254 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26255 attributes): Sort alphabetically.
26257 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
26259 * config/avr/genmultilib.awk: Use gsub instead of gensub.
26261 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
26264 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
26265 V2DF/V2DI splat into two separate patterns, one that handles
26266 registers, and the other that only handles memory. Drop support
26267 for splatting from a GPR on ISA 2.07 and then splitting the
26268 splat into direct move and splat.
26269 (vsx_splat_<mode>_reg): Likewise.
26270 (vsx_splat_<mode>_mem): Likewise.
26272 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
26274 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
26276 2017-05-22 Jakub Jelinek <jakub@redhat.com>
26278 PR middle-end/80809
26279 * omp-low.c (finish_taskreg_remap): New function.
26280 (finish_taskreg_scan): If unit size of ctx->record_type
26281 is non-constant, unshare the size expression and replace
26282 decls in it with possible outer var refs.
26284 PR middle-end/80809
26285 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
26286 GOVD_SHARED rather than GOVD_PRIVATE with it.
26287 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
26288 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
26290 PR middle-end/80853
26291 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
26292 as last argument to build_outer_var_ref for pointer bases of array
26293 section reductions.
26295 2017-05-19 Martin Sebor <msebor@redhat.com>
26297 * print-tree.c (print_node): Print DECL_READ_P flag.
26299 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26301 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
26302 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
26303 * cgraph.c: Likewise.
26304 * cgraphunit.c: Likewise.
26305 * gengtype.c: Likewise.
26306 * ipa-cp.c: Likewise.
26307 * ipa-devirt.c: Likewise.
26308 * ipa-icf.c: Likewise.
26309 * ipa-predicate.c: Likewise.
26310 * ipa-profile.c: Likewise.
26311 * ipa-prop.c: Likewise.
26312 * ipa-split.c: Likewise.
26314 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
26315 edge_predicate_pool, dump_inline_hints,
26316 inline_summary::account_size_time, redirect_to_unreachable,
26317 edge_set_predicate, set_hint_predicate,
26318 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
26319 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
26320 inline_summary_t::remove, remap_hint_predicate_after_duplication,
26321 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
26322 ipa_call_summary_t::remove, initialize_growth_caches,
26323 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
26324 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
26325 mark_modified, unmodified_parm_1, unmodified_parm,
26326 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
26327 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
26328 compute_bb_predicates, will_be_nonconstant_expr_predicate,
26329 will_be_nonconstant_predicate, record_modified_bb_info,
26330 get_minimal_bb, record_modified, param_change_prob,
26331 phi_result_unknown_predicate, predicate_for_phi_result,
26332 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
26333 estimate_function_body_sizes, compute_inline_parameters,
26334 compute_inline_parameters_for_curren, pass_data_inline_parameters,
26335 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
26336 inline_update_callee_summaries, remap_edge_change_prob,
26337 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
26338 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
26339 inline_analyze_function, inline_summary_t::insert,
26340 inline_generate_summary, read_ipa_call_summary, inline_read_section,
26341 inline_read_summary, write_ipa_call_summary, inline_write_summary,
26342 inline_free_summary): Move to ipa-fnsummary.h
26343 (predicate_t): Remove.
26344 * ipa-fnsummary.c: New file.
26345 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
26346 (enum inline_hints_vals, inline_hints, agg_position_info,
26347 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
26348 inline_summaries, ipa_call_summary, ipa_call_summary_t,
26349 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
26350 dump_inline_summary, dump_inline_hints, inline_generate_summary,
26351 inline_read_summary, inline_write_summary, inline_free_summary,
26352 inline_analyze_function, initialize_inline_failed,
26353 inline_merge_summary, inline_update_overall_summary,
26354 compute_inline_parameters): Move to ipa-fnsummary.h
26355 * ipa-fnsummary.h: New file.
26356 * ipa-inline-transform.h: Include ipa-inline.h.
26357 * ipa-inline.c: LIkewise.
26359 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26361 * ipa-inline.c (edge_badness): Use inlined_time instead of
26362 inline_summaries->get.
26364 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26366 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
26368 2017-05-22 Nathan Sidwell <nathan@acm.org>
26370 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
26371 (fdump-lang): Document 'raw' option.
26372 * dumpfile.h (TDI_tu): Delete.
26373 * dumpfile.c (dump_files): Remove translation-unit.
26374 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
26376 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
26378 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
26379 command option from $(AWK) call.
26380 * config/avr/genmultilib.awk: Simplify and rewrite so that it
26381 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
26382 [FORMAT]: Remove handling of variable.
26383 * config/avr/t-multilib: Regenerate.
26385 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26387 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
26389 (dump_inline_summary): Do not print self_time.
26390 (estimate_function_body_sizes): Do not set self_time.
26391 (compute_inline_parameters): Likewise.
26392 (inline_read_section, inline_write_summary): Do not stream self_time.
26393 * ipa-inline.h (inline_summary): Drop self_time.
26395 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26397 * ipa-inline-analysis.c (account_size_time): Rename to ...
26398 (inline_summary::account_size_time): ... this one.
26399 (reset_ipa_call_summary): Turn to ...
26400 (ipa_call_summary::reset): ... this one.
26401 (reset_inline_summary): Turn to ...
26402 (inline_summary::reset): ... this one.
26403 (inline_summary_t::remove): Update.
26404 (inline_summary_t::duplicate): Update.
26405 (ipa_call_summary_t::remove): Update.
26406 (dump_inline_summary): Update.
26407 (estimate_function_body_sizes): Update.
26408 (compute_inline_parameters): Update.
26409 (estimate_node_size_and_time): Update.
26410 (inline_merge_summary): Update.
26411 (inline_update_overall_summary): Update.
26412 (inline_read_section): Update.
26413 (inline_write_summary): Update.
26414 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
26415 add account_size_time and reset member functions.
26416 (ipa_call_summary): Add reset function.
26417 * ipa-predicate.h (predicate::operator &): Constify.
26419 2017-05-22 Richard Biener <rguenther@suse.de>
26421 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
26423 2017-05-19 Jason Merrill <jason@redhat.com>
26425 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
26427 2017-05-19 Marek Polacek <polacek@redhat.com>
26430 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
26431 TYPE_OVERFLOW_WRAPS checks.
26433 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
26435 * tree-core.h (enum omp_clause_default_kind): Add
26436 "OMP_CLAUSE_DEFAULT_PRESENT".
26437 * tree-pretty-print.c (dump_omp_clause): Handle it.
26438 * gimplify.c (enum gimplify_omp_var_data): Add
26439 "GOVD_MAP_FORCE_PRESENT".
26440 (gimplify_adjust_omp_clauses_1): Map it to
26441 "GOMP_MAP_FORCE_PRESENT".
26442 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
26444 * gimplify.c (oacc_default_clause): Clarify.
26446 2017-05-19 Nathan Sidwell <nathan@acm.org>
26448 LANG_HOOK_REGISTER_DUMPS
26449 * toplev.c (general_init): Call register dump lang hook.
26450 * doc/invoke.texi: Document -fdump-lang option family.
26451 * dumpfile.c (dump_files): Remove class dump here.
26452 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
26453 * dumpfile.h (tree_dump_index): Remove TDI_class.
26454 * langhooks-def.h (lhd_register_dumps): Declare.
26455 (LANG_HOOKS_REGISTER_DUMPS): Define.
26456 (LANG_HOOKS_INITIALIZER): Add it.
26457 * langhooks.c (lhd_register_dumps): Define.
26458 * langhooks.h (struct lang_hooks): Add register_dumps.
26460 2017-05-19 Nathan Sidwell <nathan@acm.org>
26462 * context.h (context::set_passes): New.
26463 * context.c (context::context): Do not create pass manager.
26464 * toplev.c (general_init): Create pass manager here.
26466 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
26468 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
26469 use this splitter if two add or or instructions would also work for
26470 the constant we want to generate.
26472 2017-05-19 Richard Biener <rguenther@suse.de>
26475 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
26476 predicate evaluation.
26478 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26480 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
26482 * ipa-inline.c (want_inline_small_function_p): Do not cast to
26485 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26487 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26488 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26489 (inline_edge_summary_vec): Turn into ...
26490 (ipa_call_summaries): ... this one.
26491 (redirect_to_unreachable, edge_set_predicate,
26492 evaluate_properties_for_edge, inline_summary_alloc,
26493 reset_ipa_call_summary, reset_inline_summary,
26494 inline_summary_t::duplicate): Update.
26495 (inline_edge_duplication_hook): Turn to ...
26496 (ipa_call_summary_t::duplicate): ... this one.
26497 (inline_edge_removal_hook): Turn to ...
26498 (ipa_call_summary_t::remove): ... this one.
26499 (dump_inline_edge_summary): Turn to ...
26500 (dump_ipa_call_summary): ... this one.
26501 (estimate_function_body_sizes): Update.
26502 (inline_update_callee_summaries): Update.
26503 (remap_edge_change_prob): Update.
26504 (remap_edge_summaries): Update.
26505 (inline_merge_summary): Update.
26506 (do_estimate_edge_time): Update.
26507 (inline_generate_summary): Update.
26508 (inline_read_section): Update.
26509 (inline_read_summary): Update.
26510 (inline_free_summary): Update.
26511 * ipa-inline.c (can_inline_edge_p): Update.
26512 (compute_inlined_call_time): Update.
26513 (want_inline_small_function_p): Update.
26514 (edge_badness): Update.
26515 (early_inliner): Update.
26516 * ipa-inline.h (inline_edge_summary): Turn to ...
26517 (ipa_call_summary): ... this one.
26518 (ipa_call_summary_t): New class.
26519 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26520 (ipa_call_summaries): New.
26521 (inline_edge_summary): Remove.
26522 (estimate_edge_growth): Update.
26523 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26524 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26525 * ipa-split.c (execute_split_functions): Update.
26526 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26528 2017-05-19 Richard Biener <rguenther@suse.de>
26530 PR middle-end/80764
26531 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
26533 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
26535 * config/rs6000/rs6000.c (struct machine_function): Add field
26536 fpr_is_wrapped_separately.
26537 (rs6000_get_separate_components): Use 64 components. Handle the
26538 new FPR components.
26539 (rs6000_components_for_bb): Handle the FPR components.
26540 (rs6000_emit_prologue_components): Handle the FPR components.
26541 (rs6000_emit_epilogue_components): Handle the FPR components.
26542 (rs6000_set_handled_components): Handle the FPR components.
26543 (rs6000_emit_prologue): Don't output prologue code for those FPRs
26544 that are already separately shrink-wrapped.
26545 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
26546 that are already separately shrink-wrapped.
26548 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26551 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
26554 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
26555 (define_peephole2 for Altivec d-form load): Add peepholes to catch
26556 cases where the register allocator uses a move and an offsettable
26557 memory operation to/from a FPR register on ISA 2.06/2.07.
26558 (define_peephole2 for Altivec d-form store): Likewise.
26560 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
26563 * config/i386/mmx.md (*mov<mode>_internal): Enable
26564 alternatives 11, 12, 13 and 14 also for 32bit targets.
26565 Remove alternatives 15, 16, 17 and 18.
26566 * config/i386/sse.md (vec_concatv2di): Change
26567 alternative (!x, *y) to (x, ?!*Yn).
26569 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
26571 * dumpfile.h (enum dump_kind): Remove stray comma.
26573 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26575 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
26576 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
26577 predicate::num_conditions
26578 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
26579 (CHANGED): turn into predicate::changed.
26580 (agg_position_info): Move to ipa-predicate.h
26581 (add_condition, predicate::add_clause, predicate::operator &=,
26582 predicate::or_with, predicate::evaluate, predicate::probability,
26583 dump_condition, dump_clause, predicate::dump,
26584 predicate::remap_after_duplication, predicate::remap_after_inlining,
26585 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
26586 (evaluate_conditions_for_known_args): Update.
26587 (set_cond_stmt_execution_predicate): Update.
26588 * ipa-inline.h: Include ipa-predicate.h
26589 (condition, inline_param_summary, conditions, agg_position_info,
26590 predicate): Move to ipa-predicate.h
26591 * ipa-predicate.c: New file.
26592 * ipa-predicate.h: New file.
26594 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
26596 * final.c (leaf_function_p): Check we are not in a sequence.
26598 2017-05-18 Martin Liska <mliska@suse.cz>
26600 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
26601 * dumpfile.c (dump_register): Use new enum dump_kind.
26602 (get_dump_file_name): Likewise.
26603 (dump_enable_all): Likewise.
26604 (dump_switch_p_1): Likewise.
26605 (enable_rtl_dump_file): Remove usage of TDF_RTL.
26606 * dumpfile.h (enum dump_kind): New enum type.
26607 (struct dump_file_info): Create constructor and
26608 format fields and comments.
26609 * passes.c (pass_manager::register_one_dump_file):
26611 * statistics.c (statistics_early_init): Likewise.
26612 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
26613 TDF_TREE with TDF_SLIM.
26614 (gather_memory_references_ref): Likewise.
26616 2017-05-18 Martin Liska <mliska@suse.cz>
26618 * vec.h (struct vnull): Use it.
26620 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26622 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
26623 (true_predicate, false_predicate, true_predicate_p,
26624 false_predicate_p): Remove.
26625 (single_cond_predicate, not_inlined_predicate): Turn to member function
26627 (add_condition): Update.
26628 (add_clause): Turn to...
26629 (predicate::add_clause): ... this one; update; allow passing NULL
26631 (and_predicates): Turn to ...
26632 (predicate::operator &=): ... this one.
26633 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
26634 (or_predicates): Turn to ...
26635 (predicate::or_with): ... this one.
26636 (evaluate_predicate): Turn to ...
26637 (predicate::evaluate): ... this one.
26638 (predicate_probability): Turn to ...
26639 (predicate::probability): ... this one.
26640 (dump_condition): Update.
26641 (dump_predicate): Turn to ...
26642 (predicate::dump): ... this one.
26643 (account_size_time): Update.
26644 (edge_set_predicate): Update.
26645 (set_hint_predicate): UPdate.
26646 (evaluate_conditions_for_known_args): Update.
26647 (evaluate_properties_for_edge): Update.
26648 (remap_predicate_after_duplication): Turn to...
26649 (predicate::remap_after_duplication): ... this one.
26650 (remap_hint_predicate_after_duplication): Update.
26651 (inline_summary_t::duplicate): UPdate.
26652 (dump_inline_edge_summary): Update.
26653 (dump_inline_summary): Update.
26654 (set_cond_stmt_execution_predicate): Update.
26655 (set_switch_stmt_execution_predicate): Update.
26656 (compute_bb_predicates): Update.
26657 (will_be_nonconstant_expr_predicate): Update.
26658 (will_be_nonconstant_predicate): Update.
26659 (phi_result_unknown_predicate): Update.
26660 (predicate_for_phi_result): Update.
26661 (array_index_predicate): Update.
26662 (estimate_function_body_sizes): Update.
26663 (estimate_node_size_and_time): Update.
26664 (estimate_ipcp_clone_size_and_time): Update.
26665 (remap_predicate): Rename to ...
26666 (predicate::remap_after_inlining): ... this one.
26667 (remap_hint_predicate): Update.
26668 (inline_merge_summary): Update.
26669 (inline_update_overall_summary): Update.
26670 (estimate_size_after_inlining): Update.
26671 (read_predicate): Rename to ...
26672 (predicate::stream_in): ... this one.
26673 (read_inline_edge_summary): Update.
26674 (write_predicate): Rename to ...
26675 (predicate::stream_out): ... this one.
26676 (write_inline_edge_summary): Update.
26677 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
26678 (clause_t): Turn to uint32_t
26679 (predicate): Turn to class; implement constructor and operators
26681 (size_time_entry): Update.
26682 (inline_summary): Update.
26683 (inline_edge_summary): Update.
26685 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
26687 * fold-const.c (fold_binary_loc): Move transformation...
26688 * match.pd (C - X CMP X): ... here.
26690 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
26692 * config/sparc/sparc.c (sparc_option_override): Set function
26693 alignment for -mcpu=niagara7 to 64 to match the I$ line.
26694 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
26696 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
26698 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
26700 2017-05-18 Marek Polacek <polacek@redhat.com>
26703 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
26704 (pass_ubsan::execute): Call gimple_assign_single_p instead of
26705 gimple_assign_load_p.
26707 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
26709 PR middle-end/80692
26710 * real.c (do_compare): Give decimal_do_compare preference over
26711 comparing just the signs.
26713 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
26715 * doc/md.texi (Canonicalization of Instructions): Describe the
26716 canonical form of instructions that inherently set a condition
26719 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
26721 PR middle-end/80775
26722 * tree-cfg.c: Move deletion of unreachable case statements to after
26723 the merging of consecutive case labels.
26725 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
26727 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
26728 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
26729 restoring of callee-saved registers.
26731 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
26733 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
26734 * config/visium/visium.c (single_set_and_flags): Likewise.
26735 * config/visium/visium.md (Substitutions): Likewise.
26737 2017-05-17 Martin Liska <mliska@suse.cz>
26739 * cfg.c: Introduce dump_flags_t type and
26740 use it instead of int type.
26742 * cfghooks.c: Likewise.
26743 * cfghooks.h (struct cfg_hooks): Likewise.
26744 * cfgrtl.c: Likewise.
26745 * cfgrtl.h: Likewise.
26746 * cgraph.c (cgraph_node::get_body): Likewise.
26747 * coretypes.h: Likewise.
26748 * domwalk.c: Likewise.
26749 * domwalk.h: Likewise.
26750 * dumpfile.c (struct dump_option_value_info): Likewise.
26751 (dump_enable_all): Likewise.
26752 (dump_switch_p_1): Likewise.
26753 (opt_info_switch_p): Likewise.
26754 * dumpfile.h (enum tree_dump_index): Likewise.
26755 (struct dump_file_info): Likewise.
26756 * genemit.c: Likewise.
26757 * generic-match-head.c: Likewise.
26758 * gengtype.c (open_base_files): Likewise.
26759 * gimple-pretty-print.c: Likewise.
26760 * gimple-pretty-print.h: Likewise.
26761 * graph.c (print_graph_cfg): Likewise.
26762 * graphite-scop-detection.c (dot_all_sese): Likewise.
26763 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
26764 * loop-unroll.c (report_unroll): Likewise.
26765 * passes.c (pass_manager::register_one_dump_file): Likewise.
26766 * print-tree.c: Likewise.
26767 * statistics.c: Likewise.
26768 * tree-cfg.c: Likewise.
26769 * tree-cfg.h: Likewise.
26770 * tree-dfa.c: Likewise.
26771 * tree-dfa.h: Likewise.
26772 * tree-dump.c (dump_function): Likewise.
26773 * tree-dump.h (struct dump_info): Likewise.
26774 * tree-pretty-print.c: Likewise.
26775 * tree-pretty-print.h: Likewise.
26776 * tree-ssa-live.c: Likewise.
26777 * tree-ssa-live.h: Likewise.
26778 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
26779 * tree-vect-loop.c: Likewise.
26780 * tree-vect-slp.c: Likewise.
26782 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
26783 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26785 PR tree-optimization/80457
26786 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
26787 of all arguments to a statement as scalar_to_vec operations.
26788 (vectorizable_call): Adjust call to vect_model_simple_cost for
26790 (vectorizable_conversion): Likewise.
26791 (vectorizable_assignment): Likewise.
26792 (vectorizable_shift): Likewise.
26793 (vectorizable_operation): Likewise.
26794 (vectorizable_comparison): Likewise.
26795 (vect_is_simple_cond): Record the def types for operands.
26796 (vectorizable_condition): Likewise, call vect_model_simple_cost.
26797 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
26798 for statement argument count.
26800 2017-05-16 Carl Love <cel@us.ibm.com>
26802 * config/rs6000/rs6000-c: Add support for built-in functions
26803 vector unsigned long long vec_bperm (vector unsigned long long,
26804 vector unsigned char)
26805 vector signed long long vec_mule (vector signed int,
26807 vector unsigned long long vec_mule (vector unsigned int,
26808 vector unsigned int)
26809 vector signed long long vec_mulo (vector signed int,
26811 vector unsigned long long vec_mulo (vector unsigned int,
26812 vector unsigned int)
26813 vector signed char vec_sldw (vector signed char,
26814 vector signed char,
26816 vector unsigned char vec_sldw (vector unsigned char,
26817 vector unsigned char,
26819 vector signed short vec_sldw (vector signed short,
26820 vector signed short,
26822 vector unsigned short vec_sldw (vector unsigned short,
26823 vector unsigned short,
26825 vector signed int vec_sldw (vector signed int,
26828 vector unsigned int vec_sldw (vector unsigned int,
26829 vector unsigned int,
26831 vector signed long long vec_sldw (vector signed long long,
26832 vector signed long long,
26834 vector unsigned long long vec_sldw (vector unsigned long long,
26835 vector unsigned long long,
26837 * config/rs6000/rs6000-c: Add support for built-in functions
26838 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
26839 * config/rs6000/altivec.h: Add defintion for vec_sldw.
26840 * doc/extend.texi: Update the built-in documentation for the
26841 new built-in functions.
26843 2017-05-16 Marek Polacek <polacek@redhat.com>
26847 * tree.c (save_expr): Don't fold the expression.
26849 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
26851 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
26852 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
26853 and (?*y,m). Update insn attributes.
26855 2017-05-16 Martin Liska <mliska@suse.cz>
26857 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
26858 flags argument of print_gimple_stmt, print_gimple_expr,
26859 print_generic_stmt and print_generic_expr.
26860 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
26861 * coretypes.h: Likewise.
26862 * except.c (dump_eh_tree): Likewise.
26863 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
26864 * gimple-pretty-print.h: Likewise.
26865 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
26866 (backprop::push_to_worklist): Likewise.
26867 (backprop::pop_from_worklist): Likewise.
26868 (backprop::process_use): Likewise.
26869 (backprop::intersect_uses): Likewise.
26870 (note_replacement): Likewise.
26871 * gimple-ssa-store-merging.c
26872 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
26873 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
26874 (pass_store_merging::execute): Likewise.
26875 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
26876 (ssa_base_cand_dump_callback): Likewise.
26877 (dump_incr_vec): Likewise.
26878 (replace_refs): Likewise.
26879 (replace_mult_candidate): Likewise.
26880 (create_add_on_incoming_edge): Likewise.
26881 (create_phi_basis): Likewise.
26882 (insert_initializers): Likewise.
26883 (all_phi_incrs_profitable): Likewise.
26884 (introduce_cast_before_cand): Likewise.
26885 (replace_one_candidate): Likewise.
26886 * gimplify.c (gimplify_expr): Likewise.
26887 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
26888 (set_rename): Likewise.
26889 (rename_uses): Likewise.
26890 (copy_loop_phi_nodes): Likewise.
26891 (add_close_phis_to_merge_points): Likewise.
26892 (copy_loop_close_phi_args): Likewise.
26893 (copy_cond_phi_args): Likewise.
26894 (graphite_copy_stmts_from_block): Likewise.
26895 (translate_pending_phi_nodes): Likewise.
26896 * graphite-poly.c (print_pdr): Likewise.
26897 (dump_gbb_cases): Likewise.
26898 (dump_gbb_conditions): Likewise.
26899 (print_scop_params): Likewise.
26900 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
26901 (build_cross_bb_scalars_use): Likewise.
26902 (gather_bbs::before_dom_children): Likewise.
26903 * hsa-dump.c (dump_hsa_immed): Likewise.
26904 * ipa-cp.c (print_ipcp_constant_value): Likewise.
26905 (get_replacement_map): Likewise.
26906 * ipa-inline-analysis.c (dump_condition): Likewise.
26907 (estimate_function_body_sizes): Likewise.
26908 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
26909 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
26910 * ipa-prop.c (ipa_dump_param): Likewise.
26911 (ipa_print_node_jump_functions_for_edge): Likewise.
26912 (ipa_modify_call_arguments): Likewise.
26913 (ipa_modify_expr): Likewise.
26914 (ipa_dump_param_adjustments): Likewise.
26915 (ipa_dump_agg_replacement_values): Likewise.
26916 (ipcp_modif_dom_walker::before_dom_children): Likewise.
26917 * ipa-pure-const.c (check_stmt): Likewise.
26918 (pass_nothrow::execute): Likewise.
26919 * ipa-split.c (execute_split_functions): Likewise.
26920 * omp-offload.c (dump_oacc_loop_part): Likewise.
26921 (dump_oacc_loop): Likewise.
26922 * trans-mem.c (tm_log_emit): Likewise.
26923 (tm_memopt_accumulate_memops): Likewise.
26924 (dump_tm_memopt_set): Likewise.
26925 (dump_tm_memopt_transform): Likewise.
26926 * tree-cfg.c (gimple_verify_flow_info): Likewise.
26927 (print_loop): Likewise.
26928 * tree-chkp-opt.c (chkp_print_addr): Likewise.
26929 (chkp_gather_checks_info): Likewise.
26930 (chkp_get_check_result): Likewise.
26931 (chkp_remove_check_if_pass): Likewise.
26932 (chkp_use_outer_bounds_if_possible): Likewise.
26933 (chkp_reduce_bounds_lifetime): Likewise.
26934 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
26935 (chkp_mark_completed_bounds): Likewise.
26936 (chkp_register_incomplete_bounds): Likewise.
26937 (chkp_mark_invalid_bounds): Likewise.
26938 (chkp_maybe_copy_and_register_bounds): Likewise.
26939 (chkp_build_returned_bound): Likewise.
26940 (chkp_get_bound_for_parm): Likewise.
26941 (chkp_build_bndldx): Likewise.
26942 (chkp_get_bounds_by_definition): Likewise.
26943 (chkp_generate_extern_var_bounds): Likewise.
26944 (chkp_get_bounds_for_decl_addr): Likewise.
26945 * tree-chrec.c (chrec_apply): Likewise.
26946 * tree-data-ref.c (dump_data_reference): Likewise.
26947 (dump_subscript): Likewise.
26948 (dump_data_dependence_relation): Likewise.
26949 (analyze_overlapping_iterations): Likewise.
26950 * tree-inline.c (expand_call_inline): Likewise.
26951 (tree_function_versioning): Likewise.
26952 * tree-into-ssa.c (dump_defs_stack): Likewise.
26953 (dump_currdefs): Likewise.
26954 (dump_names_replaced_by): Likewise.
26955 (dump_update_ssa): Likewise.
26956 (update_ssa): Likewise.
26957 * tree-object-size.c (pass_object_sizes::execute): Likewise.
26958 * tree-parloops.c (build_new_reduction): Likewise.
26959 (try_create_reduction_list): Likewise.
26960 (ref_conflicts_with_region): Likewise.
26961 (oacc_entry_exit_ok_1): Likewise.
26962 (oacc_entry_exit_single_gang): Likewise.
26963 * tree-pretty-print.h: Likewise.
26964 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
26965 (get_scalar_evolution): Likewise.
26966 (add_to_evolution): Likewise.
26967 (get_loop_exit_condition): Likewise.
26968 (analyze_evolution_in_loop): Likewise.
26969 (analyze_initial_condition): Likewise.
26970 (analyze_scalar_evolution): Likewise.
26971 (instantiate_scev): Likewise.
26972 (number_of_latch_executions): Likewise.
26973 (gather_chrec_stats): Likewise.
26974 (final_value_replacement_loop): Likewise.
26975 (scev_const_prop): Likewise.
26976 * tree-sra.c (dump_access): Likewise.
26977 (disqualify_candidate): Likewise.
26978 (create_access): Likewise.
26979 (reject): Likewise.
26980 (maybe_add_sra_candidate): Likewise.
26981 (create_access_replacement): Likewise.
26982 (analyze_access_subtree): Likewise.
26983 (analyze_all_variable_accesses): Likewise.
26984 (sra_modify_assign): Likewise.
26985 (initialize_constant_pool_replacements): Likewise.
26986 (find_param_candidates): Likewise.
26987 (decide_one_param_reduction): Likewise.
26988 (replace_removed_params_ssa_names): Likewise.
26989 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
26990 * tree-ssa-copy.c (dump_copy_of): Likewise.
26991 (copy_prop_visit_cond_stmt): Likewise.
26992 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
26993 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
26994 (record_equivalences_from_stmt): Likewise.
26995 * tree-ssa-dse.c (compute_trims): Likewise.
26996 (delete_dead_call): Likewise.
26997 (delete_dead_assignment): Likewise.
26998 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
26999 (forward_propagate_into_cond): Likewise.
27000 (pass_forwprop::execute): Likewise.
27001 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
27002 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
27004 (move_computations_worker): Likewise.
27005 (execute_sm): Likewise.
27006 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
27007 (remove_exits_and_undefined_stmts): Likewise.
27008 (remove_redundant_iv_tests): Likewise.
27009 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
27010 (adjust_iv_update_pos): Likewise.
27011 * tree-ssa-math-opts.c (bswap_replace): Likewise.
27012 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
27013 (value_replacement): Likewise.
27014 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
27015 * tree-ssa-pre.c (print_pre_expr): Likewise.
27016 (get_representative_for): Likewise.
27017 (create_expression_by_pieces): Likewise.
27018 (insert_into_preds_of_block): Likewise.
27019 (eliminate_insert): Likewise.
27020 (eliminate_dom_walker::before_dom_children): Likewise.
27021 (eliminate): Likewise.
27022 (remove_dead_inserted_code): Likewise.
27023 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
27024 * tree-ssa-reassoc.c (get_rank): Likewise.
27025 (eliminate_duplicate_pair): Likewise.
27026 (eliminate_plus_minus_pair): Likewise.
27027 (eliminate_not_pairs): Likewise.
27028 (undistribute_ops_list): Likewise.
27029 (eliminate_redundant_comparison): Likewise.
27030 (update_range_test): Likewise.
27031 (optimize_range_tests_var_bound): Likewise.
27032 (optimize_vec_cond_expr): Likewise.
27033 (rewrite_expr_tree): Likewise.
27034 (rewrite_expr_tree_parallel): Likewise.
27035 (linearize_expr): Likewise.
27036 (break_up_subtract): Likewise.
27037 (linearize_expr_tree): Likewise.
27038 (attempt_builtin_powi): Likewise.
27039 (attempt_builtin_copysign): Likewise.
27040 (transform_stmt_to_copy): Likewise.
27041 (transform_stmt_to_multiply): Likewise.
27042 (dump_ops_vector): Likewise.
27043 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
27044 (print_scc): Likewise.
27045 (set_ssa_val_to): Likewise.
27046 (visit_reference_op_store): Likewise.
27047 (visit_use): Likewise.
27048 (sccvn_dom_walker::before_dom_children): Likewise.
27049 (run_scc_vn): Likewise.
27050 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
27052 (expr_hash_elt::print): Likewise.
27053 (const_and_copies::pop_to_marker): Likewise.
27054 (const_and_copies::record_const_or_copy_raw): Likewise.
27055 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27056 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27057 (dump_predicates): Likewise.
27058 (find_uninit_use): Likewise.
27059 (warn_uninitialized_phi): Likewise.
27060 (pass_late_warn_uninitialized::execute): Likewise.
27061 * tree-ssa.c (verify_vssa): Likewise.
27062 (verify_ssa): Likewise.
27063 (maybe_optimize_var): Likewise.
27064 * tree-vrp.c (dump_value_range): Likewise.
27065 (dump_all_value_ranges): Likewise.
27066 (dump_asserts_for): Likewise.
27067 (register_edge_assert_for_2): Likewise.
27068 (vrp_visit_cond_stmt): Likewise.
27069 (vrp_visit_switch_stmt): Likewise.
27070 (vrp_visit_stmt): Likewise.
27071 (vrp_visit_phi_node): Likewise.
27072 (simplify_cond_using_ranges_1): Likewise.
27073 (fold_predicate_in): Likewise.
27074 (evrp_dom_walker::before_dom_children): Likewise.
27075 (evrp_dom_walker::push_value_range): Likewise.
27076 (evrp_dom_walker::pop_value_range): Likewise.
27077 (execute_early_vrp): Likewise.
27079 2017-05-16 Richard Biener <rguenther@suse.de>
27081 * dwarf2out.c (loc_list_from_tree_1): Do not create
27082 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27084 2017-05-16 Richard Biener <rguenther@suse.de>
27086 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27088 (note_variable_value_in_expr): If we resolved the decl ref
27089 do not push to the stack.
27091 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27093 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27094 operations in fast-math mode.
27095 (vaddq_f16): Likewise.
27096 (vmul_f16): Likewise.
27097 (vmulq_f16): Likewise.
27098 (vsub_f16): Likewise.
27099 (vsubq_f16): Likewise.
27100 * config/arm/neon.md (add<mode>3): New.
27102 (fma:<VH:mode>3): New. Also remove outdated comment.
27105 2017-05-16 Martin Liska <mliska@suse.cz>
27109 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27110 (odr_types_equivalent_p): Likewise.
27112 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
27114 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
27116 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
27119 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
27120 non-interunit SSE move alternatives with '?'.
27121 (zero-extendsidi peephole2): New peephole to skip intermediate
27122 general register in SSE zero-extend sequence.
27124 2017-05-15 Jeff Law <law@redhat.com>
27126 * reorg.c (relax_delay_slots): Create a new variable to hold
27127 the temporary target rather than clobbering TARGET_LABEL.
27129 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
27130 missing argument to extract_bit_field call.
27131 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
27133 2017-05-15 Martin Liska <mliska@suse.cz>
27136 * gcc.c (process_command): Do not allow empty argument of -o option.
27138 2017-05-15 Renlin Li <renlin.li@arm.com>
27140 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
27141 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
27142 * config/aarch64/constraints.md (Usf): Add long call check.
27143 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
27144 (call_value): Likewise.
27145 (sibcall): Likewise.
27146 (sibcall_value): Likewise.
27148 (call_value_insn): New.
27149 (sibcall_insn): Update rtx pattern.
27150 (sibcall_value_insn): Likewise.
27151 (call_internal): Remove.
27152 (call_value_internal): Likewise.
27153 (sibcall_internal): Likewise.
27154 (sibcall_value_internal): Likewise.
27155 (call_reg): Likewise.
27156 (call_symbol): Likewise.
27157 (call_value_reg): Likewise.
27158 (call_value_symbol): Likewise.
27160 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
27163 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
27165 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
27167 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
27168 compatible with CCGOCmode and with CCZmode.
27170 2017-05-14 Martin Sebor <msebor@redhat.com>
27172 PR middle-end/77671
27173 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
27174 (gimple_fold_builtin_snprintf): Same.
27175 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
27176 (gimple_fold_builtin_snprintf): Same.
27177 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
27178 of character types.
27179 (is_call_safe): New function.
27180 (try_substitute_return_value): Call it.
27181 (try_simplify_call): New function.
27182 (pass_sprintf_length::handle_gimple_call): Call it.
27184 2017-05-14 Martin Sebor <msebor@redhat.com>
27186 PR middle-end/80669
27187 * builtins.c (expand_builtin_stpncpy): Simplify.
27189 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
27191 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
27192 * config/i386/i386.h
27193 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
27194 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
27195 (struct machine_function): Add new members call_ms2sysv,
27196 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
27197 (struct machine_frame_state): New fields sp_realigned and
27198 sp_realigned_offset.
27199 * config/i386/i386.c
27200 (enum xlogue_stub): New enum.
27201 (enum xlogue_stub_sets): New enum.
27202 (class xlogue_layout): New class.
27203 (struct ix86_frame): New fields stack_realign_allocate_offset,
27204 stack_realign_offset and outlined_save_offset. Modify comments to
27205 detail stack layout when using out-of-line stubs.
27206 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
27207 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
27208 -mcall-ms2sysv-xlogues.
27209 (stub_managed_regs): New static variable.
27210 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
27211 registers managed by out-of-line stub.
27212 (disable_call_ms2sysv_xlogues): New function.
27213 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
27214 m->call_ms2sysv when appropriate and compute frame layout for
27216 (sp_valid_at, fp_valid_at): New inline functions.
27217 (choose_basereg): New function.
27218 (choose_baseaddr): Add align parameter, use choose_basereg and modify
27220 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
27221 Use align parameter of choose_baseaddr to generated aligned SSE movs
27223 (pro_epilogue_adjust_stack): Modify to track
27224 machine_frame_state::sp_realigned.
27225 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
27226 (ix86_nsaved_sseregs): Likewise.
27227 (ix86_emit_save_regs): Likewise.
27228 (ix86_emit_save_regs_using_mov): Likewise.
27229 (ix86_emit_save_sse_regs_using_mov): Likewise.
27230 (get_scratch_register_on_entry): Likewise.
27231 (gen_frame_set): New function.
27232 (gen_frame_load): Likewise.
27233 (gen_frame_store): Likewise.
27234 (emit_outlined_ms2sysv_save): Likewise.
27235 (emit_outlined_ms2sysv_restore): Likewise.
27236 (ix86_expand_prologue): Modify stack re-alignment code and call
27237 emit_outlined_ms2sysv_save when appropriate.
27238 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
27239 parameter rtx_insn *insn, which allows the function to be used to only
27240 generate the notes.
27241 (ix86_expand_epilogue): Modify validity checks of frame and stack
27242 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
27243 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
27244 * config/i386/predicates.md
27245 (save_multiple): New predicate.
27246 (restore_multiple): Likewise.
27247 * config/i386/sse.md
27248 (save_multiple<mode>): New pattern.
27249 (save_multiple_realign<mode>): Likewise.
27250 (restore_multiple<mode>): Likewise.
27251 (restore_multiple_and_return<mode>): Likewise.
27252 (restore_multiple_leave_return<mode>): Likewise.
27253 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
27255 2017-05-14 Julia Koval <julia.koval@intel.com>
27257 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
27258 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
27259 (__builtin_ia32_xsetbv): New builtins.
27260 * config/i386/i386.c (ix86_expand_special_args_builtin):
27262 (ix86_expand_builtin): Special expand for new intrinsics.
27263 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
27264 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
27265 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
27267 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27269 * cfganal.c (inverted_post_order_compute): Change argument type
27271 * cfganal.h (inverted_post_order_compute): Adjust prototype.
27272 * df-core.c (rest_of_handle_df_initialize): Adjust.
27273 (rest_of_handle_df_finish): Likewise.
27274 (df_analyze_1): Likewise.
27275 (df_analyze): Likewise.
27276 (loop_inverted_post_order_compute): Change argument to be a vec *.
27277 (df_analyze_loop): Adjust.
27278 (df_get_n_blocks): Likewise.
27279 (df_get_postorder): Likewise.
27280 * df.h (struct df_d): Change field to be a vec.
27281 * lcm.c (compute_laterin): Adjust.
27282 (compute_available): Likewise.
27283 * lra-lives.c (lra_create_live_ranges_1): Likewise.
27284 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
27285 * tree-ssa-pre.c (compute_antic): Likewise.
27287 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27289 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
27290 (depth_first_search::depth_first_search): Change structure init
27291 function to this constructor.
27292 (depth_first_search::add_bb): Rename function to this member.
27293 (depth_first_search::execute): Likewise.
27294 (flow_dfs_compute_reverse_finish): Adjust.
27296 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27298 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
27299 (longest_simple_path): Likewise.
27300 * shrink-wrap.c (spread_components): Likewise.
27301 (disqualify_problematic_components): Likewise.
27302 (emit_common_heads_for_components): Likewise.
27303 (emit_common_tails_for_components): Likewise.
27304 (insert_prologue_epilogue_for_components): Likewise.
27306 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27308 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
27311 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27313 * df-core.c (df_set_blocks): Start using auto_bitmap.
27314 (df_compact_blocks): Likewise.
27315 * df-problems.c (df_rd_confluence_n): Likewise.
27316 * df-scan.c (df_insn_rescan_all): Likewise.
27317 (df_process_deferred_rescans): Likewise.
27318 (df_update_entry_block_defs): Likewise.
27319 (df_update_exit_block_uses): Likewise.
27320 (df_entry_block_bitmap_verify): Likewise.
27321 (df_exit_block_bitmap_verify): Likewise.
27322 (df_scan_verify): Likewise.
27323 * lra-constraints.c (lra_constraints): Likewise.
27324 (undo_optional_reloads): Likewise.
27325 (lra_undo_inheritance): Likewise.
27326 * lra-remat.c (calculate_gen_cands): Likewise.
27327 (do_remat): Likewise.
27328 * lra-spills.c (assign_spill_hard_regs): Likewise.
27329 (spill_pseudos): Likewise.
27330 * tree-ssa-pre.c (bitmap_set_and): Likewise.
27331 (bitmap_set_subtract_values): Likewise.
27333 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27335 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
27336 management with auto_bitmap.
27337 (fix_inter_tick): Likewise.
27338 (fix_recovery_deps): Likewise.
27339 * ira.c (add_store_equivs): Likewise.
27340 (find_moveable_pseudos): Likewise.
27341 (split_live_ranges_for_shrink_wrap): Likewise.
27342 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
27343 (rtx_reuse_manager::seen_def_p): Likewise.
27344 (rtx_reuse_manager::set_seen_def): Likewise.
27345 * print-rtl.h (class rtx_reuse_manager): Likewise.
27347 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27349 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
27351 (migrate_btr_def): Likewise.
27352 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
27353 * df-core.c (loop_post_order_compute): Likewise.
27354 (loop_inverted_post_order_compute): Likewise.
27355 * hsa-common.h: Likewise.
27356 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
27357 * init-regs.c (initialize_uninitialized_regs): Likewise.
27358 * ipa-inline.c (resolve_noninline_speculation): Likewise.
27359 (inline_small_functions): Likewise.
27360 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
27361 * ira.c (combine_and_move_insns): Likewise.
27362 (build_insn_chain): Likewise.
27363 * loop-invariant.c (find_invariants): Likewise.
27364 * lower-subreg.c (propagate_pseudo_copies): Likewise.
27365 * predict.c (tree_predict_by_opcode): Likewise.
27366 (predict_paths_leading_to): Likewise.
27367 (predict_paths_leading_to_edge): Likewise.
27368 (estimate_loops_at_level): Likewise.
27369 (estimate_loops): Likewise.
27370 * shrink-wrap.c (try_shrink_wrapping): Likewise.
27371 (spread_components): Likewise.
27372 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
27373 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
27374 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
27375 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
27376 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
27377 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
27378 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
27379 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
27380 (mark_threaded_blocks): Likewise.
27381 (thread_through_all_blocks): Likewise.
27382 * tree-ssa.c (verify_ssa): Likewise.
27383 (execute_update_addresses_taken): Likewise.
27384 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
27386 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27388 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
27390 (post_order_compute): Likewise.
27391 (inverted_post_order_compute): Likewise.
27392 (pre_and_rev_post_order_compute_fn): Likewise.
27394 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27396 * genrecog.c (int_set::int_set): Explicitly construct our
27397 auto_vec base class.
27398 * vec.h (auto_vec::auto_vec): New constructor.
27400 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27402 * bitmap.h (class auto_bitmap): New constructor taking
27403 bitmap_obstack * argument.
27405 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27407 * bitmap.h (class auto_bitmap): Change type of m_bits to
27408 bitmap_head, and adjust ctor / dtor and member operators.
27410 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
27412 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
27413 when returned register mode doesn't match original mode.
27415 2017-05-12 Jeff Law <law@redhat.com>
27416 Jakub Jelinek <jakub@redhat.com>
27418 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
27419 we look for cc setter after the compare-elim changes.
27420 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
27421 within the vector to match what compare-elim now expects.
27422 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
27423 (xorsi3_flags, one_cmplsi2_flags): Likewise.
27425 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
27426 after the compare-elim changes.
27427 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
27428 the vector to match what compare-elim now expects.
27429 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
27430 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
27431 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
27432 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
27433 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
27435 * config/visium/visium.c (single_set_and_flags): Fix where
27436 we look for cc setter after the compare-elim changes.
27437 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
27438 with the vector to match what compare-elim now expects.
27439 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
27440 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
27441 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
27442 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
27443 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
27444 (neg<mode>2_insn_set_overflow): Likewise.
27446 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
27448 PR middle-end/79794
27449 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
27450 maybe_expand_insn call, set ops[0].target. If still set after call,
27451 set alt_rtl. Add extra arg to recursive calls.
27452 (extract_bit_field): Add alt_rtl argument. Pass to
27454 * expmed.h (extract_bit_field): Fix prototype.
27455 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
27456 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
27457 to extract_bit_field_calls.
27458 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
27459 Pass alt_rtl to extract_bit_field calls.
27460 * calls.c (store_unaligned_arguments_into_psuedos)
27461 load_register_parameters): Pass extra NULL to extract_bit_field calls.
27462 * optabs.c (maybe_legitimize_operand): Clear op->target when call
27464 * optabs.h (struct expand_operand): Add target bitfield.
27466 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27468 * compare-elim.c (try_eliminate_compare): Canonicalize
27469 operation with embedded compare to
27470 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
27471 (set (reg) (operation)].
27473 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
27475 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27478 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
27479 cost of adding a carry flag for ADC instruction.
27480 [case MINUS]: Ignore the cost of subtracting a carry flag
27481 for SBB instruction.
27483 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
27485 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
27487 * config/rs6000/bmiintrin.h: New file.
27488 * config/rs6000/bmi2intrin.h: New file.
27489 * config/rs6000/x86intrin.h: New file.
27491 2017-05-12 Jeff Law <law@redhat.com>
27493 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
27496 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
27498 PR middle-end/80707
27499 * tree-cfg.c: Remove cfg edges of unreachable case statements.
27501 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27503 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27504 early expansion of vector divide builtins.
27505 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
27506 builtins identified as having unsigned arguments.
27508 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27510 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
27511 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
27512 expansion of vector logical operations (and, andc, or, xor,
27515 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27517 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
27518 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
27520 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27522 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27523 early GIMPLE expansion of vector multiplies.
27525 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27527 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
27528 TARGET_HAVE_MOVT conditional.
27529 (movt splitter): Likewise.
27531 2017-05-12 Richard Biener <rguenther@suse.de>
27533 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
27534 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27535 Fold all stmts not inplace.
27537 2017-05-12 Richard Biener <rguenther@suse.de>
27539 PR tree-optimization/80713
27540 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
27541 inserted_exprs bit for not removed stmts.
27543 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
27545 PR middle-end/69921
27546 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
27547 parallelized" attribute for parallelized OpenACC kernels.
27548 * omp-offload.c (execute_oacc_device_lower): Use it.
27550 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
27551 Set "oacc kernels" attribute.
27552 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
27553 parameter. Adjust all users.
27554 (oacc_fn_attrib_kernels_p): Remove function.
27555 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
27556 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
27557 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
27558 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
27559 assert "oacc kernels" attribute is set.
27561 2017-05-11 Carl Love <cel@us.ibm.com>
27563 * config/rs6000/rs6000-c: Add support for built-in functions
27564 vector unsigned char vec_popcnt (vector signed char)
27565 vector unsigned char vec_popcnt (vector unsigned char)
27566 vector unsigned short vec_popcnt (vector signed short)
27567 vector unsigned short vec_popcnt (vector unsigned short)
27568 vector unsigned int vec_popcnt (vector signed int)
27569 vector unsigned int vec_popcnt (vector unsigned int)
27570 vector unsigned long long vec_popcnt (vector signed long long)
27571 vector unsigned long long vec_popcnt (vector unsigned long long)
27572 vector signed long long vec_slo (vector signed long long,
27573 vector signed char)
27574 vector signed long long vec_slo (vector signed long long,
27575 vector unsigned char)
27576 vector unsigned long long vec_slo (vector unsigned long long,
27577 vector signed char)
27578 vector unsigned long long vec_slo (vector unsigned long long,
27579 vector unsigned char)
27580 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
27581 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
27582 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
27583 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
27584 * doc/extend.texi: Update the built-in documentation file for the
27585 new built-in functions.
27587 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
27589 * attribs.h (sorted_attr_string): Move machine independent
27590 functions for target clone support from the i386 port to common
27591 code. Rename ix86_function_versions to common_function_versions.
27592 Rename make_name to make_unique_name.
27593 (common_function_versions): Likewise.
27594 (make_unique_name): Likewise.
27595 (make_dispatcher_decl): Likewise.
27596 (is_function_default_version): Likewise.
27597 * attribs.c (attr_strcmp): Likewise.
27598 (sorted_attr_string): Likewise.
27599 (common_function_versions): Likewise.
27600 (make_unique_name): Likewise.
27601 (make_dispatcher_decl): Likewise.
27602 (is_function_default_version): Likewise.
27603 * config/i386/i386.c (attr_strcmp): Likewise.
27604 (sorted_attr_string): Likewise.
27605 (ix86_function_versions): Likewise.
27606 (make_name): Likewise.
27607 (make_dispatcher_decl): Likewise.
27608 (is_function_default_version): Likewise.
27609 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
27611 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27614 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
27615 Account for direct move costs for vec_construct of integer
27618 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
27621 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
27622 (UNSPEC_STX_ATOMIC): Ditto.
27623 (loaddi_via_sse): New insn.
27624 (storedi_via_sse): Ditto.
27625 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
27626 Update corresponding peephole2 patterns.
27627 (atomic_storedi_fpu): Ditto.
27629 2017-05-11 Julia Koval <julia.koval@intel.com>
27631 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
27632 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
27634 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
27635 (__builtin_ia32_rsqrt14ss_mask): New builtins.
27636 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
27638 2017-05-11 Nathan Sidwell <nathan@acm.org>
27640 * graphite-poly.c: Include dumpfile.h.
27642 * dumpfle.h (dump_function): Declare here ...
27643 * tree-dump.h (dump_function): ... not here.
27644 * dumpfile.c: #include tree-cfg.h.
27645 (dump_function): Move here from ...
27646 * tree-dump.c (dump_function): ... here.
27647 * gimplify.c: #include splay-tree.h, not tree-dump.h.
27648 * graphite-poly.c: Don't include tree-dump.h.
27649 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
27650 * print-tree.c: Likewise.
27651 * stor-layout.c: Likewise.
27652 * tree-nested.c: Likewise.
27654 * dumpfile.c (dump_start): Use TDF_FLAGS.
27655 (dump_enable_all): Fix TDF_KIND check thinko.
27657 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
27659 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27660 array entries to represent two legal parameterizations of the
27661 overloaded __builtin_cmpb function, as represented by the
27662 P6_OV_BUILTIN_CMPB constant.
27663 (altivec_resolve_overloaded_builtin): Add special case handling
27664 for the __builtin_cmpb function, as represented by the
27665 P6_OV_BUILTIN_CMPB constant.
27666 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
27667 (BU_P6_64BIT_2): New macro.
27668 (BU_P6_OVERLOAD_2): New macro
27669 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
27670 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
27671 (CMPB): Add overload support to represent both 32-bit and 64-bit
27672 compare-bytes function.
27673 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
27674 support for TARGET_CMPB.
27675 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
27676 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27677 documentation of the __builtin_cmpb overloaded built-in function.
27679 2017-05-11 Richard Biener <rguenther@suse.de>
27681 PR tree-optimization/80705
27682 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
27683 bases are not vectorizable.
27685 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27687 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
27688 when counting register pressure.
27690 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27692 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
27693 (iv_ca_more_deps): Renamed to ...
27694 (iv_ca_compare_deps): ... this.
27695 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
27697 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27699 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
27701 (determine_group_iv_costs): ... here.
27702 (find_inv_vars_cb): Record inv var if it's not recorded before.
27704 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27706 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
27707 (get_shiftadd_cost): Ditto.
27709 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27711 * tree-ssa-address.c: Include header file.
27712 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
27714 (add_to_parts): Refactor.
27715 (addr_to_parts): New parameter. Update use of move_hint_to_base.
27716 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
27719 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27721 PR tree-optimization/53090
27722 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
27724 (extract_cond_operands): Detect condition with IV on both sides
27725 and return COMP_IV_EXPR_2.
27726 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
27727 (rewrite_use_compare): Simplify by removing call to function
27728 extract_cond_operands.
27730 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27732 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
27733 (extract_cond_operands): Detect condition comparing against non-
27734 invariant bound and return appropriate enum value.
27735 (find_interesting_uses_cond): Update use of extract_cond_operands.
27736 Handle its return value accordingly.
27737 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
27739 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27741 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
27742 nonlinear iv_use computation in loop invariant sensitive way.
27744 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27746 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
27747 (find_iv_candidates): Call relate_compare_use_with_all_cands.
27749 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27751 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
27752 (dump_cand): Support iv_cand.inv_exprs.
27753 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
27755 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
27758 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27760 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
27762 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
27763 as local function. Include necessary header files.
27764 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
27766 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27768 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
27770 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27772 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
27773 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
27774 RSHIFT_EXPR and BIT_NOT_EXPR.
27776 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27778 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
27779 (adjust_setup_cost): New parameter supporting round up adjustment.
27780 (struct address_cost_data): Delete.
27781 (force_expr_to_var_cost): Don't bound cost with spill_cost.
27782 (split_address_cost, ptr_difference_cost): Delete.
27783 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
27784 (struct ainc_cost_data): New struct.
27785 (get_address_cost_ainc): New function.
27786 (get_address_cost, get_computation_cost): Reimplement.
27787 (determine_group_iv_cost_address): Record inv_expr for all uses of
27789 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
27790 (iv_ca_has_deps): Reimplemented to ...
27791 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
27793 (iv_ca_extend): Call iv_ca_more_deps.
27795 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27797 * tree-ssa-address.c (struct mem_address): Move to header file.
27798 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
27799 * tree-ssa-address.h (struct mem_address): Move from C file.
27800 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
27802 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27804 * tree-affine.h (aff_combination_type): New interface.
27805 (aff_combination_zero_p): Remove static.
27806 (aff_combination_const_p): New interface.
27807 (aff_combination_singleton_var_p): New interfaces.
27809 2017-05-11 Richard Biener <rguenther@suse.de>
27811 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27812 Skip unreachable blocks and destinations.
27813 (eliminate): Move stmt removal and fixup ...
27814 (fini_eliminate): ... here. Skip inserted exprs.
27815 (pass_pre::execute): Move fini_pre after fini_eliminate.
27816 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
27817 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
27818 PRE to get rid of dead code that has invalid SSA form and
27819 split critical edges again.
27821 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27823 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27825 2017-05-11 Richard Biener <rguenther@suse.de>
27827 * passes.c (execute_function_todo): Verify loops if they are
27828 said to be up-to-date.
27829 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
27830 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
27832 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
27835 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
27836 handle calling assemble_external ourself.
27839 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
27840 modes with zero size. Enhance comment.
27842 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27844 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
27845 built-ins for vec_xl and vec_xst with short and char pointer
27848 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27850 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
27851 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
27852 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
27853 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
27854 (_mm_maskz_min_round_ss): New intrinsics.
27855 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
27856 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
27857 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
27858 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
27859 (__builtin_ia32_minss_mask_round): New builtins.
27860 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27861 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27862 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
27864 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
27865 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
27867 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
27870 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27872 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
27873 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
27874 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
27875 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
27876 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
27877 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
27878 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
27879 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27880 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27881 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
27882 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
27883 (__builtin_ia32_mulss_mask_round): New builtins.
27884 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27885 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27886 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
27888 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
27889 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27891 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27894 2017-05-10 Julia Koval <julia.koval@intel.com>
27896 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
27897 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
27898 (_mm256_setr_m128i): New intrinsics.
27900 2017-05-10 Julia Koval <julia.koval@intel.com>
27902 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
27903 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
27904 (_mm_maskz_rcp14_ss): New intrinsics.
27905 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
27906 (__builtin_ia32_rcp14ss_mask): New builtins.
27907 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
27909 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
27911 PR tree-optimization/51513
27912 * tree-cfg.c (gimple_seq_unreachable_p): New function.
27913 (assert_unreachable_fallthru_edge_p): Use it.
27914 (group_case_labels_stmt): Likewise.
27915 * tree-cfg.h: Prototype it.
27916 * stmt.c: Include cfghooks.h and tree-cfg.h.
27917 (emit_case_dispatch_table) <gap_label>: New local variable.
27918 Use it to fill dispatch table gaps.
27919 Test for default_label before updating probabilities.
27920 (expand_case) <default_label>: Remove unneeded initialization.
27921 Test for unreachable default case statement and remove its edge.
27922 Set default_label accordingly.
27923 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
27925 2017-05-10 Carl Love <cel@us.ibm.com>
27927 * config/rs6000/rs6000-c: Add support for built-in functions
27928 vector signed char vec_neg (vector signed char)
27929 vector signed short int vec_neg (vector short int)
27930 vector signed int vec_neg (vector signed int)
27931 vector signed long long vec_neg (vector signed long long)
27932 vector float vec_neg (vector float)
27933 vector double vec_neg (vector double)
27934 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
27936 * config/rs6000/altivec.h: Add define for vec_neg
27937 * doc/extend.texi: Update the built-in documentation for the
27938 new built-in functions.
27940 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27942 PR tree-optimization/77644
27943 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
27945 2017-05-10 Nathan Sidwell <nathan@acm.org>
27947 * dumpfile.h (TDI_lang_all): New.
27948 (TDF_KIND): New. Renumber others
27949 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
27951 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
27953 (get_dump_file_name): Adjust suffix generation.
27954 (dump_enable_all): Use TDF_KIND.
27955 * doc/invoke.texi (-fdump-lang-all): Document.
27957 * dumpfile.h: Tabify.
27959 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
27962 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
27963 Move member access before delete.
27965 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
27967 * tree-inline.c (expand_call_inline): Split block at stmt
27970 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
27973 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
27974 are now unused after splitting mov{sf,sd}_hardfloat.
27975 (f32_lr2): Likewise.
27976 (f32_lm): Likewise.
27977 (f32_lm2): Likewise.
27978 (f32_li): Likewise.
27979 (f32_li2): Likewise.
27980 (f32_lv): Likewise.
27981 (f32_sr): Likewise.
27982 (f32_sr2): Likewise.
27983 (f32_sm): Likewise.
27984 (f32_sm2): Likewise.
27985 (f32_si): Likewise.
27986 (f32_si2): Likewise.
27987 (f32_sv): Likewise.
27988 (f32_dm): Likewise.
27989 (f32_vsx): Likewise.
27990 (f32_av): Likewise.
27991 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
27992 For movsf, order stores so the VSX stores occur before the GPR
27993 store which encourages the register allocator to use a traditional
27994 FPR instead of a GPR. For movsd, order the stores so that the GPR
27995 store comes before the VSX stores to allow the power6 to work.
27996 This is due to the power6 not having a 32-bit integer store
27997 instruction from a FPR.
27998 (movsf_hardfloat): Likewise.
27999 (movsd_hardfloat): Likewise.
28001 2017-05-09 Martin Sebor <msebor@redhat.com>
28003 PR translation/80280
28004 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
28007 PR translation/80280
28008 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
28009 data member added in r247778.
28010 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
28012 2017-05-09 Nathan Sidwell <nathan@acm.org>
28014 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
28016 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
28019 2017-05-09 Marek Polacek <polacek@redhat.com>
28021 * doc/invoke.texi: Fix typo.
28023 2017-05-09 Richard Biener <rguenther@suse.de>
28025 * tree-vrp.c (vrp_val_is_max): Adjust comment.
28026 (vrp_val_is_min): Likewise.
28027 (set_value_range_to_value): Likewise.
28028 (set_value_range_to_nonnegative): Likewise.
28029 (gimple_assign_nonzero_p): Likewise.
28030 (gimple_stmt_nonzero_p): Likewise.
28031 (vrp_int_const_binop): Likewise. Remove unreachable case.
28032 (adjust_range_with_scev): Adjust comments.
28033 (compare_range_with_value): Likewise.
28034 (extract_range_from_phi_node): Likewise.
28035 (test_for_singularity): Likewise.
28037 2017-05-09 Richard Biener <rguenther@suse.de>
28039 * tree-vrp.c (get_single_symbol): Add assert that we don't
28040 get overflowed constants as invariant part.
28041 (compare_values_warnv): Add comment before the TREE_NO_WARNING
28042 checks. Use wi::cmp instead of recursing for integer constants.
28043 (compare_values): Just ignore whether we assumed undefined
28044 overflow instead of failing the compare.
28045 (extract_range_for_var_from_comparison_expr): Add comment before the
28046 TREE_NO_WARNING sets.
28047 (test_for_singularity): Likewise.
28048 (extract_range_from_comparison): Do not disable optimization
28049 when we assumed undefined overflow.
28050 (extract_range_basic): Remove init of unused var.
28052 2017-05-09 Richard Biener <rguenther@suse.de>
28054 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28055 (extract_range_from_multiplicative_op_1): Adjust.
28056 (extract_range_from_binary_expr_1): Use int_const_binop.
28058 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28061 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28062 rs6000_store_data_bypass_p in seven define_bypass directives and
28063 in several comments.
28064 * config/rs6000/rs6000-protos.h: Add prototype for
28065 rs6000_store_data_bypass_p function.
28066 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28067 function implements slightly different (rs6000-specific) semantics
28068 than store_data_bypass_p, returning false rather than aborting
28069 with assertion error when arguments do not satisfy the
28070 requirements of store data bypass.
28071 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28072 rs6000_store_data_bypass_p.
28074 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28076 * config/xtensa/xtensa-protos.h
28077 (xtensa_initial_elimination_offset): New declaration.
28078 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28079 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28080 macro definition, add case for FRAME_POINTER_REGNUM when
28081 FRAME_GROWS_DOWNWARD.
28082 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28083 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28084 xtensa_initial_elimination_offset.
28086 2017-05-08 Nathan Sidwell <nathan@acm.org>
28088 * doc/invoke.texi: Alphabetize -fdump options.
28090 2017-05-08 Martin Sebor <msebor@redhat.com>
28092 PR translation/80280
28093 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28095 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28097 * target.def (compute_frame_layout): New optional target hook.
28098 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28099 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28100 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28102 * reload1.c (verify_initial_elim_offsets): Likewise.
28103 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28104 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28105 (arm_get_frame_offsets): Split up into this ...
28106 (arm_compute_frame_layout): ... and this function.
28108 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28110 * config/aarch64/constraints.md (Usa): New constraint.
28111 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
28113 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
28115 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
28116 with_multilib_list after it has been checked.
28118 2017-05-08 Richard Biener <rguenther@suse.de>
28120 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
28121 (bitmap_set_subtract_values): Likewise.
28123 2017-05-08 Richard Biener <rguenther@suse.de>
28125 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
28126 (gimple_assign_nonzero): ... this and remove strict_overflow_p
28128 (gimple_stmt_nonzero_warnv_p): Rename to ...
28129 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
28131 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
28132 (extract_range_basic): Adjust, do not disable propagation on
28133 strict overflow sensitive simplification.
28134 (vrp_visit_cond_stmt): Likewise.
28136 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
28138 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
28139 body size unconditionally.
28141 2017-05-07 Jeff Law <law@redhat.com>
28144 2017-05-06 Jeff Law <law@redhat.com>
28145 PR tree-optimization/78496
28146 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28149 PR tree-optimization/78496
28150 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28151 (simplify_stmt_using_ranges): Call it.
28152 (vrp_dom_walker::before_dom_children): Extract equivalences
28153 from an ASSERT_EXPR with an equality comparison against a
28156 2017-05-06 Jeff Law <law@redhat.com>
28158 PR tree-optimization/78496
28159 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28162 PR tree-optimization/78496
28163 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28164 (simplify_stmt_using_ranges): Call it.
28165 (vrp_dom_walker::before_dom_children): Extract equivalences
28166 from an ASSERT_EXPR with an equality comparison against a
28169 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28171 * lra-constraints.c (lra_copy_reg_equiv): New function.
28172 (split_reg): Use it to copy equivalence information from the
28173 original register to the spill register.
28175 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28177 PR rtl-optimization/75964
28178 * simplify-rtx.c (simplify_const_relational_operation): Remove
28179 invalid handling of comparisons of integer ABS.
28181 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
28183 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
28184 initialize to zero.
28185 (init_regs): Remove declaration.
28186 (function_arg_advance_32): Initialize error_p as boolean variable.
28188 2017-05-05 Nathan Sidwell <nathan@acm.org>
28190 * store-motion.c (remove_reachable_equiv_notes): Reformat long
28191 lines. Use for (;;).
28193 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28195 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
28196 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
28197 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
28198 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
28199 VF=2 that require versioning.
28201 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28203 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
28206 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28208 * diagnostic.h (diagnostic_override_option_index): Convert from
28209 macro to inline function.
28211 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28213 * diagnostic.c (last_module_changed_p): New function.
28214 (set_last_module): New function.
28215 (diagnostic_report_current_module): Convert macro usage to
28216 the above functions.
28217 * diagnostic.h (diagnostic_context::last_module): Strengthen
28218 from const line_map * to const line_map_ordinary *.
28219 (diagnostic_last_module_changed): Delete macro.
28220 (diagnostic_set_last_module): Delete macro.
28222 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28224 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
28225 with diagnostic_report_diagnostic.
28226 (diagnostic_n_impl_richloc): Likewise.
28227 * diagnostic.h (report_diagnostic): Delete macro.
28228 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
28229 with diagnostic_report_diagnostic.
28230 * substring-locations.c (format_warning_va): Likewise.
28232 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28234 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
28235 save/restor of format_spec. Move option-printing code to...
28236 (print_option_information): ...this new function, and
28237 reimplement by simply printing to the pretty_printer,
28238 rather than appending to the format string.
28240 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28242 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
28243 handling logic into...
28244 (update_effective_level_from_pragmas): ...this new function.
28246 2017-05-04 Andrew Waterman <andrew@sifive.com>
28248 * config/riscv/riscv.opt (mstrict-align): New option.
28249 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
28250 (SLOW_UNALIGNED_ACCESS): Define.
28251 (riscv_slow_unaligned_access): Declare.
28252 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
28254 (riscv_slow_unaligned_access): New variable.
28255 (rocket_tune_info): Set slow_unaligned_access to true.
28256 (optimize_size_tune_info): Set slow_unaligned_access to false.
28257 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
28258 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
28259 (riscv_option_override): Set riscv_slow_unaligned_access.
28260 * doc/invoke.texi: Add -mstrict-align to RISC-V.
28262 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
28264 * config/riscv/riscv.md: Unify indentation.
28266 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28271 * config/rs6000/rs6000.md (u code attribute): Add FIX and
28273 (extendsi<mode>2): Add support for doing sign extension via
28274 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
28275 don't have ISA 3.0 instructions.
28276 (extendsi<mode>2 splitter): Likewise.
28277 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
28278 generate the normal insns since SImode can now go in vector
28279 registers. Disallow the special UNSPECs needed for previous
28280 machines to hide SImode being used. Add new insns
28281 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
28282 (fix_trunc<mode>si2_stfiwx): Likewise.
28283 (fix_trunc<mode>si2_internal): Likewise.
28284 (fixuns_trunc<mode>si2): Likewise.
28285 (fixuns_trunc<mode>si2_stfiwx): Likewise.
28286 (fctiw<u>z_<mode>_smallint): Likewise.
28287 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
28288 of floating point to 32-bit integer from doing a direct move to
28289 the GPR registers to do a store.
28290 (fctiwz_<mode>): Break long line.
28292 2017-05-05 Bin Cheng <bin.cheng@arm.com>
28294 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
28295 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
28296 (addr_list, addr_offset_valid_p): New.
28297 (split_address_groups): Check offset validity with above function.
28298 (gt-tree-ssa-loop-ivopts.h): Include header file.
28300 2017-05-05 Nathan Sidwell <nathan@acm.org>
28302 * config.gcc (arm*-*-*): Add missing 'fi'.
28304 2017-05-05 Steve Ellcey <sellcey@cavium.com>
28306 * doc/invoke.texi (-fopt-info): Explicitly say order of options
28307 included in -fopt-info does not matter.
28308 * doc/optinfo.texi (-fopt-info): Fix description of default
28309 behavour. Explicitly say order of options included in -fopt-info
28312 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28314 * config.gcc: Allow combinations of aprofile and rmprofile values for
28315 --with-multilib-list.
28316 * config/arm/t-multilib: New file.
28317 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
28318 variables. Remove setting of ISA and floating-point ABI in
28319 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
28320 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
28321 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
28322 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
28324 * config/arm/t-rmprofile: Likewise except for the matches changes.
28325 * doc/install.texi (--with-multilib-list): Document the combination of
28326 aprofile and rmprofile values and warn about pitfalls in doing that.
28328 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28330 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
28331 (movdi_aarch64): Likewise.
28333 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28335 PR tree-optimization/80632
28336 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
28338 (build_arrays): Initialize it for virtual phis.
28339 (fix_phi_nodes): Use it for virtual phis.
28341 PR tree-optimization/80558
28342 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
28343 [x, y] op z into [x op, y op z] for op & or | if conditions
28346 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
28347 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28350 * config/arm/arm.md (use_literal_pool): Remove.
28351 (64-bit immediate split): No longer takes cost into consideration
28352 if arm_disable_literal_pool is enabled.
28353 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
28354 used when arm_disable_literal_pool is enabled.
28355 (arm_max_const_double_inline_cost): Remove use of
28356 arm_disable_literal_pool.
28357 (push_minipool_fix): Add assert.
28358 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
28359 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
28360 (no_literal_pool_sf_immediate): New.
28362 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28364 PR tree-optimization/80613
28365 * tree-ssa-dce.c (propagate_necessity): Remove cases for
28366 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
28368 2017-05-05 Richard Biener <rguenther@suse.de>
28370 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
28372 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
28374 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
28375 of this flag from insn conditions due to removal from r247495.
28377 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28379 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
28381 (arm_early_store_addr_dep_ptr): Likewise.
28382 * config/arm/aarch-common-protos.h
28383 (arm_early_load_addr_dep_ptr): Add prototype.
28384 (arm_early_store_addr_dep_ptr): Likewise.
28385 * config/arm/cortex-a53.md: Add new bypasses.
28387 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28389 * tree.c (next_type_uid): Change type to unsigned.
28390 (type_hash_canon): Decrement back next_type_uid if
28391 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
28392 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
28395 2017-05-04 Martin Sebor <msebor@redhat.com>
28397 * builtins.c: Fix a trivial typo in a comment.
28399 PR middle-end/79234
28400 * builtins.c (check_sizes): Adjust to handle reading past the end.
28401 Avoid printing excessive upper bound of ranges. Use %E to print
28402 tree nodes instead of converting them to %wu.
28403 (expand_builtin_memchr): New function.
28404 (compute_dest_size): Rename...
28405 (compute_objsize): ...to this.
28406 (expand_builtin_memcpy): Adjust.
28407 (expand_builtin_mempcpy): Adjust.
28408 (expand_builtin_strcat): Adjust.
28409 (expand_builtin_strcpy): Adjust.
28410 (check_strncat_sizes): Adjust.
28411 (expand_builtin_strncat): Adjust.
28412 (expand_builtin_strncpy): Adjust and simplify.
28413 (expand_builtin_memset): Adjust.
28414 (expand_builtin_bzero): Adjust.
28415 (expand_builtin_memcmp): Adjust.
28416 (expand_builtin): Handle memcmp.
28417 (maybe_emit_chk_warning): Check strncat just once.
28419 2017-05-04 Martin Sebor <msebor@redhat.com>
28421 PR preprocessor/79214
28422 PR middle-end/79222
28423 PR middle-end/79223
28424 * builtins.c (check_sizes): Add inlining context and issue
28425 warnings even when -Wno-system-headers is set.
28426 (check_strncat_sizes): Same.
28427 (expand_builtin_strncat): Same.
28428 (expand_builtin_memmove): New function.
28429 (expand_builtin_stpncpy): Same.
28430 (expand_builtin): Handle memmove and stpncpy.
28432 2017-05-04 Bin Cheng <bin.cheng@arm.com>
28434 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
28435 which is not used any more.
28437 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28439 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
28441 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28443 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
28444 (cortexa53_tunings): Likewise.
28445 (cortexa57_tunings): Likewise.
28446 (cortexa72_tunings): Likewise.
28447 (cortexa73_tunings): Likewise.
28449 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28451 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
28452 Set loop alignment to 8.
28454 2017-05-04 Martin Sebor <msebor@redhat.com>
28456 PR translation/80280
28457 * builtins.c (expand_builtin_object_size): Add missing quoting to
28458 %D and like directives.
28459 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
28460 (hsa_type_for_tree_type): Same.
28461 (verify_function_arguments): Same.
28462 * symtab.c (symbol_table::change_decl_assembler_name): Same.
28463 * varasm.c (get_section): Same.
28466 2017-05-04 Martin Sebor <msebor@redhat.com>
28468 PR translation/80280
28469 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
28471 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28473 * config/aarch64/aarch64.c (generic_addrcost_table):
28474 Change HI/TI mode setting.
28476 2017-05-04 Martin Jambor <mjambor@suse.cz>
28478 PR tree-optimization/80622
28479 * tree-sra.c (comes_initialized_p): New function.
28480 (build_accesses_from_assign): Only set write lazily when
28481 comes_initialized_p is false.
28482 (analyze_access_subtree): Use comes_initialized_p.
28483 (propagate_subaccesses_across_link): Assert !comes_initialized_p
28484 instead of testing for PARM_DECL.
28486 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28488 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
28489 constraint on operand 0 to allow more general addressing modes.
28490 Adjust output template.
28491 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
28493 * config/aarch64/aarch64-protos.h
28494 (aarch64_address_valid_for_prefetch_p): Declare prototype.
28495 * config/aarch64/constraints.md (Dp): New address constraint.
28496 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
28499 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
28501 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
28502 update use of estimate_ipcp_clone_size_and_time.
28503 (estimate_local_effects): Update use of
28504 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
28505 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
28506 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
28507 Return nonspecialized time.
28509 2017-05-04 Richard Biener <rguenther@suse.de>
28511 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
28512 for the last VUSE which def dominates the PHI. Directly call
28514 (get_continuation_for_phi_1): Remove.
28516 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
28518 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
28519 to explain the use of truncating division. Cap the number of
28520 iterations to the maximum given by nb_iterations_upper_bound,
28523 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28525 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
28526 * configure: Regenerate.
28527 * config.in: Regenerate.
28528 * config/i386/driver-mingw32.c: new file.
28529 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
28530 * config.host: Link driver-mingw32.o on MinGW host.
28531 * doc/install.texi: Document new --enable-mingw-wildcard configure
28534 2017-05-04 Marek Polacek <polacek@redhat.com>
28536 PR tree-optimization/80612
28537 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
28539 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28540 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
28542 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
28543 (movt splitter): Likewise.
28544 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
28545 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
28546 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
28547 block for Thumb-1 with MOVT.
28548 (thumb2_legitimate_address_p): Move code block ...
28549 (can_avoid_literal_pool_for_label_p): ... into this new function.
28550 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
28552 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
28553 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
28554 "M-profile targets with the MOVT instruction".
28556 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28558 * config/arm/arm-builtins.c (arm_init_builtins): Rename
28559 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
28560 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
28562 2017-05-04 Martin Liska <mliska@suse.cz>
28564 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
28565 variable cond_code.
28567 2017-05-04 Richard Biener <rguenther@suse.de>
28569 * tree.c (array_at_struct_end_p): Handle arrays at struct
28570 end with flexarrays more conservatively. Refactor and treat
28571 arrays of arrays or aggregates more strict. Fix
28572 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
28573 * tree.c (array_at_struct_end_p): Adjust prototype.
28574 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
28575 * gimple-fold.c (get_range_strlen): Likewise.
28576 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
28578 2017-05-04 Richard Biener <rguenther@suse.de>
28580 PR tree-optimization/31130
28581 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
28583 (supports_overflow_infinity): Likewise.
28584 (is_negative_overflow_infinity): Likewise.
28585 (is_positive_overflow_infinity): Likewise.
28586 (is_overflow_infinity): Likewise.
28587 (stmt_overflow_infinity): Likewise.
28588 (overflow_infinity_range_p): Likewise.
28589 (usable_range_p): Remove as always returning true.
28590 (make_overflow_infinity): Remove.
28591 (negative_overflow_infinity): Likewise.
28592 (positive_overflow_infinity): Likewise.
28593 (avoid_overflow_infinity): Likewise.
28594 (set_value_range): Adjust accordingly.
28595 (set_value_range_to_nonnegative): Likewise, remove now unused
28596 overflow_infinity arg.
28597 (vrp_operand_equal_p): Adjust.
28598 (update_value_range): Likewise.
28599 (range_int_cst_singleton_p): Likewise.
28600 (operand_less_p): Likewise.
28601 (compare_values_warnv): Likewise.
28602 (extract_range_for_var_from_comparison_expr): Likewise.
28603 (vrp_int_const_binop): Likewise.
28604 (zero_nonzero_bits_from_vr): Likewise.
28605 (extract_range_from_multiplicative_op_1): Likewise.
28606 (extract_range_from_binary_expr_1): Likewise.
28607 (extract_range_from_unary_expr): Likewise.
28608 (extract_range_from_comparison): Likewise.
28609 (extract_range_basic): Likewise.
28610 (adjust_range_with_scev): Likewise.
28611 (compare_ranges): Likewise.
28612 (compare_range_with_value): Likewise.
28613 (dump_value_range): Likewise.
28614 (test_for_singularity): Likewise, remove strict_overflow_p parameter
28616 (simplify_cond_using_ranges): Adjust.
28618 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28620 * brig-builtins.def: Added a builtin for class_f64.
28621 * builtin-types.def: Added a builtin type needed by class_f64.
28623 2017-05-03 Jason Merrill <jason@redhat.com>
28625 * timevar.def: Add TV_CONSTEXPR.
28627 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28629 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
28631 2017-05-03 Martin Jambor <mjambor@suse.cz>
28633 * ipa-prop.c (ipa_update_after_lto_read): Removed.
28634 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
28635 * ipa-cp.c (ipcp_propagate_stage): Do not call
28636 ipa_update_after_lto_read.
28637 * ipa-inline.c (ipa_inline): Likewise.
28639 2017-05-03 Martin Jambor <mjambor@suse.cz>
28641 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
28642 tag. Added a default constructor and a destructor.
28643 (ipa_edge_args_sum_t): New class;
28644 (ipa_edge_args_sum): Declare.
28645 (ipa_edge_args_vector): Remove declaration.
28646 (IPA_EDGE_REF): Use ipa_edge_args_sum.
28647 (ipa_free_edge_args_substructures): Remove declaration.
28648 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
28649 (ipa_edge_args_info_available_for_edge_p): Likewise.
28650 * ipa-prop.c (ipa_edge_args_vector): Removed.
28651 (edge_removal_hook_holder): Likewise.
28652 (edge_duplication_hook_holder): Likewise.
28653 (ipa_edge_args_sum): New variable.
28654 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
28655 ipa_edge_args_vector.
28656 (ipa_free_edge_args_substructures): Likewise.
28657 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
28658 ipa_edge_args_vector.
28659 (ipa_edge_removal_hook): Turned into method
28660 ipa_edge_args_sum_t::remove.
28661 (ipa_edge_duplication_hook): Turned into method
28662 ipa_edge_args_sum_t::duplicate.
28663 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
28664 registering edge hooks.
28665 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
28666 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
28667 ipa_edge_args_sum instead of ipa_edge_args_vector.
28668 * ipa-profile.c (ipa_profile): Likewise.
28670 2017-05-03 Martin Jambor <mjambor@suse.cz>
28672 * symbol-summary.h (function_summary): New method exists.
28673 (function_summary::symtab_removal): Deallocate through release.
28674 (call_summary): New class.
28675 (gt_ggc_mx): New overload.
28676 (gt_pch_nx): Likewise.
28677 (gt_pch_nx): Likewise.
28679 2017-05-03 Jeff Law <law@redhat.com>
28681 PR tree-optimization/78496
28682 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
28683 from simplify_cond_using_ranges. Split off code to walk
28684 backwards through casts into ...
28685 (simplify_cond_using_ranges_2): New function.
28686 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
28687 (execute_vrp): After identifying jump threads, call
28688 simplify_cond_using_ranges_2.
28690 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
28693 * ipa-inline.h (inline_summary): Add ctor.
28694 (create_ggc): Do not use ggc_cleared_alloc.
28696 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
28697 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28699 * gcc.c (handle_braces): Support escaping in switch matching
28701 * doc/invoke.texi (Spec Files): Document it.
28702 Remove superfluous @code markup in items.
28704 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28706 * diagnostic-show-locus.c (struct column_range): New struct.
28707 (get_affected_columns): New function.
28708 (get_printed_columns): New function.
28709 (struct correction): New struct.
28710 (correction::ensure_capacity): New function.
28711 (correction::ensure_terminated): New function.
28712 (struct line_corrections): New struct.
28713 (line_corrections::~line_corrections): New dtor.
28714 (line_corrections::add_hint): New function.
28715 (layout::print_trailing_fixits): Reimplement in terms of the new
28717 (selftest::test_overlapped_fixit_printing): New function.
28718 (selftest::diagnostic_show_locus_c_tests): Call it.
28720 2017-05-03 Nathan Sidwell <nathan@acm.org>
28722 Canonicalize canonical type hashing
28723 * tree.h (type_hash_canon_hash): Declare.
28724 * tree.c (type_hash_list, attribute_hash_list): Move into
28725 type_hash_canon_hash.
28726 (build_type_attribute_qual_variant): Break out hash code calc into
28727 type_hash_canon_hash.
28728 (type_hash_canon_hash): New. Generic type hash computation.
28729 (build_range_type_1, build_array_type_1, build_function_type,
28730 build_method_type_directly, build_offset_type, build_complex_type,
28731 make_vector_type): Call it.
28733 2017-05-03 Richard Biener <rguenther@suse.de>
28735 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
28736 When all DRs have unknown misaligned do not always peel
28737 when there is a store but apply the same costing model as if
28738 there were only loads.
28740 2017-05-03 Richard Biener <rguenther@suse.de>
28743 PR tree-optimization/80492
28744 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28745 compare_base_decls returning dont-know properly.
28747 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
28749 * config/arm/iterators.md (CCSI): New mode iterator.
28750 (arch): New mode attribute.
28751 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
28752 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
28753 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
28754 code iterator for success result mode.
28755 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
28756 the corresponding new insn generators.
28758 2017-05-03 Bin Cheng <bin.cheng@arm.com>
28761 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28762 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28764 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
28766 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
28767 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
28768 (DDR_A): Wrap DDR argument in brackets.
28769 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
28770 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
28771 (DDR_REVERSED_P): Likewise.
28773 2017-05-03 Jakub Jelinek <jakub@redhat.com>
28775 PR tree-optimization/79472
28776 * tree-switch-conversion.c (struct switch_conv_info): Add
28777 contiguous_range and default_case_nonstandard fields.
28778 (collect_switch_conv_info): Compute contiguous_range and
28779 default_case_nonstandard fields, don't clear final_bb if
28780 contiguous_range and only the default case doesn't have the required
28782 (check_all_empty_except_final): Set default_case_nonstandard instead
28783 of failing if contiguous_range and the default case doesn't have empty
28785 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
28786 and only the default case doesn't have the required constants. Skip
28788 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
28789 if default_case_nonstandard.
28790 (build_constructors): Build constant 1 just once. Assert that default
28791 values aren't inserted in between cases if contiguous_range. Skip
28793 (build_arrays): Skip virtual phis.
28794 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
28795 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
28796 Handle virtual phis.
28797 (gen_inbound_check): Handle default_case_nonstandard case.
28798 (process_switch): Adjust check_final_bb caller. Call
28799 gather_default_values with the first non-default case instead of
28800 default case if default_case_nonstandard.
28802 2017-05-02 Nathan Sidwell <nathan@acm.org>
28804 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
28805 check. Fix formatting.
28807 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
28809 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
28810 errors when comparing specialized and unspecialized times.
28812 2017-05-02 David Malcolm <dmalcolm@redhat.com>
28814 * diagnostic-show-locus.c
28815 (layout::should_print_annotation_line_p): Make private.
28816 (layout::print_annotation_line): Make private.
28817 (layout::annotation_line_showed_range_p): Make private.
28818 (layout::show_ruler): Make private.
28819 (layout::print_source_line): Make private. Pass in line and
28820 line_width, rather than calling location_get_source_line. Drop
28822 (layout::print_leading_fixits): New method.
28823 (layout::print_any_fixits): Rename to...
28824 (layout::print_trailing_fixits): ...this, and make private.
28825 Don't print newline fixits.
28826 (diagnostic_show_locus): Move logic for printing one row into...
28827 (layout::print_line): ...this new function. Move the
28828 location_get_source_line call and error-handling from
28829 print_source_line to here. Call print_leading_fixits, and rename
28830 print_any_fixits to print_trailing_fixits.
28831 (selftest::test_fixit_insert_containing_newline): Update now that
28832 newlines are partially supported.
28833 (selftest::test_fixit_insert_containing_newline_2): New test.
28834 (selftest::test_fixit_replace_containing_newline): Update comments.
28835 (selftest::diagnostic_show_locus_c_tests): Call the new test.
28836 * edit-context.c (class added_line): New class.
28837 (class edited_line): Describe newline handling in comment.
28838 (edited_line::actually_edited_p): New method.
28839 (edited_line::print_content): Delete redundant decl.
28840 (edited_line::m_predecessors): New field.
28841 (edited_file::print_content): Call edited_line::print_content.
28842 (edited_file::print_diff): Update to support newlines.
28843 (edited_file::print_diff_hunk): Likewise.
28844 (edited_file::print_run_of_changed_lines): New function.
28845 (edited_file::print_diff_line): Convert to...
28846 (print_diff_line): ...this.
28847 (edited_file::get_effective_line_count): New function.
28848 (edited_line::edited_line): Initialize new field m_predecessors.
28849 (edited_line::~edited_line): Clean up m_predecessors.
28850 (edited_line::apply_fixit): Handle newlines.
28851 (edited_line::get_effective_line_count): New function.
28852 (edited_line::print_content): New function.
28853 (edited_line::print_diff_lines): New function.
28854 (selftest::test_applying_fixits_insert_containing_newline): New
28856 (selftest::test_applying_fixits_replace_containing_newline): New
28858 (selftest::insert_line): New function.
28859 (selftest::test_applying_fixits_multiple_lines): Add example of
28861 (selftest::edit_context_c_tests): Call the new tests.
28863 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28865 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
28866 parameter cand. Update dump information.
28867 (get_computation_cost): Update uses.
28869 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28871 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
28872 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
28873 (get_computation_at, rewrite_use_address): Update use of
28874 get_computation_aff.
28876 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28878 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
28879 (get_computation): Delete.
28880 (get_computation_cost): Implement like get_computation_cost_at.
28881 Use get_computation_at.
28882 (get_computation_cost_at): Delete.
28883 (rewrite_use_nonlinear_expr): Use get_computation_at.
28884 (rewrite_use_compare, remove_unused_ivs): Ditto.
28886 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28888 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
28890 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28892 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
28893 (ivopts_global_cost_for_size): Rename parameter and update uses.
28894 (iv_ca_recount_cost): Update uses.
28895 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
28896 candidates seperately in n_invs and n_cands.
28897 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
28899 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28901 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
28902 (find_inv_vars_cb): New.
28903 (find_depends): Renamed to ...
28904 (find_inv_vars): ... this.
28905 (add_candidate_1, force_var_cost): Call find_inv_vars.
28906 (split_address_cost, determine_group_iv_cost_cond): Ditto.
28908 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28910 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
28911 inv_vars. Add inv_exprs.
28912 (struct iv_cand): Rename depends_on to inv_vars.
28913 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
28914 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
28915 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
28916 (dump_cand): Dump inv_vars.
28917 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
28918 (record_invariant, find_depends, add_candidate_1): Ditto.
28919 (set_group_iv_cost, force_var_cost): Ditto.
28920 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
28921 (get_computation_cost_at, get_computation_cost): Ditto.
28922 (determine_group_iv_cost_generic): Ditto.
28923 (determine_group_iv_cost_address): Ditto.
28924 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
28925 (determine_group_iv_costs): Ditto.
28926 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
28927 (iv_ca_set_remove_invariants): Renamed to ...
28928 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
28929 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
28930 (iv_ca_set_add_invariants): Renamed to ...
28931 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
28932 (iv_ca_set_cp): Use iv_ca_set_add_invs.
28933 (iv_ca_has_deps): Support inv_vars and inv_exprs.
28934 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
28935 (create_new_ivs): Remove useless dump.
28937 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28939 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
28941 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
28942 (iv_ca_set_no_cp, create_new_iv): Ditto.
28944 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28946 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28948 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28950 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
28951 function tree_check2.
28953 2017-05-02 Martin Liska <mliska@suse.cz>
28955 * doc/gcov.texi: Add missing preposition.
28956 * gcov.c (function_info::function_info): Properly fill up
28957 all member variables.
28959 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28961 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
28963 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28965 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
28967 2017-05-02 Martin Liska <mliska@suse.cz>
28970 * lto-streamer-in.c (lto_read_tree_1): Remove
28971 LTO_STREAMER_DEBUG.
28972 * lto-streamer.c (struct tree_hash_entry): Likewise.
28973 (struct tree_entry_hasher): Likewise.
28974 (tree_entry_hasher::hash): Likewise.
28975 (tree_entry_hasher::equal): Likewise.
28976 (lto_streamer_init): Likewise.
28977 (lto_orig_address_map): Likewise.
28978 (lto_orig_address_get): Likewise.
28979 (lto_orig_address_remove): Likewise.
28980 * lto-streamer.h: Likewise.
28981 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
28982 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
28984 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
28986 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
28987 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
28988 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
28989 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
28990 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
28991 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
28992 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
28993 (mm_maskz_sub_ss): New intrinsics.
28994 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28995 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28996 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
28997 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
28998 (__builtin_ia32_subss_mask_round): New builtins.
28999 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29000 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29001 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
29003 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
29004 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29006 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29009 2017-05-02 Martin Jambor <mjambor@suse.cz>
29011 PR tree-optimization/78687
29012 * tree-sra.c (access): New field parent.
29013 (process_subtree_disqualification): New function.
29014 (disqualify_candidate): Call it.
29015 (build_accesses_from_assign): Reset write flag if creating an
29017 (build_access_subtree): Fill in parent field and also prpagate
29018 down grp_write flag.
29019 (create_artificial_child_access): New parameter set_grp_write, set
29020 grp_write to its value.
29021 (propagate_subaccesses_across_link): Also propagate grp_write flag
29023 (propagate_all_subaccesses): Push the closest parent back to work
29024 queue if add_access_to_work_queue returned true.
29026 2017-05-02 Richard Biener <rguenther@suse.de>
29028 * common.opt (fstrict-overflow): Alias negative to fwrapv.
29029 * doc/invoke.texi (fstrict-overflow): Remove all traces of
29030 -fstrict-overflow documentation.
29031 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
29032 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
29033 flag_strict_overflow.
29034 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
29035 * lto-opts.c (lto_write_options): Do not stream it.
29036 * lto-wrapper.c (merge_and_complain): Do not handle it.
29037 * opts.c (default_options_table): Do not set -fstrict-overflow.
29038 (finish_options): Likewise do not clear it when sanitizing.
29039 * simplify-rtx.c (simplify_const_relational_operation): Do not
29040 test flag_strict_overflow.
29042 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29044 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
29045 using enabled attribute.
29046 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
29047 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
29048 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
29049 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
29050 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
29051 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
29052 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
29053 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29054 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29055 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29057 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29059 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29061 2017-05-02 Richard Biener <rguenther@suse.de>
29063 PR tree-optimization/80591
29065 2017-04-10 Richard Biener <rguenther@suse.de>
29067 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29070 2017-05-02 Richard Biener <rguenther@suse.de>
29072 PR tree-optimization/80549
29073 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29074 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29075 headers if they do not have a preheader.
29077 2017-05-02 Martin Liska <mliska@suse.cz>
29080 * common.opt: Fix typo.
29081 * doc/invoke.texi: Likewise.
29083 2017-05-01 Jan Beulich <jbeulich@suse.com>
29085 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29086 swapping, add (x,x,m,x,n) alternative.
29088 2017-05-01 Nathan Sidwell <nathan@acm.org>
29090 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29091 unnecessary unadjusted_alignment check.
29093 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29096 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29098 * gimplify.c (gimplify_cilk_detach): New function.
29099 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29100 * tree-core.h: Document EXPR_CILK_SPAWN.
29101 * tree.h (EXPR_CILK_SPAWN): Define.
29103 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29105 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29106 to use new fixit_hint representation, using the "replace" logic.
29107 (get_line_span_for_fixit_hint): Likewise.
29108 (layout::print_any_fixits): Likewise.
29109 (selftest::test_one_liner_many_fixits): Rename to...
29110 (selftest::test_one_liner_many_fixits_1): ...this, and update
29111 comment and expected output to reflect that the multiple fix-it
29112 hints are now consolidated into one insertion.
29113 (selftest::test_one_liner_many_fixits_2): New test.
29114 (selftest::test_diagnostic_show_locus_one_liner): Update for
29116 (selftest::test_fixit_consolidation): Update for fix-it API
29118 * diagnostic.c (print_parseable_fixits): Likewise.
29119 * edit-context.c (edited_line::m_line_events): Convert from
29120 auto_vec <line_event *> to auto_vec <line_event>.
29121 (class line_event): Convert from abstract base class to a concrete
29122 class, taking over the role of replace_event.
29123 (class insert_event): Delete.
29124 (class replace_event): Rename to class line_event. Convert to
29126 (edit_context::add_fixits): Reimplement.
29127 (edit_context::apply_insert): Delete.
29128 (edit_context::apply_replace): Rename to...
29129 (edit_context::apply_fixit): ...this. Convert to half-open range.
29130 (edited_file::apply_insert): Delete.
29131 (edited_file::apply_replace): Rename to...
29132 (edited_file::apply_fixit): ...this.
29133 (edited_line::~edited_line): Drop deletion of events.
29134 (edited_line::apply_insert): Delete.
29135 (edited_line::apply_replace): Rename to...
29136 (edited_line::apply_fixit): ...this. Convert to half-open range.
29137 Update for change to type of m_line_events.
29138 * edit-context.h (edit_context::apply_insert): Delete.
29139 (edit_context::apply_replace): Rename to...
29140 (edit_context::apply_fixit): ...this.
29142 2017-05-01 Martin Sebor <msebor@redhat.com>
29144 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
29147 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
29150 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
29151 __get_cpuid_max returns 0.
29152 (__get_cpuid_count): Ditto.
29154 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
29156 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
29157 replacement expression is another instance of one of its arguments.
29159 2017-05-01 Jakub Jelinek <jakub@redhat.com>
29162 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
29163 check for stack push/pop autoinc.
29164 * config/i386/i386.c (ix86_agi_dependent): Return false
29165 if the only reason why modified_in_p returned true is that
29166 addr is SP based and set_insn is a push or pop.
29168 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29170 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
29173 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29176 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
29177 (account_size_time): Use two predicates - exec_pred and
29179 (evaluate_conditions_for_known_args): Compute both clause and
29181 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
29182 (inline_summary_t::duplicate): Update.
29183 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
29185 (compute_inline_parameters): Likewise.
29186 (estimate_edge_size_and_time): Update caluclation of time.
29187 (estimate_node_size_and_time): Compute both time and nonspecialized
29189 (estimate_ipcp_clone_size_and_time): Update.
29190 (inline_merge_summary): Update.
29191 (do_estimate_edge_time): Update.
29192 (do_estimate_edge_size): Update.
29193 (do_estimate_edge_hints): Update.
29194 (inline_read_section, inline_write_summary): Stream both new predicates.
29195 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
29197 (compute_inlined_call_time): Cleanup.
29198 (big_speedup_p): Update.
29199 (edge_badness): Update.
29200 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
29201 (size_time_entry): Replace predicate by exec_predicate and
29202 nonconst_predicate.
29203 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
29204 (estimate_edge_time): Return also nonspec_time.
29205 (reset_edge_growth_cache): Update.
29207 2017-04-29 Jakub Jelinek <jakub@redhat.com>
29209 PR rtl-optimization/80491
29210 * ifcvt.c (noce_process_if_block): When looking for x setter
29211 with missing else_bb, don't check only the insn right before
29212 cond_earliest, but look for the last insn that x is modified in
29213 within the same bb.
29215 PR rtl-optimization/80491
29216 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
29218 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
29220 PR tree-optimization/80487
29221 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
29223 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29225 PR tree-optimization/79697
29226 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
29227 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
29228 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
29230 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
29231 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
29233 2017-04-28 Martin Sebor <msebor@redhat.com>
29235 PR tree-optimization/80523
29236 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
29237 (init_target_to_host_charmap, target_to_host, target_strtol10): New
29239 (maybe_warn, format_directive, parse_directive): Use new functions.
29240 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
29242 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
29244 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
29246 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
29248 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
29249 target_header_dir): Set correctly.
29250 * configure: Regenerated.
29251 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29252 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29253 instead of SYSTEM_HEADER_DIR.
29255 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
29257 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
29258 (estimate_local_effects): Likewise.
29259 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
29260 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
29261 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
29262 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
29263 do_estimate_edge_time, estimate_edge_time): Likewise.
29264 * ipa-inline-analysis.c (estimate_node_size_and_time,
29265 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
29266 (estimate_time_after_inlining): Remove.
29268 2017-04-28 Martin Liska <mliska@suse.cz>
29270 * doc/gcov.texi: Enhance documentation of gcov.
29272 2017-04-28 Martin Liska <mliska@suse.cz>
29274 * doc/gcov.texi: Sort options in alphabetic order.
29275 * doc/gcov-dump.texi: Likewise.
29276 * doc/gcov-tool.texi: Likewise.
29277 * gcov.c (print_usage): Likewise.
29278 * gcov-dump.c (print_usage): Likewise.
29279 * gcov-tool.c (print_merge_usage_message): Likewise.
29280 (print_rewrite_usage_message): Likewise.
29281 (print_overlap_usage_message): Likewise.
29283 2017-04-28 Martin Liska <mliska@suse.cz>
29285 PR gcov-profile/53915
29286 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
29288 2017-04-28 Martin Liska <mliska@suse.cz>
29290 PR gcov-profile/79891
29291 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
29292 is marked by compiler as living on a line.
29293 (get_cycles_count): Remove usage of the union.
29294 (output_intermediate_file): Likewise.
29295 (find_source): Fix GNU coding style.
29296 (accumulate_line_counts): Remove old non-all block mode.
29297 (output_lines): Remove usage of the union.
29298 * profile.c (output_location): Include all BBs, even if
29299 belonging to a same line (and file) as a previous BB.
29301 2017-04-28 Martin Liska <mliska@suse.cz>
29303 * gcov.c (process_args): Handle new argument 'w'.
29304 (read_graph_file): Assign ID to BBs.
29305 (output_branch_count): Display BB # if verbose flag is set.
29306 (output_lines): Likewise for arcs.
29307 (print_usage): Add '--verbose' option help.
29308 * doc/gcov.texi: Document --verbose (-w) option.
29310 2017-04-28 Martin Liska <mliska@suse.cz>
29312 * gcov.c (struct block_location_info): New struct.
29313 (process_file): Fill up the new structure.
29314 (read_graph_file): Replace usage of encoding by the newly added
29316 (add_line_counts): Likewise.
29317 (accumulate_line_counts): Remove usage of the union.
29318 (function_info::function_info): New function.
29319 (function_info::~function_info): Likewise.
29320 (process_file): Call delete instead of release_function.
29321 (release_function): Release the function.
29322 (release_structures): Call delete instead of release_function.
29323 (solve_flow_graph): Replace usage of num_blocks.
29324 (find_exception_blocks): Likewise.
29325 (output_lines): Fix GNU coding style.
29327 2017-04-28 Martin Liska <mliska@suse.cz>
29330 * coverage.c (coverage_remove_note_file): New function.
29331 * coverage.h: Declare the function.
29332 * toplev.c (finalize): Clean if an error has been seen.
29334 2017-04-28 Martin Liska <mliska@suse.cz>
29336 PR gcov-profile/80031
29337 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
29338 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
29339 * gcov.c (read_graph_file): Read just number of blocks.
29340 * profile.c (branch_prob): Do not stream 0 flags per a basic
29343 2017-04-28 Martin Liska <mliska@suse.cz>
29345 * gcov-dump.c (tag_*): Add new argument to declarations.
29346 (dump_gcov_file): Likewise.
29347 (tag_blocks): Add and use new argument depth.
29348 (tag_arcs): Likewise.
29349 (tag_lines): Likewise.
29350 (tag_counters): Likewise.
29351 (tag_summary): Likewise.
29352 (dump_working_sets): Use depth to do a proper indentation.
29354 2017-04-28 Jakub Jelinek <jakub@redhat.com>
29357 * cgraph.h (symtab_node::debug_symtab): No longer inline.
29358 * symtab.c (symtab_node::debug_symtab): Move definition here.
29360 2017-04-28 Richard Biener <rguenther@suse.de>
29362 * lto-streamer.h (LTO_major_version): Bump to 7.
29364 2017-04-28 Richard Biener <rguenther@suse.de>
29366 * tree-vrp.c (assert_info): New struct.
29367 (add_assert_info): New helper.
29368 (register_edge_assert_for_2): Refactor to add asserts to a vector
29370 (register_edge_assert_for_1): Likewise.
29371 (register_edge_assert_for): Likewise.
29372 (finish_register_edge_assert_for): New helper actually registering
29373 asserts where live on edge.
29374 (find_conditional_asserts): Adjust.
29375 (find_switch_asserts): Likewise.
29376 (evrp_dom_walker::try_find_new_range): Generalize.
29377 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
29379 2017-04-27 Marek Polacek <polacek@redhat.com>
29382 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
29383 arg10 and arg11 to itype.
29385 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
29387 * doc/extend.texi (Object Size Checking): Improve grammar.
29389 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
29392 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
29393 that the logic for permitting reciprocal estimates matches that
29396 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29399 * tree.c (type_cache_hasher::equal): Only compare
29400 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
29401 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
29402 non-aggregate element types.
29403 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
29404 about the flag on ARRAY_TYPEs in the comment, formatting fix.
29406 2017-04-27 Richard Biener <rguenther@suse.de>
29408 PR middle-end/80533
29409 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
29410 stripping ARRAY_REFs from MEM_EXPR make sure we're not
29411 keeping a reference to a trailing array.
29413 2017-04-27 Richard Biener <rguenther@suse.de>
29415 PR middle-end/80539
29416 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
29417 being in loop-closed SSA form conservatively.
29418 (chrec_fold_multiply_poly_poly): Likewise.
29420 2017-04-27 Tamar Christina <tamar.christina@arm.com>
29422 PR middle-end/79665
29423 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
29424 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
29426 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29429 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
29430 (aarch64_function_arg_alignment): Return unsigned int again, but still
29431 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
29432 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29433 Don't emit -Wpsabi note.
29434 (aarch64_function_arg_boundary): Likewise.
29435 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29438 2017-04-26 Nathan Sidwell <nathan@acm.org>
29440 * tree.h (crc32_unsigned_n): Declare.
29441 (crc32_unsigned, crc32_unsigned): Make inline.
29442 * tree.c (crc32_unsigned_bits): Replace with ...
29443 (crc32_unsigned_n): ... this.
29444 (crc32_unsigned, crc32_byte): Remove.
29445 (crc32_string): Remove unnecessary braces.
29447 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29449 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
29450 * ipa-inline-analysis.c (MAX_TIME): Remove.
29451 (account_size_time): Use sreal for time.
29452 (dump_inline_summary): Update.
29453 (estimate_function_body_sizes): Update.
29454 (estimate_edge_size_and_time): Update.
29455 (estimate_calls_size_and_time): Update.
29456 (estimate_node_size_and_time): Update.
29457 (inline_merge_summary): Update.
29458 (inline_update_overall_summary): Update.
29459 (estimate_time_after_inlining): Update.
29460 (inline_read_section): Update.
29461 (inline_write_summary): Update.
29462 * ipa-inline.c (compute_uninlined_call_time): Update.
29463 (compute_inlined_call_time): Update.
29464 (recursive_inlining): Update.
29465 (inline_small_functions): Update.
29466 (dump_overall_stats): Update.
29467 * ipa-inline.h: Include sreal.h.
29468 (size_time_entry): Turn time to sreal.
29469 (inline_summary): Turn self_time nad time to sreal.
29471 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29473 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
29475 (sreal::stream_out, sreal::stream_in): New.
29476 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
29478 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29480 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
29483 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
29486 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
29488 (dimode_scalar_chain::compute_convert_gain): Ditto.
29489 (dimode_scalar_chain::make_vector_copies): Ditto.
29490 (dimode_scalar_chain::convert_reg): Ditto.
29491 (dimode_scalar_chain::convert_insn): Ditto.
29492 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
29493 (VI248_AVX512BW_1): New mode iterator.
29494 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
29495 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
29498 2017-04-25 Martin Sebor <msebor@redhat.com>
29500 PR tree-optimization/80497
29501 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
29502 constants are representable in HOST_WIDE_INT.
29503 (parse_directive): Ditto.
29505 2017-04-25 Martin Sebor <msebor@redhat.com>
29508 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
29509 (new_zero_array): Adjust signature.
29510 (dom_info::dom_init): Used unsigned rather that size_t.
29511 (dom_info::dom_info): Same.
29513 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29514 Jakub Jelinek <jakub@redhat.com>
29517 * config/arm/arm.c: Include gimple.h.
29518 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29519 returns negative, increment ncrn only if it returned positive.
29520 (arm_needs_doubleword_align): Return int instead of bool,
29521 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
29522 members, but if there is any such non-FIELD_DECL
29523 > PARM_BOUNDARY aligned decl, return -1 instead of false.
29524 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29525 returns negative, increment nregs only if it returned positive.
29526 (arm_setup_incoming_varargs): Likewise.
29527 (arm_function_arg_boundary): Emit -Wpsabi note if
29528 arm_needs_doubleword_align returns negative, return
29529 DOUBLEWORD_ALIGNMENT only if it returned positive.
29531 2017-04-25 Marek Polacek <polacek@redhat.com>
29534 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29535 first argument to type.
29537 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
29540 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
29541 type checks to test for compatibility instead of equality.
29543 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29544 Jakub Jelinek <jakub@redhat.com>
29547 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
29549 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
29550 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
29551 the alignment computation, but return their maximum in warn_alignment.
29552 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29553 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
29555 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
29556 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29559 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29561 * config/arc/simdext.md (dmpyh): Fix typo.
29563 2017-04-25 Richard Biener <rguenther@suse.de>
29565 PR tree-optimization/80492
29566 * alias.c (compare_base_decls): Handle registers with asm
29567 specification conservatively.
29568 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29569 compare_base_decls returning dont-know properly.
29571 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29573 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
29574 (legitimate_offset_address_p): New function.
29575 (arc_legitimate_address_p): Use above function.
29577 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29579 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
29581 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29583 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
29584 ACCH registers whenever they are available.
29586 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29588 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
29589 double regs fix when not used.
29591 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29593 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
29595 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
29596 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
29598 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29600 * config/arc/arc.c (arc_output_addsi): Check for h-register class
29601 when emitting short ADD instructions.
29603 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29605 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
29607 (cmpsi_cc_c_insn): Likewise.
29608 (cbranchsi4_scratch): Compute proper instruction length using
29609 compact_hreg_operand.
29610 * config/arc/predicates.md (compact_hreg_operand): New predicate.
29612 2017-04-25 Richard Biener <rguenther@suse.de>
29614 PR middle-end/80509
29615 * passes.c (pass_manager::pass_manager): Initialize
29616 m_name_to_pass_map.
29618 2017-04-25 Richard Biener <rguenther@suse.de>
29620 PR tree-optimization/79201
29621 * tree-ssa-sink.c (statement_sink_location): Handle calls.
29623 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29626 * config/s390/vector.md: Split MEM->GPR vector moves for
29627 non-s_operand addresses.
29629 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29632 * config/s390/predicates.md (reload_const_wide_int_operand): New
29634 * config/s390/s390.md ("movti"): Remove d/P alternative.
29635 ("movti_bigconst"): New pattern definition.
29637 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29640 * s390-protos.h (s390_expand_cs_hqi): Removed.
29641 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
29642 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
29643 modes as well as CCZ1mode and CCZmode.
29644 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
29645 signature of s390_emit_compare_and_swap.
29646 (s390_expand_cs_hqi): Likewise, make static.
29647 (s390_expand_cs_tdsi): Generate an explicit compare before trying
29648 compare-and-swap, in some cases.
29649 (s390_expand_cs): Wrapper function.
29650 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
29652 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
29653 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
29654 patterns for small and large integers. Forbid symref memory operands.
29655 Move expander to s390.c. Require cc register.
29656 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
29657 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
29658 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
29659 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
29660 symref memory operands. Remove CC mode and call s390_match_ccmode
29662 ("atomic_exchange<mode>"): Allow and implement all integer modes.
29664 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29666 * config/s390/s390.md (define_peephole2): New peephole to help
29667 combining the load-and-test pattern with volatile memory.
29669 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29671 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
29672 with CCZmode for TARGET_Z196.
29674 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29676 PR rtl-optimization/80501
29677 * combine.c (make_compound_operation_int): Set subreg_code to SET
29678 even for AND with mask of the sign bit of mode.
29680 PR rtl-optimization/80500
29681 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
29682 sum's initial value.
29684 2017-04-25 Julian Brown <julian@codesourcery.com>
29685 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29687 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
29689 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
29691 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
29693 2017-04-25 Julian Brown <julian@codesourcery.com>
29694 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29696 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
29697 (thunderx2t99_sha): New Reservation.
29699 2017-04-25 Julian Brown <julian@codesourcery.com>
29700 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29702 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
29703 type for 1-element load.
29705 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
29707 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
29709 2017-04-24 Martin Jambor <mjambor@suse.cz>
29711 PR tree-optimization/80293
29712 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
29713 char arrays not totally scalarizable if it is false.
29714 (analyze_all_variable_accesses): Pass correct value in the new
29715 parameter. Add a statistics counter.
29717 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
29719 PR middle-end/79931
29720 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
29722 2017-04-24 Richard Biener <rguenther@suse.de>
29724 PR tree-optimization/80494
29725 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
29726 out for complex types.
29728 2017-04-24 Richard Biener <rguenther@suse.de>
29730 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
29731 * tree-ssa-sccvn.c (print_scc): Print SCC size.
29732 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
29733 (DFS): Adjust and never fail.
29734 (sccvn_dom_walker::fail): Remove.
29735 (sccvn_dom_walker::before_dom_children): Adjust.
29736 (run_scc_vn): Likewise and never fail.
29737 * tree-ssa-pre.c (pass_pre::execute): Adjust.
29738 (pass_fre::execute): Likewise.
29740 2017-04-24 Richard Biener <rguenther@suse.de>
29742 PR tree-optimization/79725
29743 * tree-ssa-sink.c (statement_sink_location): Return whether
29744 failure reason was zero uses. Move that check later.
29745 (sink_code_in_bb): Deal with zero uses by removing the stmt
29748 2017-04-24 Richard Biener <rguenther@suse.de>
29751 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
29752 pointer-based references.
29754 2017-04-24 Richard Biener <rguenther@suse.de>
29757 * pass_manager.h (pass_manager::operator new): Remove.
29758 (pass_manager::operator delete): Likewise.
29759 * passes.c (pass_manager::operator new): Remove.
29760 (pass_manager::operator delete): Likewise.
29761 (pass_manager::pass_manager): Zero individual pass members.
29763 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
29766 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
29767 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
29768 Check "XEXP (src, 1)" operand here.
29769 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
29770 Check "XEXP (src, 1)" operand here.
29771 (dimode_scalar_chain::make_vector_copies): Detect count register
29772 of a shift instruction. Zero extend count register from QImode
29773 to DImode to satisfy vector shift pattern count operand predicate.
29774 Substitute vector shift count operand with a DImode copy.
29775 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
29778 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
29780 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
29781 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29782 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
29783 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29784 (UNSPEC_NOREX_MEM): Remove definition.
29786 2017-04-21 Richard Biener <rguenther@suse.de>
29788 PR tree-optimization/79547
29789 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
29790 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
29791 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
29792 without any constraints.
29794 2017-04-21 Richard Biener <rguenther@suse.de>
29796 PR tree-optimization/78847
29797 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
29799 2017-04-21 Richard Biener <rguenther@suse.de>
29801 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
29802 (build_distinct_type_copy): Likewise.
29803 (build_variant_type_copy): Likewise.
29804 * tree.c (build_qualified_type): Pass down mem-stat info.
29805 (build_distinct_type_copy): Likewise.
29806 (build_variant_type_copy): Likewise.
29808 2017-04-21 Richard Biener <rguenther@suse.de>
29810 PR tree-optimization/80237
29811 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
29813 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
29814 for a simplified result.
29816 2016-04-21 Richard Biener <rguenther@suse.de>
29818 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
29819 sth as strict as a simple_iv but a chrec without symbols and an
29820 operand defined in the loop we are peeling (and not some subloop).
29821 (propagate_constants_for_unrolling): Propagate all constants.
29823 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29826 * config/i386/i386.c (print_reg): Remove assert for disalowed
29827 regno values, call output_operand_lossage instead.
29829 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29832 * config/i386/constraints.md (Yc): New register constraint.
29833 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
29834 Use Yc constraint for alternative 2 of operand 0. Remove
29835 preferred_for_speed attribute.
29837 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29839 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
29840 lastprivate clauses in SIMT case.
29842 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
29844 * doc/invoke.texi (-Wextra-semi): Document new warning option.
29846 2017-04-20 Richard Biener <rguenther@suse.de>
29848 PR tree-optimization/57796
29849 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
29850 as N scalar stores.
29851 (vect_model_load_cost): Cost gathers as N scalar loads.
29853 2017-04-20 Richard Biener <rguenther@suse.de>
29855 * ggc-page.c (ggc_allocated_p): Rename to ...
29856 (safe_lookup_page_table_entry): ... this and return the lookup
29858 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
29860 2017-04-20 Richard Biener <rguenther@suse.de>
29862 PR tree-optimization/80453
29863 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
29864 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
29865 from the conditions.
29866 (vn_phi_eq): Pass them down.
29867 (vn_phi_lookup): Record them.
29868 (vn_phi_insert): Likewise.
29870 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
29872 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
29873 uninitialized variable warning to avoid buffer overrun.
29875 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29878 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
29879 is suppressed for '{ 0 }' in C.
29881 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29883 * BASE-VER: Set to 8.0.0.
29885 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
29887 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
29888 priority .init_array and .fini_array section with SECTION_NOTYPE
29891 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29893 PR middle-end/80423
29894 * tree.h (build_array_type): Add typeless_storage default argument.
29895 * tree.c (type_cache_hasher::equal): Also compare
29896 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
29897 (build_array_type): Add typeless_storage argument, set
29898 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
29900 (build_nonshared_array_type): Adjust build_array_type_1 caller.
29901 (build_array_type): Likewise. Add typeless_storage argument.
29903 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29904 Jakub Jelinek <jakub@redhat.com>
29906 PR tree-optimization/80426
29907 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
29908 operation on symbolic operands, also compute the overflow for the
29909 invariant part when the operation degenerates into a negation.
29911 2017-04-19 Jakub Jelinek <jakub@redhat.com>
29914 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
29915 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
29918 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
29920 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
29923 * config/avr/avr.c (tree.h): Include it.
29924 (cgraph.h): Include it.
29925 (avr_encode_section_info): Don't warn for uninitialized progmem
29926 variable if it's just an alias.
29928 2017-04-19 Richard Biener <rguenther@suse.de>
29931 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
29932 when needed by AutoPGO.
29934 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
29937 * doc/lto.texi: Remove an extra 'that'.
29939 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
29941 PR rtl-optimization/80429
29942 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
29943 are only used in debug insns.
29945 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29946 Vladimir Makarov <vmakarov@redhat.com>
29948 * config/sparc/predicates.md (input_operand): Add comment. Return
29949 true for any memory operand when LRA is in progress.
29950 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
29952 2017-04-18 Jeff Law <law@redhat.com>
29955 * mips.md ({return,simple_return}_internal): Do not overwrite
29958 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29960 PR tree-optimization/80443
29961 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
29962 instead of adding 1, subtract -1 and similarly instead of subtracting
29965 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
29967 PR rtl-optimization/80357
29968 * haifa-sched.c (tmp_bitmap): New variable.
29969 (model_recompute): Handle duplicate use records.
29970 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
29971 (free_global_sched_pressure_data): Free it.
29973 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
29976 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29977 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29978 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29979 instead of SYSTEM_HEADER_DIR.
29981 2017-04-18 Jeff Law <law@redhat.com>
29983 PR middle-end/80422
29984 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
29985 predecessors after walking up the insn chain.
29987 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29990 * dwarf2out.c (modified_type_die): Try harder not to emit internal
29991 sizetype type into debug info.
29993 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29996 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
29997 unneeded test for TARGET_UPPER_REGS_SF.
29998 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
30000 2017-04-18 Jakub Jelinek <jakub@redhat.com>
30003 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
30004 instead of gsi_after_labels.
30006 2017-04-18 Jeff Law <law@redhat.com>
30008 * regcprop.c (maybe_mode_change): Avoid creating copies of the
30012 2017-04-13 Jeff Law <law@redhat.com>
30013 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30014 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30016 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
30019 * config/avr/avr.c (intl.h): Include it.
30020 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
30022 2017-04-18 Martin Liska <mliska@suse.cz>
30024 PR gcov-profile/78783
30025 * gcov-tool.c (gcov_output_files): Validate that destination
30026 file is either removed by the tool or by a user.
30028 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30029 Guy Benyei <guybe@mellanox.com>
30031 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
30032 block, and do not negate it, the stored id is already negative.
30034 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30036 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
30038 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30041 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
30042 masks of options that should be turned off if the VSX vector
30043 options are turned off.
30044 (OTHER_P8_VECTOR_MASKS): Likewise.
30045 (OTHER_VSX_VECTOR_MASKS): Likewise.
30046 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
30047 rs6000_disable_incompatible_switches to validate no type switches
30049 (rs6000_incompatible_switch): New function to disallow turning on
30050 other vector options if -mno-vsx, -mno-power8-vector, or
30051 -mno-power9-vector are specified.
30053 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30055 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30057 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30059 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30060 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30061 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30062 (ARG_POINTER_CFA_OFFSET): Likewise.
30064 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30066 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30067 conditions to take advantage of various optimizations.
30069 2017-04-13 Jeff Law <law@redhat.com>
30071 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30072 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30073 (zero_extendsidi2_dext): Likewise.
30075 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30078 * fold-const.c (fold_ternary_loc): Revert
30079 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30082 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30084 PR rtl-optimization/80343
30085 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30086 new scratch pseudo.
30088 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30091 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30092 to ubsan_encode_value.
30094 2017-04-13 Jeff Law <law@redhat.com>
30096 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30097 appearing in DEBUG_INSNs.
30099 2017-04-13 Martin Liska <mliska@suse.cz>
30101 PR gcov-profile/80413
30102 * gcov-io.c (gcov_write_string): Copy to buffer just when
30103 allocated size is greater than zero.
30105 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30108 * dwarf2out.c (decls_for_scope): Ignore declarations of
30109 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30111 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
30114 * ipa-visibility.c (non_local_p): Fix typos.
30115 (localize_node): When localizing symbol in same comdat group,
30116 dissolve the group only when we know external symbols are going
30118 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
30120 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30122 PR tree-optimization/79390
30123 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
30124 order does not result in usable sequence, retry with reversed operand
30130 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
30131 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
30132 op0 instead of fold_convert_loc (loc, type, arg0).
30134 2017-04-12 Jeff Law <law@redhat.com>
30136 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
30137 has a delay slot in the generated code.
30139 * config/cris/cris.md (cris_preferred_reload_class): Return
30140 GENNONACR_REGS rather than GENERAL_REGS.
30142 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30145 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
30146 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
30147 signedness of the result type.
30149 2017-04-12 Richard Biener <rguenther@suse.de>
30150 Jeff Law <law@redhat.com>
30152 PR tree-optimization/80359
30153 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
30154 trim stores to TARGET_MEM_REFs.
30156 2017-04-12 Richard Biener <rguenther@suse.de>
30158 PR tree-optimization/79390
30159 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
30160 threading case even more.
30162 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
30165 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
30166 for quad_address_p for TImode, instead of just not indexed_address.
30168 2017-04-12 Richard Biener <rguenther@suse.de>
30169 Bernd Edlinger <bernd.edlinger@hotmail.de>
30171 PR middle-end/79671
30172 * alias.c (component_uses_parent_alias_set_from): Handle
30173 TYPE_TYPELESS_STORAGE.
30174 (get_alias_set): Likewise.
30175 * tree-core.h (tree_type_common): Add typeless_storage flag.
30176 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
30177 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
30178 for types containing members with TYPE_TYPELESS_STORAGE.
30179 (place_field): Likewise.
30180 (layout_type): Likewise for ARRAY_TYPE.
30181 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
30182 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
30183 TYPE_TYPELESS_STORAGE.
30184 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
30186 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30189 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30190 first argument to type.
30192 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30196 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
30197 CONST0_RTX (mode) rather than const0_rtx where appropriate.
30198 (rs6000_expand_binop_builtin): Likewise.
30199 (rs6000_expand_ternop_builtin): Likewise; also add missing
30200 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
30201 vshasigma built-ins.
30202 * doc/extend.texi: Document that vec_xxpermdi's third argument
30203 must be a constant.
30205 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
30207 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
30208 Use shift_const cost parameter when calculating gain of STV shifts.
30210 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
30212 PR rtl-optimization/70478
30213 * lra-constraints.c (process_alt_operands): Check memory for
30214 disfavoring memory insn operand.
30216 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30218 PR middle-end/80100
30219 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
30220 left shift in unsigned HOST_WIDE_INT type.
30222 PR rtl-optimization/80385
30223 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
30224 (not (neg X)) into (plus X -1) for complex or non-integral modes.
30227 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
30228 if they have any depend clauses.
30230 2017-04-11 Martin Liska <mliska@suse.cz>
30233 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
30234 * ipa-split.c (split_function): Create a local comdat symbol
30235 if caller is in a comdat group.
30237 2017-04-11 Martin Liska <mliska@suse.cz>
30240 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
30243 2017-04-11 Martin Sebor <msebor@redhat.com>
30245 PR middle-end/80364
30246 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
30247 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
30249 (directive::set_width, directive::set_precision, format_character):
30251 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
30254 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
30257 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
30258 conflict, set target->arch_name instead of target->cpu_name.
30260 2017-04-11 Richard Biener <rguenther@suse.de>
30262 PR tree-optimization/80374
30263 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
30264 build_zero_cst, remove fold_convertible_p check again.
30266 2017-04-11 Martin Liska <mliska@suse.cz>
30269 * ubsan.c (instrument_object_size): Do not instrument register
30272 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30275 * config/i386/i386-builtin-types.def
30276 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
30277 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
30278 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
30279 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
30280 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
30281 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
30282 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
30283 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
30284 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
30285 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
30286 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
30287 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
30288 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
30289 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
30290 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
30291 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
30292 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
30293 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
30294 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
30295 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
30296 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
30297 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
30298 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
30299 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
30300 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
30301 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
30302 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
30303 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
30304 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
30305 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
30306 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
30307 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
30308 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
30309 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
30310 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
30311 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
30312 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
30313 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
30314 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
30315 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
30316 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
30317 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
30318 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
30319 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
30320 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
30321 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
30323 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
30324 flag to second_arg_count, handle 4 argument function type _COUNT
30325 aliases, handle second_arg_count on second argument rather than last.
30327 2017-04-10 Jeff Law <law@redhat.com>
30329 PR tree-optimization/80374
30330 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
30331 record anything if we can not convert integer_zero_node to the
30334 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30337 * config/rs6000/rs6000.c (rs6000_option_override_internal):
30338 Enhance special handling given to the TARGET_P9_MINMAX option in
30339 relation to certain other options.
30341 2017-04-10 Bin Cheng <bin.cheng@arm.com>
30343 PR tree-optimization/80153
30344 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
30345 remove POINTER_PLUS_EXPR's base part directly, rather than through
30348 2017-04-10 Richard Biener <rguenther@suse.de>
30349 Bin Cheng <bin.cheng@arm.com>
30351 PR tree-optimization/80153
30352 * tree-affine.c (aff_combination_to_tree): Get base pointer from
30353 the first element of pointer type aff_tree. Build result expr in
30355 (add_elt_to_tree): Convert to type unconditionally. Remove other
30356 fold_convert calls.
30357 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
30358 (rewrite_use_nonlinear_expr): Check invariant using iv information.
30360 2017-04-10 Richard Biener <rguenther@suse.de>
30362 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30365 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
30367 PR rtl-optimization/70478
30368 * lra-constraints.c (curr_small_class_check): New.
30369 (update_and_check_small_class_inputs): New.
30370 (process_alt_operands): Update curr_small_class_check. Disfavor
30371 alternative insn memory operands. Check available regs for small
30374 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30377 * config/mips/mips.opt (-mvirt): Update description.
30378 * doc/invoke.texi (-mvirt): Likewise.
30380 2017-04-10 Richard Biener <rguenther@suse.de>
30382 PR middle-end/80362
30383 * fold-const.c (fold_binary_loc): Look at unstripped ops when
30384 looking for NEGATE_EXPR in -A / -B to A / B folding.
30386 2017-04-10 Martin Liska <mliska@suse.cz>
30388 PR gcov-profile/80224
30389 * gcov.c (print_usage): Fix usage string.
30390 (get_gcov_intermediate_filename): Remove.
30391 (output_gcov_file): Use both for normal and intermediate format.
30392 (generate_results): Do not initialize special file for
30393 intermediate format.
30395 2017-04-10 Richard Biener <rguenther@suse.de>
30397 PR tree-optimization/80304
30398 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
30401 2017-04-10 Nathan Sidwell <nathan@acm.org>
30404 * config/rs6000/rs6000.c (rs6000_vector_type): New.
30405 (rs6000_init_builtins): Use it.
30407 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30409 * config/arm/arm.md (<mrc>): Add mode to SET source.
30410 (<mrrc>): Likewise.
30412 2017-04-10 Richard Biener <rguenther@suse.de>
30414 PR middle-end/80344
30415 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
30417 2017-04-10 Jakub Jelinek <jakub@redhat.com>
30420 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
30421 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
30422 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
30423 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
30424 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
30425 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
30426 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
30427 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
30428 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
30429 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
30430 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
30431 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
30432 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
30433 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
30434 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
30435 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
30436 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
30437 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
30438 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
30439 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
30440 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
30441 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
30442 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
30444 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
30446 PR rtl-optimization/70478
30447 * lra-constraints.c: Reverse the last patch.
30449 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
30451 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
30452 Add comment for WCHAR_T.
30454 2017-04-08 Martin Liska <mliska@suse.cz>
30457 2017-04-07 Martin Liska <mliska@suse.cz>
30460 * ipa-split.c (split_function): Add function part to a same comdat
30463 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30466 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
30468 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
30470 * rs6000/rs6000.c (vec_load_pendulum): Rename...
30471 (vec_pairing): ...to this.
30472 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
30473 (rs6000_sched_init): Adjust for name change.
30474 (struct rs6000_sched_context): Likewise.
30475 (rs6000_init_sched_context): Likewise.
30476 (rs6000_set_sched_context): Likewise.
30478 2017-04-07 Jakub Jelinek <jakub@redhat.com>
30484 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
30486 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
30487 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
30488 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
30490 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
30492 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
30494 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30496 PR rtl-optimization/70703
30497 * ira-color.c (update_conflict_hard_regno_costs): Use
30498 int64_t instead of HOST_WIDE_INT.
30500 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30502 PR rtl-optimization/70478
30503 * lra-constraints.c (process_alt_operands): Disfavor alternative
30504 insn memory operands.
30506 2017-04-07 Jeff Law <law@redhat.com>
30508 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
30509 CALL and NOTE_INSN_CALL_ARG_LOCATION.
30511 2017-04-07 Martin Liska <mliska@suse.cz>
30514 * config/aarch64/aarch64.c (aarch64_process_target_attr):
30515 Show error message instead of an ICE.
30517 2017-04-07 Martin Liska <mliska@suse.cz>
30520 * ipa-split.c (split_function): Add function part to a same comdat
30523 2017-04-07 Richard Biener <rguenther@suse.de>
30525 PR middle-end/80341
30526 * tree.c (get_unwidened): Also handle ! for_type case for
30528 * convert.c (do_narrow): Split out from ...
30529 (convert_to_integer_1): ... here. Do not pass final truncation
30530 type to get_unwidened for TRUNC_DIV_EXPR.
30532 2017-04-07 Richard Biener <rguenther@suse.de>
30534 * tree-affine.c (wide_int_ext_for_comb): Take type rather
30536 (aff_combination_const): Adjust.
30537 (aff_combination_scale): Likewise.
30538 (aff_combination_add_elt): Likewise.
30539 (aff_combination_add_cst): Likewise.
30540 (aff_combination_convert): Likewise.
30541 (add_elt_to_tree): Likewise. Remove unused argument.
30542 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
30544 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
30546 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
30548 * config/arm/arm.c (arm_default_short_enums): Use
30549 ARM_DEFAULT_SHORT_ENUMS.
30550 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
30552 2017-04-06 Jakub Jelinek <jakub@redhat.com>
30555 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
30556 members with redundant out-of-class redeclaration.
30558 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30561 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
30562 * config/i386/i386.md (*zero_extendsidi2):
30563 Add (?*x,*x) and (?*v,*v) alternatives.
30565 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30568 * config/i386/i386.c (ix86_expand_builtin)
30569 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
30570 mode from insn data. Convert operands to insn operand mode.
30571 Copy operands that don't satisfy insn predicate to a register.
30573 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
30575 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
30578 2017-04-06 Richard Biener <rguenther@suse.de>
30580 PR tree-optimization/80334
30581 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
30582 preserve alignment of accesses.
30584 2017-04-06 Richard Biener <rguenther@suse.de>
30586 PR tree-optimization/80262
30587 * tree-sra.c (build_ref_for_offset): Preserve address-space
30589 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
30590 Drop useless address-space information on MEM_REF offsets.
30592 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
30594 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
30596 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30598 PR rtl-optimization/70703
30599 * ira-color.c (update_conflict_hard_regno_costs): Use
30600 HOST_WIDE_INT instead of long.
30602 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
30605 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
30606 not defined for x86_64 target. Add -mmmx target option when __SSE2__
30608 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
30609 for x86_64 target. Handle -m3dnowa option.
30611 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30613 PR rtl-optimization/70703
30614 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
30615 (update_conflict_hard_regno_costs): Use long instead of unsigned
30616 arithmetic for cost calculation.
30618 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30619 Bernd Edlinger <bernd.edlinger@hotmail.de>
30622 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
30625 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
30628 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
30629 ptr_mode with Pmode throughout.
30630 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
30631 into probe_stack_range and use DImode.
30633 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
30636 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
30637 call_eh_return is true.
30639 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30641 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
30642 Initialize last_match_fntype_index.
30644 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30647 * tree-nvr.c: Include internal-fn.h.
30648 (pass_return_slot::execute): Ignore internal calls without
30651 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30652 Richard Biener <rguenther@suse.de>
30655 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
30656 captures used multiple times, except for the last use.
30657 * generic-match-head.c: Include gimplify.h.
30659 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30661 PR tree-optimization/79390
30662 * target.h (struct noce_if_info): Declare.
30663 * targhooks.h (default_noce_conversion_profitable_p): Declare.
30664 * target.def (noce_conversion_profitable_p): New target hook.
30665 * ifcvt.h (struct noce_if_info): New type, moved from ...
30666 * ifcvt.c (struct noce_if_info): ... here.
30667 (noce_conversion_profitable_p): Renamed to ...
30668 (default_noce_conversion_profitable_p): ... this. No longer
30670 (noce_try_store_flag_constants, noce_try_addcc,
30671 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
30672 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
30673 instead of noce_conversion_profitable_p.
30674 * config/i386/i386.c: Include ifcvt.h.
30675 (ix86_option_override_internal): Don't override
30676 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
30677 (ix86_noce_conversion_profitable_p): New function.
30678 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
30679 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
30680 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
30681 * doc/tm.texi: Regenerated.
30683 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30685 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
30688 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
30691 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
30692 instructions for small multiply cores.
30694 2017-04-04 Jeff Law <law@redhat.com>
30696 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
30698 (mips_expand_vec_perm_const): Initialize elements in orig_perm
30699 that are not set by the loop over the elements.
30701 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30704 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
30705 int mode, convert_modes it to mode as unsigned, otherwise use
30706 lowpart_subreg to mode rather than SImode.
30707 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
30708 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
30709 Use DImode instead of SImode for the shift count operand.
30710 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
30713 2017-04-04 Richard Biener <rguenther@suse.de>
30715 PR middle-end/80281
30716 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
30717 arithmetic done for the negate or the plus. Simplify.
30718 (A - (-B) -> A + B): Likewise.
30719 * fold-const.c (split_tree): Make sure to not negate pointers.
30721 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
30723 PR rtl-optimization/60818
30724 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
30725 a compare of comparisons with the thing compared if this results
30726 in a different machine mode.
30728 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
30730 * alias.c (base_alias_check): Fix typo in comment.
30731 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
30732 * cgraphunit.c (symbol_table::compile): Likewise.
30733 * collect2.c (maybe_run_lto_and_relink): Likewise.
30734 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
30735 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
30736 * config/avr/avr.c (avr_map_op_t): Likewise.
30737 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
30738 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
30739 * config/epiphany/epiphany.md (movcc): Likewise.
30740 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
30741 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
30743 * config/mips/mips.c (mips_save_restore_reg): Likewise.
30744 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
30745 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
30746 * config/sh/sh.c (sh_rtx_costs): Likewise.
30747 * fold-const.c (fold_truth_andor): Likewise.
30748 * genautomata.c (collapse_flag): Likewise.
30749 * gengtype.h (struct type::u::s): Likewise.
30750 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
30751 * input.c (FORMAT_AMOUNT): Likewise.
30752 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
30753 (known_aggs_to_agg_replacement_list): Likewise.
30754 * ipa-inline-analysis.c: Likewise.
30755 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
30756 * ipa-polymorphic-call.c
30757 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
30758 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
30759 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
30761 * modulo-sched.c (apply_reg_moves): Likewise.
30762 * omp-expand.c (build_omp_regions_1): Likewise.
30763 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
30764 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
30765 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
30766 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
30767 * value-prof.c: Likewise.
30768 * var-tracking.c (val_reset): Likewise.
30770 2017-04-03 Richard Biener <rguenther@suse.de>
30772 PR tree-optimization/80275
30773 * fold-const.c (split_address_to_core_and_offset): Handle
30776 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
30778 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
30779 descriptors is at least equal to that of functions.
30781 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30783 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
30785 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30788 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
30789 (mov<IMOD4:mode>): New expander.
30790 (*mov<IMOD4:mode>_internal): New insn and split pattern.
30792 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
30794 PR rtl-optimization/79405
30795 * fwprop.c (propagations_left): New variable.
30796 (forward_propagate_into): Decrement it.
30797 (fwprop_init): Initialize it.
30798 (fw_prop): If the variable has reached zero, stop propagating.
30799 (fwprop_addr): Ditto.
30801 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30804 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
30805 a FUNCTION_DECL, pass it as decl instead of origin to
30808 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
30810 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
30813 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
30816 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
30817 TARGET_VSX_SMALL_INTEGER.
30819 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30821 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30822 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
30824 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30826 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
30827 extraction from odd-numbered MSA register.
30829 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30831 PR middle-end/80173
30832 * expmed.c (store_bit_field_1): Don't attempt to create
30833 a word subreg out of hard registers wider than word if they
30834 have HARD_REGNO_NREGS of 1 for their mode.
30836 PR middle-end/80163
30837 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
30838 conversions to integer types wider than word and pointer.
30841 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
30842 (rtx_equal_for_cselib_p): Pass 0 to it.
30843 * cselib.c (cselib_hasher::equal): Likewise.
30844 (rtx_equal_for_cselib_1): Add depth argument. If depth
30845 is 128, don't look up VALUE locs and punt. Increment
30846 depth in recursive calls when walking VALUE locs.
30848 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
30850 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
30851 (make_gcov_file_name): Use the canonical path name for generating
30853 (read_line): Fix handling of files with ascii null bytes.
30855 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
30857 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
30858 to initialise a vector register instead
30859 of using a const_int.
30861 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30863 PR translation/80189
30864 * gimplify.c (omp_default_clause): Use %qs instead of %s in
30865 diagnostic messages.
30867 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
30870 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
30871 (dfp_diex_<mode>): Update mode of operand 1.
30872 * doc/extend.texi (dxex, dxexq): Document change to return type.
30873 (diex, diexq): Document change to argument type.
30875 2017-03-30 Martin Jambor <mjambor@suse.cz>
30878 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
30879 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
30880 it reflects the signature changes performed at the callee side.
30881 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
30882 to cgraph_build_function_type_skip_args.
30883 (build_function_decl_skip_args): Adjust call to the above function.
30885 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30888 * config/i386/sse.md
30889 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
30890 register as dest whenever it is a MEM not rtx_equal_p to the
30891 corresponding dup operand, and when forcing into reg move the
30892 reg into the memory afterwards.
30893 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
30894 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
30895 for the force_reg mode.
30896 (avx512vl_vextractf128<mode>): Use register as dest either
30897 always when a MEM, or when it is a MEM not rtx_equal_p to the
30898 corresponding dup operand, or even not when it is a CONST_VECTOR
30899 depending on the mode and lo vs. hi.
30900 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
30902 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
30903 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
30904 Likewise. Require that operands[2] is even.
30905 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
30906 Remove extraneous parens. Require that operands[2] is a multiple
30908 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
30909 operands[0] is a MEM if <mask_applied>, the predicates/constraints
30910 disallow memory then.
30912 2017-03-30 Richard Biener <rguenther@suse.de>
30914 PR tree-optimization/77498
30915 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
30916 to non-constants over backedges.
30918 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
30920 PR rtl-optimization/80233
30921 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
30922 as last_combined_insn. Do not test for BARRIER_P separately.
30924 2017-03-29 Andreas Schwab <schwab@suse.de>
30927 * calls.c (prepare_call_address): Convert funexp to Pmode before
30928 copying to temp reg.
30930 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30932 PR tree-optimization/80158
30933 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
30934 Handle possible future case of more than one alternate
30936 (replace_rhs_if_not_dup): Likewise.
30937 (replace_one_candidate): Likewise.
30939 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
30941 PR rtl-optimization/80193
30942 * ira.c (ira): Do not check allocation for LRA.
30944 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
30946 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
30947 (nvptx_output_simt_exit): Declare.
30948 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
30949 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
30950 (init_softstack_frame): Move initialization of crtl->is_leaf to...
30951 (nvptx_declare_function_name): ...here. Emit declaration of local
30952 memory space buffer for omp_simt_enter insn.
30953 (nvptx_output_unisimt_switch): New.
30954 (nvptx_output_softstack_switch): New.
30955 (nvptx_output_simt_enter): New.
30956 (nvptx_output_simt_exit): New.
30957 * config/nvptx/nvptx.h (struct machine_function): New fields
30958 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
30959 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
30960 (UNSPECV_SIMT_EXIT): Ditto.
30961 (omp_simt_enter_insn): New insn.
30962 (omp_simt_enter): New expansion.
30963 (omp_simt_exit): New insn.
30964 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
30966 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
30967 (expand_GOMP_SIMT_ENTER_ALLOC): New.
30968 (expand_GOMP_SIMT_EXIT): New.
30969 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
30970 (GOMP_SIMT_ENTER_ALLOC): Ditto.
30971 (GOMP_SIMT_EXIT): Ditto.
30972 * target-insns.def (omp_simt_enter): New insn.
30973 (omp_simt_exit): Ditto.
30974 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
30976 (lower_rec_simd_input_clauses): Implement SIMT privatization.
30977 (lower_rec_input_clauses): Likewise.
30978 (lower_lastprivate_clauses): Handle SIMT privatization.
30980 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
30981 (ompdevlow_adjust_simt_enter): New.
30982 (find_simtpriv_var_op): New.
30983 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
30984 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
30986 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
30987 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
30988 (copy_decl_for_dup_finish): Ditto.
30990 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
30992 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
30995 * config/i386/i386.c (ix86_option_override_internal): Always
30996 allow -mpreferred-stack-boundary=3 for 64-bit targets.
30998 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31000 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
31002 2017-03-28 Bin Cheng <bin.cheng@arm.com>
31004 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
31005 mark new edge's irreducible flag accordign to it.
31006 (vect_do_peeling): Check loop preheader edge's irreducible flag
31007 and pass it to function slpeel_add_loop_guard.
31009 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
31011 PR tree-optimization/80218
31012 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31013 Update block frequencies and counts.
31015 2017-03-28 Richard Biener <rguenther@suse.de>
31017 PR tree-optimization/78644
31018 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
31019 of a simplification result we may not use it at all.
31021 2017-03-28 Richard Biener <rguenther@suse.de>
31024 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
31025 without arguments, generate default definition of a SSA name.
31027 2017-03-28 Richard Biener <rguenther@suse.de>
31029 PR middle-end/80222
31030 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
31031 TYPE_REF_CAN_ALIAS_ALL references.
31032 * fold-const.c (fold_indirect_ref_1): Likewise.
31034 2017-03-28 Martin Liska <mliska@suse.cz>
31037 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
31038 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
31040 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31041 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
31043 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
31044 (EXTRA_SPECS): Define.
31045 (SUBTARGET_EXTRA_SPECS): Likewise.
31046 (SUBTARGET_CPP_SPEC): Likewise.
31047 * config/arc/elf.h (EXTRA_SPECS): Renamed to
31048 SUBTARGET_EXTRA_SPECS.
31049 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
31051 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31053 * config/arc/simdext.md (vst64_insn): Update pattern.
31054 (vld32wh_insn): Likewise.
31055 (vld32wl_insn): Likewise.
31056 (vld64_insn): Likewise.
31057 (vld32_insn): Likewise.
31059 2017-03-28 Marek Polacek <polacek@redhat.com>
31062 * fold-const.c (fold_comparison): Use protected_set_expr_location
31063 instead of SET_EXPR_LOCATION.
31065 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31067 * tree.c (add_expr): Avoid name lookup warning.
31069 2017-03-27 Jeff Law <law@redhat.com>
31071 PR tree-optimization/80216
31072 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31073 function name. Limit recursion depth.
31074 (record_temporary_equivalences): Corresponding changes.
31076 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31078 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31081 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31084 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31086 * cfgcleanup.c (reg_note_cfa_p): New array.
31087 (insns_have_identical_cfa_notes): New function.
31088 (old_insns_match_p): Don't cross-jump in between /f
31089 and non-/f instructions. If both i1 and i2 are frame related,
31090 verify all CFA notes, their order and content.
31092 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31095 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31096 HImode and SImode with zero extend to DImode to one insn.
31097 (bswap<mode>2_extenddi): Likewise.
31098 (bswapsi2_extenddi): Likewise.
31099 (bswaphi2_extendsi): Likewise.
31100 (bswaphi2): Combine bswap HImode and SImode into one insn.
31101 Separate memory insns from swapping register.
31102 (bswapsi2): Likewise.
31103 (bswap<mode>2): Likewise.
31104 (bswaphi2_internal): Delete, no longer used.
31105 (bswapsi2_internal): Likewise.
31106 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31107 store, and gpr<-gpr swap insns.
31108 (bswap<mode>2_store): Likewise.
31109 (bswaphi2_reg): Register only splitter, combine with the splitter.
31110 (bswaphi2 splitter): Likewise.
31111 (bswapsi2_reg): Likewise.
31112 (bswapsi2 splitter): Likewise.
31113 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
31114 the insns into load, store, and register/register insns.
31115 (bswapdi2_ldbrx): Likewise.
31116 (bswapdi2_load): Likewise.
31117 (bswapdi2_store): Likewise.
31118 (bswapdi2_reg): Likewise.
31120 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
31122 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
31123 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
31125 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
31128 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
31130 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31131 special handling for target option conflicts between dform
31132 options (-mpower9-dform, -mpower9-dform-vector,
31133 -mpower9-dform-scalar) and -mno-direct-move.
31135 2017-03-27 Richard Biener <rguenther@suse.de>
31137 PR tree-optimization/80181
31138 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
31140 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31142 * config/arc/predicates.md (move_double_src_operand): Replace the
31143 call to move_double_src_operand with a call to address_operand.
31145 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31147 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
31148 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
31149 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
31151 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31153 * config/arc/predicates.md (long_immediate_loadstore_operand):
31154 Consider scaled addresses cases.
31156 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31158 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
31159 restored when in interrupt.
31160 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
31161 doesn't have delay slot.
31163 2017-03-27 Richard Biener <rguenther@suse.de>
31166 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
31167 inlined thunk clones.
31169 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31172 * asan.c (instrument_derefs): Copy over last operand from
31173 original COMPONENT_REF to the new COMPONENT_REF with
31174 DECL_BIT_FIELD_REPRESENTATIVE.
31175 * ubsan.c (instrument_object_size): Likewise.
31177 2017-03-27 Richard Biener <rguenther@suse.de>
31179 PR tree-optimization/80170
31180 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
31181 sure DR/SCEV didnt fold in constants we do not see when looking
31182 at the reference base alignment.
31184 2017-03-27 Richard Biener <rguenther@suse.de>
31186 PR middle-end/80171
31187 * gimple-fold.c (fold_ctor_reference): Properly guard against
31188 NULL return value from canonicalize_constructor_val.
31190 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
31193 * config/i386/i386.c (ix86_expand_builtin)
31194 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
31195 flags reg setting and flags reg using instructions.
31196 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
31197 clobbering instructions to zero extend op2.
31199 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
31201 * doc/install.texi (Configuration) <--with-aix-soname>:
31202 Update link to AIX ld.
31204 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
31206 PR rtl-optimization/80160
31207 PR rtl-optimization/80159
31208 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
31209 reg_alternate_class into account.
31211 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
31214 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
31215 to consider in curr_insn_transform.
31217 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31219 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
31220 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
31221 and emit_mode_inner.
31223 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31225 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
31226 argument to the overloaded builtin variants. Use the new flag to
31227 deprecate certain builtin variants.
31228 * config/s390/s390-builtin-types.def: Add new builtin types.
31229 * config/s390/s390-builtins.h: Support new flags field for
31230 overloaded builtins.
31231 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
31232 (s390_macro_to_expand): Enable vector float data type.
31233 (s390_cpu_cpp_builtins_internal): Indicate support of the new
31234 builtins by incrementing the __VEC__ version number.
31235 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
31237 (s390_resolve_overloaded_builtin): Emit error messages depending
31238 on the builtin flags.
31239 * config/s390/s390.c (s390_expand_builtin): Support additional
31240 flags argument. Change error message to match the messages
31241 emitted in s390-c.c.
31242 * config/s390/s390.md: New UNSPEC_* constants.
31243 (op_type): Add new instruction types.
31244 * config/s390/vecintrin.h: Add new builtins and test data class
31246 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
31247 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
31248 (VEC_INEXACT, VEC_NOINEXACT): New constants.
31249 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
31250 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
31251 ("vec_mergel<mode>"): V_HW -> VEC_HW.
31253 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
31254 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
31255 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
31256 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
31258 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
31259 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
31260 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
31261 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
31263 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
31264 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
31265 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
31266 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
31267 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
31268 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
31269 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
31271 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
31272 ("vec_scatter_element<V_HW_4:mode>_DI")
31273 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
31274 ("vec_fpint<mode>", "vflls")
31275 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
31276 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
31277 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
31278 ("*vec_cmphe<mode>_cc"): ... these.
31280 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
31281 mode constant instead of magic value.
31283 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31285 * config/s390/s390.c (s390_expand_vec_compare): Support other
31286 vector floating point modes than just V2DF.
31287 (s390_expand_vcond): Likewise.
31288 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
31289 (s390_cannot_change_mode_class): Prevent mode changes between TF
31290 and V1TF in vector registers.
31291 * config/s390/s390.md (DF, SF): New mode attributes.
31292 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
31293 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
31294 SFmode support for VRs.
31295 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
31297 (VFT, VF_HW): New mode iterators.
31298 (vw, sdx): New mode attributes.
31299 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
31300 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
31301 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
31302 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
31303 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
31304 also the new vector floating point modes. Renaming to ...
31306 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
31307 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
31308 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
31309 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
31310 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
31311 ("vec_unordered<mode>"): ... these.
31313 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
31314 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
31315 ("*vec_extendv2df"): New insn definitions.
31317 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31319 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
31320 ("mulditi3_2", "*muldi3_sign"): New patterns.
31321 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
31322 rename the pattern definition.
31324 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31326 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
31328 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
31330 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31332 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
31333 instruction if possible.
31334 * config/s390/vector.md (vec_halfnumelts): New mode
31336 ("*vec_vllezlf<mode>"): New pattern.
31338 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31340 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
31341 ("popcountv4si2", "popcountv2di2"): Rename to ...
31342 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
31343 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
31345 ("popcount<mode>2_vxe"): New pattern.
31347 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31349 * common/config/s390/s390-common.c (processor_flags_table): Add
31351 * config.gcc: Add arch12.
31352 * config/s390/driver-native.c (s390_host_detect_local_cpu):
31353 Default to arch12 for unknown CPU model numbers.
31354 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
31355 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
31356 PROCESSOR_max sanity check.
31357 * config/s390/s390-opts.h (enum processor_type): Add
31359 * config/s390/s390.c (processor_table): Add arch12.
31360 (s390_expand_builtin): Add check for B_VXE flag.
31361 (s390_issue_rate): Add PROCESSOR_ARCH12.
31362 (s390_get_sched_attrmask): Likewise.
31363 (s390_get_unit_mask): Likewise.
31364 (s390_sched_score): Enable z13 scheduling for arch12.
31365 (s390_sched_reorder): Likewise.
31366 (s390_sched_variable_issue): Likewise.
31367 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
31369 (s390_tune_attr): Use z13 scheduling also for arch12.
31370 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
31371 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
31372 (TARGET_VXE_P): New macros.
31373 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
31374 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
31375 * config/s390/s390.opt: Add arch12 as processor_type.
31377 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31379 * config/s390/s390.md
31380 ("fixuns_truncdddi2", "fixuns_trunctddi2")
31381 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
31382 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
31384 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
31385 Rename expanders to ...
31387 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
31388 ("fixuns_truncdddi2_emu"): ... these.
31390 ("fixuns_trunc<mode>si2_emu"): New expander.
31392 ("*fixuns_truncdfdi2_z13"): Rename to ...
31393 ("*fixuns_truncdfdi2_vx"): ... this.
31395 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31397 * config/s390/2964.md: Remove the single element vector compare
31398 instructions which are no longer used.
31399 * config/s390/s390.c (s390_select_ccmode): Remove handling of
31401 (s390_canonicalize_comparison): Remove handling of DFmode
31403 (s390_expand_vec_compare_scalar): Remove function.
31404 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
31405 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
31407 ("*cmp<mode>_ccs"): Add wfcdb instruction.
31409 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31411 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
31413 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
31414 will anyway by matched by mov<mode>_64dfp.
31416 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31418 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
31419 vlef/vstef. Add missing operand to vleif.
31421 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31423 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
31424 pair for all vector types with 64 bit elements.
31425 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
31426 * config/s390/vector.md (V_HW_64): ... here.
31427 (V_128_NOSINGLE): New mode iterator.
31428 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
31429 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
31430 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
31431 ("*vec_load_pairv2di"): Change to ...
31432 ("*vec_load_pair<mode>"): ... this one.
31434 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31436 * config/s390/constraints.md: Add comments.
31437 (jKK): Reject element sizes > 8 bytes.
31438 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
31440 * config/s390/s390.md: Add the s_operand checks formerly in
31441 s390_split_ok_p to various splitters where they are still
31443 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
31444 for 128 bit vectors. Plus two splitters.
31446 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31448 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
31451 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31454 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
31455 error if the boundary argument is not constant.
31457 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31459 PR rtl-optimization/80112
31460 * loop-doloop.c (doloop_condition_get): Don't check condition
31461 if cmp isn't SET with IF_THEN_ELSE src.
31463 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31465 PR tree-optimization/80158
31466 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
31467 replacing a candidate statement, also replace it for the
31468 candidate's alternate interpretation.
31469 (replace_rhs_if_not_dup): Likewise.
31470 (replace_one_candidate): Likewise.
31472 2017-03-24 Richard Biener <rguenther@suse.de>
31474 PR tree-optimization/80167
31475 * graphite-isl-ast-to-gimple.c
31476 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
31478 (translate_isl_ast_to_gimple::get_rename): Likewise.
31480 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31482 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
31483 handling of certain combinations of target options, including the
31484 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
31485 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
31487 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31490 * config/arm/arm.md (*load_multiple): Add reload_completed to
31491 matching condition.
31493 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31494 Richard Biener <rguenth@suse.de>
31496 PR tree-optimization/79908
31497 PR tree-optimization/80136
31498 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31499 been cast away, gimplify_and_add suffices.
31501 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
31503 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
31505 2017-03-23 Richard Biener <rguenther@suse.de>
31507 PR tree-optimization/80032
31508 * gimplify.c (gimple_push_cleanup): Forced unconditional
31509 cleanups still have to go to the conditional_cleanups
31512 2017-03-22 Jakub Jelinek <jakub@redhat.com>
31514 PR tree-optimization/80072
31515 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
31517 (next_operand_entry_id): Change type to unsigned int.
31518 (sort_by_operand_rank): Make sure to return the right return value
31519 even if unsigned fields are bigger than INT_MAX.
31520 (struct oecount): Change cnt and id type to unsigned int.
31521 (oecount_hasher::equal): Formatting fix.
31522 (oecount_cmp): Make sure to return the right return value
31523 even if unsigned fields are bigger than INT_MAX.
31524 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
31527 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
31528 TREE_READONLY on result if writing it more than once.
31531 * doc/invoke.texi (-fsanitize=thread): Document that with
31532 -fnon-call-exceptions atomics are not able to throw
31536 * tsan.c: Include tree-eh.h.
31537 (instrument_builtin_call): Call maybe_clean_eh_stmt or
31538 maybe_clean_or_replace_eh_stmt where needed.
31539 (instrument_memory_accesses): Add cfg_changed argument.
31540 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
31541 if it returned true.
31542 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
31544 PR rtl-optimization/63191
31545 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
31546 wrapper function, moved the whole old content into ...
31547 (ix86_delegitimize_address_1): ... this. New inline function.
31548 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
31549 true as last argument instead of ix86_delegitimize_address.
31551 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31553 * config/aarch64/aarch64.c (generic_branch_cost): Copy
31554 cortexa57_branch_cost.
31556 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31558 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
31560 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31563 * doc/md.texi (Constraints): Document wA constraint.
31564 * config/rs6000/constraints.md (wA): New.
31565 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
31566 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
31567 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
31568 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
31570 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
31573 * gimplify.c (is_oacc_declared): New function.
31574 (oacc_default_clause): Use it to set default flags for acc declared
31575 variables inside parallel regions.
31576 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
31577 declared variables.
31578 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
31579 declare attribute to any decl as necessary.
31581 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
31584 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
31585 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
31586 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
31587 (arm_arch_lpae): This.
31588 * config/arm/arm.c (arm_arch7ve): Rename into ...
31589 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
31590 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
31593 2017-03-22 Martin Liska <mliska@suse.cz>
31596 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
31597 error message instead of an ICE.
31599 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31601 * doc/extend.texi (6.11 Additional Floating Types): Revise.
31603 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
31605 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31607 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31610 2017-03-21 Martin Sebor <msebor@redhat.com>
31612 * doc/extend.texi: Use "cannot" instead of "can't."
31613 * doc/hostconfig.texi: Same.
31614 * doc/install.texi: Same.
31615 * doc/invoke.texi: Same.
31616 * doc/loop.texi: Same.
31617 * doc/md.texi: Same.
31618 * doc/objc.texi: Same.
31619 * doc/rtl.texi: Same.
31620 * doc/tm.texi: Same.
31621 * doc/tm.texi.in: Same.
31622 * doc/trouble.texi: Same.
31624 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
31627 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
31628 (collect_checksum_attributes): Set it.
31629 (die_checksum_ordered): Use it.
31631 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31633 PR tree-optimization/79908
31634 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
31635 change: For a VA_ARG whose LHS has been cast away, use
31636 force_gimple_operand to construct the side effects.
31638 2017-03-21 David Malcolm <dmalcolm@redhat.com>
31640 PR translation/80001
31641 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
31642 more amenable to translation.
31643 (oacc_loop_auto_partitions): Likewise.
31645 2017-03-21 Marek Polacek <polacek@redhat.com>
31646 Martin Sebor <msebor@redhat.com>
31648 PR tree-optimization/80109
31649 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
31650 on INTEGRAL_TYPE_P.
31652 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31653 Segher Boessenkool <segher@kernel.crashing.org>
31656 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
31657 check reg_used_between_p between insn and one of succ or succ2
31658 depending on if succ is artificial insn not inserted into insn
31661 2017-03-21 Martin Liska <mliska@suse.cz>
31663 PR gcov-profile/80081
31664 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
31665 * doc/gcc.texi: Include gcov-dump stuff.
31666 * doc/gcov-dump.texi: New file.
31668 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
31670 PR rtl-optimization/79150
31671 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
31672 conditional jump, if the jump is the last insn of the loop.
31674 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31675 Richard Biener <rguenth@suse.de>
31677 PR tree-optimization/79908
31678 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31679 been cast away, use force_gimple_operand to construct the side
31682 2017-03-21 Martin Liska <mliska@suse.cz>
31684 PR libfortran/79956
31685 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
31688 2017-03-21 Brad Spengler <spender@grsecurity.net>
31691 * plugin.c (htab_hash_plugin): New function.
31692 (add_new_plugin): Use it and adjust.
31693 (parse_plugin_arg_opt): Adjust.
31694 (init_one_plugin): Likewise.
31696 2017-03-21 Richard Biener <rguenther@suse.de>
31698 PR tree-optimization/80032
31699 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
31700 if set force the cleanup to happen unconditionally.
31701 (gimplify_target_expr): Push inserted clobbers with force_uncond
31702 to avoid them being removed by control-dependent DCE.
31704 2017-03-21 Richard Biener <rguenther@suse.de>
31706 PR tree-optimization/80122
31707 * tree-inline.c (copy_bb): Do not expans va-arg packs or
31708 va_arg_pack_len when the inlined call stmt requires pack
31710 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
31712 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31715 * tsan.c (instrument_builtin_call): If the memory model argument
31716 is not a constant, assume it is valid.
31719 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
31722 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31724 PR rtl-optimization/79910
31725 * combine.c (can_combine_p): Do not allow combining an I0 or I1
31726 if its dest is used by an insn before I2 (other than the combined
31727 insns themselves, which are properly handled already).
31729 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31732 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31734 * combine.c (record_used_regs): New static function.
31735 (try_combine): Handle situations where there is an additional
31736 instruction between I2 and I3 which needs to have a LOG_LINK
31740 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31742 * combine.c (try_combine): Delete redundant i1 test. Call
31743 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31745 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31748 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
31749 alternatives 13/14.
31751 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31753 PR tree-optimization/80054
31754 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
31755 the optimization if a PHI or any of its arguments is not dominated
31756 by the candidate's basis. Use gphi* rather than gimple* as
31758 (replace_profitable_candidates): Clean up a gimple* variable that
31759 should be a gphi* variable.
31761 2017-03-20 Martin Sebor <msebor@redhat.com>
31764 * doc/extend.texi (attribute constructor): Document present limitation.
31766 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
31769 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
31770 __POWER9_VECTOR__ #ifdef control, change template definition to
31771 use Power9-specific built-in function.
31772 (vec_any_eq): Likewise.
31773 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
31774 to control outcomes from this test.
31775 (vector_ae_<mode>p): For VEC_F modes, likewise.
31777 2017-03-20 Ian Lance Taylor <iant@google.com>
31779 * config/i386/i386.c (ix86_function_regparm): Save an extra
31780 register for -fsplit-stack with DECL_STATIC_CHAIN.
31782 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31785 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
31786 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
31788 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31790 * config/riscv/riscv.c (riscv_print_operand): Use "fence
31792 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
31795 2017-03-20 Marek Polacek <polacek@redhat.com>
31798 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
31800 2017-03-20 Richard Biener <rguenther@suse.de>
31802 PR tree-optimization/80113
31803 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
31804 allocate extra SSA name for PHI def.
31805 (add_close_phis_to_outer_loops): Likewise.
31806 (add_close_phis_to_merge_points): Likewise.
31807 (copy_loop_close_phi_args): Likewise.
31808 (copy_cond_phi_nodes): Likewise.
31810 2017-03-20 Martin Liska <mliska@suse.cz>
31812 PR middle-end/79753
31813 * tree-chkp.c (chkp_build_returned_bound): Do not build
31814 returned bounds for a LHS that's not a BOUNDED_P type.
31816 2017-03-20 Martin Liska <mliska@suse.cz>
31820 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
31821 COMPLEX_CST and VECTOR_CST.
31823 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31826 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
31827 target operand. A new splitter adds the clobber statement in case
31828 the target operand is dead anyway.
31830 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
31832 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
31833 to age-old versions of binutils and glibc.
31835 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
31837 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
31839 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31841 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
31843 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31845 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
31846 requirement for binutils 2.13.
31848 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31850 * combine.c (try_combine): Delete redundant i1 test. Call
31851 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31853 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
31855 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
31856 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
31858 <riscv64-*-elf>: Re-arrange section
31859 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
31860 <riscv32-*-linux>: Likewise.
31861 <riscv64-*-elf>: Likewise
31862 <riscv64-*-linux>: Likewise.
31864 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
31867 * aarch64.opt(verbose-cost-dump): Fix typo.
31869 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
31872 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
31873 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
31875 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31877 * reload.c (find_reloads): When reloading a nonoffsettable address,
31878 use RELOAD_OTHER for it and its address reloads.
31880 PR rtl-optimization/79910
31881 * combine.c (record_used_regs): New static function.
31882 (try_combine): Handle situations where there is an additional
31883 instruction between I2 and I3 which needs to have a LOG_LINK
31886 2017-03-17 Jeff Law <law@redhat.com>
31888 PR tree-optimization/71437
31889 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
31890 conditional in the hash table first.
31891 (vrp_dom_walker::before_dom_children): Extract condition from
31892 ASSERT_EXPR. Record condition, its inverion and any implied
31893 conditions as well.
31895 2017-03-17 Marek Polacek <polacek@redhat.com>
31896 Markus Trippelsdorf <markus@trippelsdorf.de>
31898 PR tree-optimization/80079
31899 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
31902 2017-03-17 Richard Biener <rguenther@suse.de>
31904 PR middle-end/80075
31905 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
31906 Properly verify the LHS before the RHS possibly claims to be
31908 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
31911 2017-03-17 Martin Jambor <mjambor@suse.cz>
31913 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
31914 (List of -O2 options): Likewise.
31915 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
31918 2017-03-17 Tom de Vries <tom@codesourcery.com>
31920 * gcov-dump.c (print_usage): Print bug_report_url.
31922 2017-03-17 Richard Biener <rguenther@suse.de>
31924 PR middle-end/80050
31925 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
31926 (parser::peek): Likewise.
31928 2017-03-17 Richard Biener <rguenther@suse.de>
31930 PR tree-optimization/80048
31931 * sese.c (free_sese_info): Properly release rename_map and
31932 copied_bb_map elements.
31934 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
31936 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
31937 Add linked-list forward and backlinks. Insert on
31938 construction, remove on destruction.
31939 (class pass_store_merging): Add m_stores_head field.
31940 (pass_store_merging::terminate_and_process_all_chains):
31941 Iterate over m_stores_head list.
31942 (pass_store_merging::terminate_all_aliasing_chains):
31944 (pass_store_merging::execute): Check for debug stmts first.
31945 Push new chains onto the m_stores_head stack.
31947 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
31950 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
31951 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
31952 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
31954 2017-03-16 Jeff Law <law@redhat.com>
31956 PR tree-optimization/71437
31957 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
31958 member function. Implementation moved into after_dom_children
31959 member function and into the threader's thread_outgoing_edges
31961 (dom_opt_dom_walker::after_dom_children): Simplify by moving
31962 some code into new thread_outgoing_edges.
31963 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
31964 definition. Simplify marker handling (do it here). Assume we always
31965 have the available expression and the const/copies tables.
31966 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
31968 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
31969 * tree-vrp.c (equiv_stack): No longer file scoped.
31970 (vrp_dom_walker): New class.
31971 (vrp_dom_walker::before_dom_children): New member function.
31972 (vrp_dom_walker::after_dom_children): Likewise.
31973 (identify_jump_threads): Setup domwalker. Use it rather than
31974 walking edges in a random order by hand. Simplify setup/finalization.
31975 (finalize_jump_threads): Remove.
31976 (vrp_finalize): Do not call identify_jump_threads here.
31977 (execute_vrp): Do it here instead and call thread_through_all_blocks
31980 PR tree-optimization/71437
31981 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
31983 (simplify_stmt_for_jump_threading): Add basic_block argument. All
31985 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
31986 (dom_opt_dom_walker::thread_across_edge): Remove
31987 handle_dominating_asserts argument. All callers changed.
31988 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
31989 changes. Remove calls to lhs_of_dominating_assert. Other
31990 uses of handle_dominating_asserts turn into unconditional code
31991 (simplify_control_stmt_condition_1): Likewise.
31992 (simplify_control_stmt_condition): Likewise.
31993 (thread_through_normal_block, thread_across_edge): Likewise.
31994 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
31995 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
31996 object if it is not an SSA_NAME.
31997 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
31998 before calling into the VRP specific simplifiers.
31999 (identify_jump_threads): Remove handle_dominating_asserts
32002 2017-03-16 Jakub Jelinek <jakub@redhat.com>
32005 * tree-diagnostic.c (default_tree_printer): No longer static.
32006 * tree-diagnostic.h (default_tree_printer): New prototype.
32008 2017-03-16 Tamar Christina <tamar.christina@arm.com>
32010 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
32011 Change ins into fmov.
32013 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32015 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
32016 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
32017 Use h_con constraint for operand 1.
32018 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
32019 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
32021 2017-03-15 Jeff Law <law@redhat.com>
32023 PR tree-optimization/71437
32024 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
32025 (record_temporary_equivalences): Use it.
32027 PR tree-optimization/71437
32028 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
32029 tree-ssa-scopedtables.
32030 (lookup_avail_expr, build_and_record_new_cond): Likewise.
32031 (record_conditions, record_cond, vuse_eq): Likewise.
32032 (record_edge_info): Adjust to API tweak of record_conditions.
32033 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
32034 (record_temporary_equivalences, optimize_stmt): Likewise.
32035 (eliminate_redundant_computations): Likewise.
32036 (record_equivalences_from_stmt): Likewise.
32037 * tree-ssa-scopedtables.c: Include options.h and params.h.
32038 (vuse_eq): New function, moved from tree-ssa-dom.c
32039 (build_and_record_new_cond): Likewise.
32040 (record_conditions): Likewise. Accept vector of conditions rather
32041 than edge_equivalence structure for first argument.
32042 for the first argument.
32043 (avail_exprs_stack::lookup_avail_expr): New member function, moved
32044 from tree-ssa-dom.c.
32045 (avail_exprs_stack::record_cond): Likewise.
32046 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
32047 from tree-ssa-dom.c.
32048 (avail_exprs_stack): Add new member functions lookup_avail_expr
32050 (record_conditions): Declare.
32052 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32055 * lra-constraints.c (process_alt_operands): Increase reject for
32056 reloading an input/output operand.
32058 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32061 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32062 insns to convert from signed/unsigned char/short to IEEE 128-bit
32064 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32066 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32069 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32070 subreg of inner mode for values already in registers.
32072 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32074 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32075 iteration reg is used after the loop.
32077 2017-03-14 Martin Sebor <msebor@redhat.com>
32079 PR tree-optimization/79800
32080 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32081 precision in negative-positive range.
32082 (format_floating): Call non-const overload with adjusted precision.
32084 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32087 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32090 2017-03-14 Martin Sebor <msebor@redhat.com>
32092 PR middle-end/80020
32093 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32094 * builtins.def (aligned_alloc): Use it.
32097 * Makefile.in (GTFILES): Add calls.c.
32098 * calls.c: Include "gt-calls.h".
32100 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32102 PR rtl-optimization/79728
32103 * regs.h (struct target_regs): New field
32104 x_contains_allocatable_regs_of_mode.
32105 (contains_allocatable_regs_of_mode): New macro.
32106 * reginfo.c (init_reg_sets_1): Initialize it, and change
32107 contains_reg_of_mode so it includes global regs as well.
32108 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32109 rather than contains_regs_of_mode.
32111 2017-03-14 Martin Liska <mliska@suse.cz>
32113 * doc/invoke.texi: Document options that can't be combined with
32114 -fcheck-pointer-bounds.
32116 2017-03-14 Martin Liska <mliska@suse.cz>
32118 PR middle-end/79831
32119 * doc/invoke.texi (-Wchkp): Document the option.
32121 2017-03-14 Martin Liska <mliska@suse.cz>
32123 * Makefile.in: Install gcov-dump.
32125 2017-03-14 Martin Liska <mliska@suse.cz>
32127 * multiple_target.c (expand_target_clones): Bail out for
32128 an invalid attribute.
32130 2017-03-14 Richard Biener <rguenther@suse.de>
32132 * alias.c (struct alias_set_entry): Pack properly.
32133 * cfgloop.h (struct loop): Likewise.
32134 * cse.c (struct set): Likewise.
32135 * ipa-utils.c (struct searchc_env): Likewise.
32136 * loop-invariant.c (struct invariant): Likewise.
32137 * lra-remat.c (struct cand): Likewise.
32138 * recog.c (struct change_t): Likewise.
32139 * rtl.h (struct address_info): Likewise.
32140 * symbol-summary.h (function_summary): Likewise.
32141 * tree-loop-distribution.c (struct partition): Likewise.
32142 * tree-object-size.c (struct object_size_info): Likewise.
32143 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
32144 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
32145 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
32146 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
32147 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
32148 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
32149 (struct _stmt_vec_info): Likewise.
32151 2017-03-14 Martin Liska <mliska@suse.cz>
32154 * multiple_target.c (create_dispatcher_calls): Check that
32155 a target can create a function dispatcher.
32157 2017-03-14 Martin Liska <mliska@suse.cz>
32160 * multiple_target.c (expand_target_clones): Drop local.local
32161 flag for default implementation.
32163 2017-03-14 Richard Biener <rguenther@suse.de>
32165 PR tree-optimization/80030
32166 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
32168 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
32170 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
32171 gcc_fallthrough() instead of __attribute__((fallthrough));
32173 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32175 * doc/gcc.texi: Remove "up" link to (DIR).
32176 * doc/gccint.texi: Ditto.
32178 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32180 * doc/install.texi (Specific) <avr>: Remove reference to
32183 2017-03-13 Jeff Law <law@redhat.com>
32185 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
32186 attribute rather than comments.
32188 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
32189 match_scratch operand is highest.
32191 2017-03-13 Martin Liska <mliska@suse.cz>
32193 PR middle-end/78339
32194 * ipa-pure-const.c (warn_function_noreturn): If the declarations
32195 is a CHKP clone, use original declaration.
32197 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32199 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
32200 (arc_conditional_register_usage): Use a different allocation order
32201 when optimizing for size.
32202 * common/config/arc/arc-common.c (arc_option_optimization_table):
32203 Section anchors default on when optimizing for size.
32205 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32207 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
32209 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32211 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
32212 * config/arc/arc.md (cpu_facility): Add cd variant.
32213 (*movqi_insn): Add code density variant.
32214 (*movhi_insn): Likewise.
32215 (*movqi_insn): Likewise.
32216 (*addsi3_mixed): Likewise.
32217 (subsi3_insn): Likewise.
32219 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32221 * config/arc/arc.md (movsi_cond_exec): Update constraint.
32223 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32225 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
32226 expressions with MINUS and UNARY ops.
32228 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32231 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
32233 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
32234 between vec_select and vector argument.
32235 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
32236 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
32237 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
32238 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
32239 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
32240 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
32242 2017-03-13 Richard Biener <rguenther@suse.de>
32245 * params.def (vect-max-peeling-for-alignment): Fix typo.
32247 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32249 * doc/install.texi (Specific) <mips-*-*>: Remove description of
32250 issue that only occurred with binutils below 2.18.
32252 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32254 * doc/install.texi (Specific) <cris-axis-elf>: No longer
32255 refer to binutils 2.11/2.12 minimum.
32257 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32259 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
32260 ftp.kernel.org and simplify binutils requirement.
32262 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
32264 * doc/invoke.texi (Warning Options): Fix spelling of link-time
32266 (Optimize Options): Ditto. Also remove redundancy.
32268 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32270 PR translation/79848
32271 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
32273 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
32274 to G_ to avoid double translation.
32276 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32278 PR translation/79923
32279 * auto-profile.c (get_combined_location): Convert leading
32280 character of diagnostics to lower case and remove trailing period.
32281 (read_profile): Likewise for various diagnostics.
32282 * config/arm/arm.c (arm_option_override): Remove trailing period
32283 from various diagnostics.
32284 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
32285 (msp430_expand_delay_cycles): Likewise.
32287 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32290 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
32291 full command-line argument, rather than just "str".
32292 (aarch64_validate_march): Likewise.
32293 (aarch64_validate_mtune): Likewise.
32295 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
32297 PR rtl-optimization/78911
32298 * lra-assigns.c (must_not_spill_p): New function.
32299 (spill_for): Use it.
32301 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32303 PR tree-optimization/79981
32304 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
32305 ATOMIC_COMPARE_EXCHANGE ifn result.
32306 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
32307 IFN_ATOMIC_COMPARE_EXCHANGE.
32309 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32312 * opts.c (parse_sanitizer_options): Add missing question mark to
32313 "did you mean" message.
32315 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32317 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
32319 (VMULEUH_UNS): Likewise.
32320 (VMULOUB_UNS): Likewise.
32321 (VMULOUH_UNS): Likewise.
32322 * config/rs6000/rs6000.c (builtin_function_type): Remove
32323 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
32325 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32328 * read-rtl-function.c (function_reader::read_rtx_operand): Update
32329 x with result of extra_parsing_for_operand_code_0.
32330 (function_reader::extra_parsing_for_operand_code_0): Convert
32331 return type from void to rtx, returning x. When reading
32332 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
32333 larger size containing struct block_symbol.
32335 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
32337 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
32338 -mfloat128-hardware without -m64.
32340 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
32343 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
32344 entries to the case statement that marks unsigned arguments to
32345 overloaded functions.
32347 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32349 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
32350 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
32352 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
32355 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
32356 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
32358 2017-03-10 Martin Liska <mliska@suse.cz>
32362 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
32363 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
32366 2017-03-10 Olivier Hainque <hainque@adacore.com>
32368 * tree-switch-conversion (array_value_type): Start by resetting
32369 candidate type to it's main variant.
32371 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32373 PR rtl-optimization/79909
32374 * combine.c (try_combine): Use simplify_replace_rtx on individual
32375 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
32376 of the whole CALL_INSN_FUNCTION_USAGE.
32378 PR tree-optimization/79972
32379 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
32380 get_range_info on SSA_NAMEs. Formatting fixes.
32382 2017-03-10 Richard Biener <rguenther@suse.de>
32383 Jakub Jelinek <jakub@redhat.com>
32385 PR tree-optimization/77975
32386 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
32387 edge to be constant.
32388 (get_val_for): For constant x return it. Formatting fix.
32389 (loop_niter_by_eval): Avoid pointless looping if the next iteration
32390 would use the same bases as the current one.
32392 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32394 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
32395 instead of vec_select for V1TImode.
32396 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
32398 (VSX_LE_128): Add V1TI to this mode iterator.
32399 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
32400 (*vsx_le_perm_store_<mode>): Likewise.
32401 (pre-reload splitter for VSX stores): Likewise.
32402 (post-reload splitter for VSX stores): Likewise.
32403 (*vsx_xxpermdi2_le_<mode>): Likewise.
32404 (*vsx_lxvd2x2_le_<mode>): Likewise.
32405 (*vsx_stxvd2x2_le_<mode>): Likewise.
32407 2017-03-09 Michael Eager <eager@eagercon.com>
32409 Correct failures with --enable-checking=yes,rtl.
32411 * config/microblaze/microblaze.c (microblaze_expand_shift):
32412 Replace GET_CODE test with CONST_INT_P and INTVAL test with
32413 test for const0_rtx.
32414 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
32415 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
32417 2017-03-09 Richard Biener <rguenther@suse.de>
32419 PR tree-optimization/79977
32420 * graphite-scop-detection.c (scop_detection::merge_sese):
32421 Handle the case of extra exits to blocks dominating the entry.
32423 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
32425 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
32428 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
32430 PR rtl-optimization/79949
32431 * lra-constraints.c (process_alt_operands): Check memory when
32432 trying to predict a cycle. Print about the overall increase.
32434 2017-03-09 Richard Biener <rguenther@suse.de>
32436 PR middle-end/79971
32437 * gimple-expr.c (useless_type_conversion_p): Preserve
32438 TYPE_SATURATING for fixed-point types.
32440 2017-03-09 Richard Biener <rguenther@suse.de>
32443 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
32444 alignment of BLKmode params.
32446 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32449 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
32450 (VALL_NO_V2Q): Likewise.
32452 * config/aarch64/aarch64-simd.md
32453 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
32455 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
32456 VALL_NO_V2Q mode iterator.
32457 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
32459 2017-03-09 Martin Liska <mliska@suse.cz>
32461 PR tree-optimization/79631
32462 * tree-chkp-opt.c (chkp_is_constant_addr): Call
32463 tree_int_cst_sign_bit just for INTEGER constants.
32465 2017-03-09 Martin Liska <mliska@suse.cz>
32469 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
32472 2017-03-09 Marek Polacek <polacek@redhat.com>
32475 * tree.c (inchash::add_expr): Handle TREE_VEC.
32477 2017-03-09 Martin Liska <mliska@suse.cz>
32480 (chkp_narrow_size_and_offset): New function.
32481 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
32482 (void chkp_parse_bit_field_ref): New function.
32483 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
32484 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
32486 2017-03-09 Martin Liska <mliska@suse.cz>
32489 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
32490 (chkp_find_bounds_1): Remove gcc_unreachable.
32492 2017-03-09 Jakub Jelinek <jakub@redhat.com>
32495 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
32496 BUILT_IN_SYNC*, determine the access type from the size suffix and
32497 always build a MEM_REF with that type. Handle forgotten
32498 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
32501 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
32502 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
32503 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
32504 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
32505 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
32506 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
32507 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
32508 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
32509 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
32510 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
32511 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
32512 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
32513 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
32514 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
32515 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
32516 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
32517 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
32518 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
32519 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
32520 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
32521 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
32522 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
32523 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
32524 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
32525 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
32526 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
32527 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
32528 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
32529 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
32530 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
32531 definitions outside of __OPTIMIZE__ guarded section.
32534 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
32535 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
32536 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
32537 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
32540 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32542 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
32543 ("vfenez<mode>"): Add missing constraints.
32545 2017-03-08 Martin Sebor <msebor@redhat.com>
32548 * config/nds32/nds32.c (nds32_option_override):
32549 Fix misspelled diagnostic.
32551 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32554 * gimplify.c (gimplify_omp_for): Replace index var in outer
32555 taskloop statement with an artificial variable and add
32556 OMP_CLAUSE_PRIVATE clause for it.
32558 2017-03-08 Richard Biener <rguenther@suse.de>
32560 PR tree-optimization/79955
32561 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
32562 for accesses that are completely outside of the variable.
32564 2017-03-08 Andrew Haley <aph@redhat.com>
32566 PR tree-optimization/79943
32567 * tree-ssa-loop-split.c (compute_new_first_bound): When
32568 calculating the new upper bound, (END-BEG) should be added, not
32571 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32573 * config/avr/avr.md (setmemhi): Make sure match_dup
32574 operand number comes before match_scratch.
32576 2017-03-08 Richard Biener <rguenther@suse.de>
32578 PR tree-optimization/79920
32579 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
32580 with ncopies == 1 to ...
32581 (vect_transform_slp_perm_load): ... here. Properly compute
32582 all element loads by iterating VF times over the group. Do
32583 not handle ncopies (computed in a broken way) in
32584 vect_create_mask_and_perm.
32586 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32589 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
32590 is a uniform vector, use uniform_vector_p return value instead of
32591 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
32593 2017-03-07 Marek Polacek <polacek@redhat.com>
32595 PR middle-end/79809
32596 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
32597 (alloca_call_type): Likewise.
32599 2017-03-07 Martin Liska <mliska@suse.cz>
32601 * gcov.c (process_args): Put comment to correct location.
32603 2017-03-07 Martin Liska <mliska@suse.cz>
32605 PR middle-end/68270
32606 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
32607 Use array_at_struct_end_p instead of DECL_CHAIN (field).
32608 (chkp_narrow_bounds_for_field): Likewise.
32609 (chkp_parse_array_and_component_ref): Pass one more argument to
32612 2017-03-07 Richard Biener <rguenther@suse.de>
32614 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
32617 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
32619 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
32620 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
32622 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32625 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
32626 to end of description.
32627 (PARAM_MAX_STORES_TO_MERGE): Likewise.
32629 2017-03-07 Jakub Jelinek <jakub@redhat.com>
32631 PR rtl-optimization/79901
32632 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
32634 (*avx512f_<code><mode>3<mask_name>): ... this.
32635 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
32636 iterator instead of VI8_AVX2_AVX512BW.
32638 PR rtl-optimization/79901
32639 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
32640 min/max expander, expand it using expand_vec_cond_expr.
32643 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
32646 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32649 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
32650 to void * for PCH reasons.
32651 * dwarf2out.c (output_loc_operands, output_die): Cast
32652 v.val_vec.array to unsigned char *.
32654 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
32657 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
32660 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
32662 PR rtl-optimization/79571
32663 * lra-constraints.c (process_alt_operands): Calculate static
32664 reject and subtract it from overall when only addresses will be
32667 2017-03-06 Julia Koval <julia.koval@intel.com>
32670 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
32671 incoming stack boundary to 128 for 64-bit targets.
32673 2017-03-06 Richard Biener <rguenther@suse.de>
32675 PR tree-optimization/79894
32676 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
32677 to NULL after folding it.
32679 2017-03-06 Richard Biener <rguenther@suse.de>
32681 PR tree-optimization/79824
32682 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
32683 check disabling peeling for gaps.
32685 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
32687 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
32688 attributes): Document gettimeofday.
32690 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32692 * config/s390/s390.c (s390_option_override_internal): Set
32693 PARAM_MIN_VECT_LOOP_BOUND
32695 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32697 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
32698 * config/s390/s390.md: Likewise.
32700 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32703 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
32704 (<avx2_avx512>_perm<mode>): Rename to ...
32705 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
32707 (<avx512>_perm<mode>_mask): Rename to ...
32708 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
32710 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
32711 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
32712 instead of VI8F_256_512.
32713 (avx512f_perm<mode>): New define_expand.
32714 (avx512f_perm<mode>_mask): Likewise.
32715 (avx512f_perm<mode>_1<mask_name>): New define_insn.
32716 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
32718 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32720 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
32721 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
32723 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
32725 2017-03-06 Martin Liska <mliska@suse.cz>
32728 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
32729 when having a SSA NAME w/o VAR_DECL assigned to it.
32731 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32733 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
32734 msa_dpsub_<su>_d): Fix MODE for vec_select.
32736 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32738 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
32740 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
32742 2017-03-06 Richard Biener <rguenther@suse.de>
32744 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
32745 * plugin.c (register_plugin_info): Likewise.
32746 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
32748 2017-03-05 Jakub Jelinek <jakub@redhat.com>
32750 * config/i386/sse.md (sse_storehps, sse_storelps,
32751 avx_<castmode><avxsizesuffix>_<castmode>,
32752 avx512f_<castmode><avxsizesuffix>_<castmode>,
32753 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
32754 in condition that at least one operand is not a MEM.
32756 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32758 PR middle-end/79805
32759 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
32760 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
32762 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
32763 gimple_call_nothrow_p flag based on whether original builtin can throw.
32764 If it can, emit following stmts on the fallthrough edge.
32765 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
32766 don't create new bb if inserting just debug stmts on the edge, try to
32767 insert them on the fallthru bb or just reset debug stmts.
32769 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
32772 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
32773 restore recog_data (including the operand rtxes inside it) around
32774 the call to get_insn_template.
32776 2017-03-03 Martin Sebor <msebor@redhat.com>
32778 PR tree-optimization/79699
32779 * context.c (context::~context): Free MPFR caches to avoid
32780 a memory leak on program exit.
32782 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32784 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
32785 Use wide_int::ulow () instead of .elt (0).
32787 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32789 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
32790 (*pushxf): Limit oF constraint to 32bit targets and add oC
32791 constraint for 64bit targets.
32792 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
32793 (*pushdf): Change rmF constraint to rmC.
32795 2017-03-03 Martin Liska <mliska@suse.cz>
32797 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
32798 Remove unused variable.
32800 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32803 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
32804 is a memory operand, increase num_memory.
32805 (ix86_expand_args_builtin): Likewise.
32807 2017-03-03 Jan Hubicka <jh@suse.cz>
32810 * ipa-devirt.c (maybe_record_node): Properly handle
32811 __cxa_pure_virtual visibility.
32813 2017-03-03 Martin Liska <mliska@suse.cz>
32815 PR tree-optimization/79803
32816 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
32818 (pass_loop_prefetch::execute): Disabled optimization if an
32819 assumption about L1 cache size is not met.
32821 2017-03-03 Martin Liska <mliska@suse.cz>
32823 PR rtl-optimization/79574
32824 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
32825 (hash_scan_set): Likewise.
32826 (dump_hash_table): Likewise.
32827 (hoist_code): Likewise.
32829 2017-03-03 Richard Biener <rguenther@suse.de>
32831 * fixed-value.c (fixed_from_string): Restore use of elt (1)
32832 in place of uhigh ().
32833 (fixed_convert_from_real): Likewise.
32835 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32838 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
32840 2017-03-03 Richard Biener <rguenther@suse.de>
32842 PR middle-end/79818
32843 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
32844 TYPE_OVERFLOW_UNDEFINED check.
32846 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32848 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
32850 (vector_ae_<mode>_p): Likewise.
32851 (vector_nez_<mode>_p): Likewise.
32852 (vector_ne_v2di_p): Likewise.
32853 (vector_ae_v2di_p): Likewise.
32854 (vector_ne_<mode>_p): Likewise.
32855 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
32857 (vsx_tsqrt<mode>2_fe): Likewise.
32859 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
32862 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
32864 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32866 PR rtl-optimization/79780
32867 * cprop.c (one_cprop_pass): When second and further conditional trap
32868 in a single basic block is turned into an unconditional trap, turn it
32869 into a deleted note to avoid RTL verification failures.
32871 2017-03-02 Richard Biener <rguenther@suse.de>
32873 * fold-const.c (const_binop): Use ulow () instead of elt (0).
32875 2017-03-02 Richard Biener <rguenther@suse.de>
32877 PR tree-optimization/79345
32879 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
32880 param and abort the walk, returning -1 if it is hit.
32881 (walk_aliased_vdefs): Take a limit param and pass it on.
32882 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
32883 defaulting to 0 and return a signed int.
32884 * tree-ssa-uninit.c (struct check_defs_data): New struct.
32885 (check_defs): New helper.
32886 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
32887 about uninitialized memory.
32888 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
32889 bogus uninitialized warning.
32890 (fixed_convert_from_real): Likewise.
32892 2017-03-02 Bin Cheng <bin.cheng@arm.com>
32894 PR tree-optimization/66768
32895 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
32896 iv_use if base object can't be determined.
32898 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32900 PR tree-optimization/79345
32901 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
32902 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
32903 (get_pattern_stats): Initialize it.
32904 * genemit.c (gen_expand): Verify match_scratch numbers come after
32905 match_operand/match_dup numbers.
32906 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
32907 match_scratch numbers.
32908 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
32910 * config/s390/s390.md (trunctdsd2): Likewise.
32912 2017-03-02 Richard Biener <rguenther@suse.de>
32914 * wide-int.h (wide_int_storage::operator=): Implement in terms
32917 2017-03-02 Richard Biener <rguenther@suse.de>
32919 PR tree-optimization/79777
32920 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
32921 the to insert expression to sth existing.
32923 2017-03-01 Martin Sebor <msebor@redhat.com>
32925 PR middle-end/79692
32926 * gimple-ssa-sprintf.c
32927 (directive::known_width_and_precision): New function.
32928 (format_integer): Use it.
32929 (get_mpfr_format_length): Consider the full range of precision
32930 when computing %g output with the # flag. Set the likely byte
32931 count to 3 rather than 1 when precision is indeterminate.
32932 (format_floating): Correct the lower bound of precision.
32934 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32936 * doc/invoke.texi: Document default code model for 64-bit Linux.
32938 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32941 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
32942 udiv rather than div since input pattern is unsigned.
32944 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
32946 * config/i386/i386.c (print_reg): Warn for values of
32947 unsupported size in integer register.
32949 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
32952 * config/rs6000/predicates.md (current_file_function_operand): Do
32953 not allow self calls to be local if the function is replaceable.
32955 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
32958 * config/rs6000/altivec.h (vec_ctz and others): Change the
32959 preprocessor macro that controls conditional compilation from
32960 _ARCH_PWR9 to __POWER9_VECTOR__.
32961 (vec_all_ne): Change parameterization of __altivec_scalar_pred
32962 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
32963 control (instead of _ARCH_PWR9 control) so that template
32964 definition uses power9-specific function.
32965 (vec_any_eq): Likewise.
32966 (vec_all_ne): Change macro definition to use a power9-specific
32967 expansion under #ifdef __POWER9_VECTOR__ control (instead of
32968 _ARCH_PWR9 control).
32969 (vec_any_eq) Likewise.
32970 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
32971 expansion for CMPNEF to remove support for xvcmpnesp instruction.
32972 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
32973 support for xvcmpnedp instruction.
32974 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
32975 macro expansion so that Power9 implementation of vec_all_ne does
32976 not use the AltiVec predicate framework.
32977 (VCMPNEH_P): Likewise.
32978 (VCMPNEW_P): Likewise.
32979 (VCMPNED_P): Likewise.
32980 (VCMPNEFP_P): Likewise.
32981 (VCMPNEDP_P): Likewise.
32982 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
32983 implementation of vec_any_eq to not use AltiVec predicate
32985 (VCMPAEH_P): Likewise.
32986 (VCMPAEW_P): Likewise.
32987 (VCMPAED_P): Likewise.
32988 (VCMPAEFP_P): Likewise.
32989 (VCMPAEDP_P): Likewise.
32990 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
32991 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
32992 not use the AltiVec predicate framework.
32993 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
32994 of vec_any_eq to not use AltiVec predicate framework.
32995 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32996 support for predefined __POWER9_VECTOR__ macro to indicate that
32997 Power9 instruction selection is enabled.
32998 (altivec_overloaded_builtins): Remove extraneous
32999 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
33000 function argument types RS6000_BTI_bool_V16QI and
33001 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
33002 entry for overloaded function argument types RS6000_BTI_bool_V4SI
33003 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
33004 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
33005 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
33006 Power9 for implementations of vec_cmpne. Change the signature for
33007 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
33008 (representing vec_all_ne) to remove the previously described first
33009 argument of type RS6000_BTI_INTSI, as this was an artifact of
33010 reliance on the AltiVec predicate framework, which is no longer
33011 used in the implementation of these functions. Add
33012 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
33013 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
33014 since, unlike the AltiVec predicate framework implementation, we
33015 do not share function descriptors between vec_alle and vec_anyeq.
33016 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
33017 set of modes that receive special treatment even when
33018 TARGET_P9_VECTOR is true. The special treatment emits code that
33019 does not depend on Power9 instructions.
33020 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
33021 define_expand to not rely on AltiVec predicate framework.
33022 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33024 (vector_ne_v2di_p): Change this define_expand to not rely on
33025 AltiVec predicate framework.
33026 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
33028 (vector_ne_<mode>_p): Change this define_expand to not rely on
33029 AltiVec predicate framework.
33030 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33032 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
33033 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
33034 define_insn pattern.
33035 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
33036 define_insn pattern because the xvcmpne<VSs>. instruction is not
33038 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
33039 instruction is not supported.
33041 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33043 * config/nvptx/nvptx.c: Include intl.h.
33045 2017-03-01 Martin Jambor <mjambor@suse.cz>
33048 * ipa-prop.h (ipa_bits): Removed field known.
33049 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
33050 to pointers. Adjusted their comments to warn about their sharing.
33051 (ipcp_transformation_summary): Change bits to a vector of pointers.
33052 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
33053 (ipa_get_ipa_bits_for_value): Declare.
33054 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33055 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33056 (ipa_bits_hash_table): Likewise.
33057 (ipa_vr_ggc_hash_traits): Likewise.
33058 (ipa_vr_hash_table): Likewise.
33059 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33060 being pointers and vr_known being removed.
33061 (ipa_set_jf_unknown): Likewise.
33062 (ipa_get_ipa_bits_for_value): New function.
33063 (ipa_set_jfunc_bits): Likewise.
33064 (ipa_get_value_range): New overloaded functions.
33065 (ipa_set_jfunc_vr): Likewise.
33066 (ipa_compute_jump_functions_for_edge): Use the above functions to
33067 construct bits and vr parts of jump functions.
33068 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33069 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33071 (ipcp_grow_transformations_if_necessary): Also allocate
33072 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33074 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33075 them. Fix too long lines.
33076 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33077 vr_known being removed.
33078 (ipa_read_jump_function): Use new setter functions to construct bits
33079 and vr parts of jump functions or set them to NULL.
33080 (write_ipcp_transformation_info): Adjust for bits being pointers.
33081 (read_ipcp_transformation_info): Likewise.
33082 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33084 Include gt-ipa-prop.h.
33085 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33087 (ipcp_store_bits_results): Likewise.
33088 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33089 Do not write to existing jump functions but use a temporary instead.
33091 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33094 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33095 attempt to use its first operand as BIT_FIELD_REF base.
33097 2017-03-01 Richard Biener <rguenther@suse.de>
33099 PR middle-end/79721
33100 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33101 interpolating formula in wrapping arithmetic.
33102 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33104 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33106 PR tree-optimization/79734
33107 * tree-vect-generic.c (expand_vector_condition): Optimize
33108 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33109 Handle VEC_COND_EXPR where comparison has different inner width from
33110 type's inner width.
33112 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
33114 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
33115 markup, and similar issues. Remove @opindex entries for things
33116 that aren't options. Add missing -mmpy-option entries.
33118 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33120 PR tree-optimization/79737
33121 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
33122 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
33123 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
33124 instead of byte_size. Formatting fix.
33125 (shift_bytes_in_array_right): Formatting fix.
33127 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
33130 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
33131 condition on optimize for the leaf function test.
33133 2017-02-28 Martin Liska <mliska@suse.cz>
33136 * read-rtl-function.c (function_reader::handle_unknown_directive):
33137 Bail out when one uses -flto.
33139 2017-02-28 Martin Liska <mliska@suse.cz>
33141 * common.opt: Replace space with tabular for options of <number>
33143 * config/i386/i386.opt: Show <number> value for
33144 -mlarge-data-threshold.
33145 * opts.c (print_filtered_help): Do not display number in hexadecimal
33148 2017-02-28 Martin Liska <mliska@suse.cz>
33150 * common.opt: Fix --help=option -Q for options which are of
33153 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
33155 * config/i386/i386.c (print_reg): Error out for values
33156 of 8-bit size in invalid integer register.
33158 2017-02-28 Martin Sebor <msebor@redhat.com>
33160 PR tree-optimization/79691
33161 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
33163 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33166 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
33167 gcc_unreachable with output_operand_lossage.
33169 2017-02-28 Richard Biener <rguenther@suse.de>
33171 PR tree-optimization/79740
33172 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
33174 (visit_nary_op): Insert the nary into the hashtable if we
33175 pattern-matched sth.
33176 * tree-ssa-pre.c (eliminate_insert): Robustify.
33178 2017-02-28 Richard Biener <rguenther@suse.de>
33180 PR middle-end/79731
33181 * fold-const.c (decode_field_reference): Reject out-of-bound
33184 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33186 * config/i386/i386.c: Include intl.h.
33187 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
33188 instead of just cond ? "..." : "...".
33189 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
33190 * coverage.c (read_counts_file): Likewise.
33191 * omp-offload.c: Include intl.h.
33192 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
33193 of just cond ? "..." : "...".
33194 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
33195 of just cond ? "..." : "...".
33197 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
33200 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
33202 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
33203 'tune for' CPU name.
33204 * config/arm/arm-cpu-data.h: Regenerated.
33206 2017-02-28 Richard Biener <rguenther@suse.de>
33208 PR tree-optimization/79732
33209 * tree-inline.c (expand_call_inline): Do not shadow var.
33211 2017-02-28 Richard Biener <rguenther@suse.de>
33213 PR tree-optimization/79723
33214 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
33215 address-space properly.
33217 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
33219 * doc/optinfo.texi (Optimization groups): Fix option used for
33221 * doc/invoke.texi (-fopt-info): Document "omp".
33222 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
33223 (OPTGROUP_ALL): Add OPTGROUP_OMP.
33224 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
33225 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
33226 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
33228 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
33230 * dumpfile.c (optgroup_options): Instead of "openmp", associate
33231 OPTGROUP_OMP with "omp".
33233 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
33236 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
33237 for arithmetic shift of unsigned V2DI.
33239 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
33241 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
33242 arc/linux.h headers.
33243 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
33244 (LINK_SPEC): Likewise.
33245 (ARC_TLS_EXTRA_START_SPEC): Likewise.
33246 (EXTRA_SPECS): Likewise.
33247 (STARTFILE_SPEC): Likewise.
33248 (ENDFILE_SPEC): Likewise.
33249 (LIB_SPEC): Likewise.
33250 (TARGET_SDATA_DEFAULT): Likewise.
33251 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
33252 (MULTILIB_DEFAULTS): Likewise.
33253 (DWARF2_UNWIND_INFO): Likewise.
33254 * config/arc/big.h: New file.
33255 * config/arc/elf.h: Likewise.
33256 * config/arc/linux.h: Likewise.
33257 * config/arc/t-uClibc: Remove.
33259 2017-02-27 Bin Cheng <bin.cheng@arm.com>
33261 PR tree-optimization/77536
33262 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
33263 (tree_transform_and_unroll_loop): Use above function to compute the
33264 estimated niter of unrolled loop and use it when scaling profile.
33265 Also use count info rather than frequency if it's non-zero.
33266 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
33267 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
33268 (vect_transform_loop): Call above function.
33270 2017-02-27 Richard Biener <rguenther@suse.de>
33272 PR tree-optimization/45397
33273 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
33274 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
33275 (visit_nary_op): Add pattern matching for CSEing sign-changed
33276 or truncated operations with wider ones.
33278 2017-02-27 Richard Biener <rguenther@suse.de>
33280 PR tree-optimization/79690
33281 * tree-vect-stmts.c (vectorizable_store): Use vector type
33282 built from the DR with address-space.
33284 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
33286 * doc/invoke.texi (Optimize Options): Refine the description
33287 of asan-use-after-return.
33289 2017-02-25 Alan Modra <amodra@gmail.com>
33291 PR rtl-optimization/79584
33292 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
33293 base, not ad->base_term, the reg within base. Remove assertion
33294 that ad->base == ad->base_term. Replace gen_int_mode using
33295 bogus mode with const0_rtx.
33297 2017-02-25 Jakub Jelinek <jakub@redhat.com>
33299 PR middle-end/79396
33300 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
33301 FMA_EXPR like tcc_binary or tcc_unary.
33303 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
33306 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
33308 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
33309 (output_loc_operands): Handle DW_OP_call_ref and
33310 DW_OP_GNU_variable_value.
33311 (struct variable_value_struct): New type.
33312 (struct variable_value_hasher): Likewise.
33313 (variable_value_hash): New variable.
33314 (string_types): Remove.
33315 (copy_loc_descr): New function.
33316 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
33317 (prepend_loc_descr_to_each): New function.
33318 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
33319 instead of add_loc_descr_to_each if the first argument is single
33320 location list and the second has multiple.
33321 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
33322 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
33323 when looking for variable value which doesn't have other location info.
33324 (loc_list_from_tree): Formatting fix.
33325 (gen_array_type_die): Simplify DW_AT_string_length handling.
33326 (adjust_string_types): Remove.
33327 (gen_subprogram_die): Don't call adjust_string_types nor test/set
33328 string_types. Call resolve_variable_values.
33329 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
33330 (resolve_addr_in_expr): Likewise. Add A argument.
33331 (copy_deref_exprloc): Remove deref argument. Adjust for the
33332 original expression being DW_OP_GNU_variable_value with optionally
33333 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
33334 optionally after it.
33335 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
33336 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
33337 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
33338 (variable_value_hasher::hash, variable_value_hasher::equal): New
33340 (resolve_variable_value_in_expr, resolve_variable_value,
33341 resolve_variable_values, note_variable_value_in_expr,
33342 note_variable_value): New functions.
33343 (dwarf2out_early_finish): Call note_variable_value on all toplevel
33346 2017-02-24 Jakub Jelinek <jakub@redhat.com>
33349 * opts.h (handle_generated_option): Add GENERATED_P argument.
33350 * opts-common.c (handle_option): Adjust function comment.
33351 (handle_generated_option): Add GENERATED_P argument, pass it to
33353 (control_warning_option): Pass false to handle_generated_option
33355 * opts.c (maybe_default_option): Pass true to handle_generated_option
33357 * optc-gen.awk: Likewise.
33359 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33361 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
33362 a REG, look at the REG it is a SUBREG of.
33363 (splitter for cmpeqsi_t): Ditto.
33365 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33367 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
33368 the special USEs with the pattern of the insn, not the insn itself.
33370 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
33373 * doc/invoke.texi: Document -mload-store-pairs.
33375 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33376 Sandra Loosemore <sandra@codesourcery.com>
33378 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
33379 argument isn't a CONST_INT.
33380 (nios2_alternate_compare_const): Assert op is a CONST_INT.
33381 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
33382 (nios2_validate_compare): Bypass alternate compare logic if *op2
33383 is not a CONST_INT.
33384 (ldstwm_operation_p): Return false if first_base is not a REG or
33385 if first_offset is not a CONST_INT.
33387 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33389 * config/cris/cris.md: Use correct operand in a define_peephole2.
33391 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33393 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
33395 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33397 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
33398 this_insn if it is an INSN or JUMP_INSN.
33399 (force_offsettable): Look at base, not at addr.
33400 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
33401 on things that aren't necessarily CONST_INTs.
33403 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
33405 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
33406 -mfpmath=sse is the default also for x86-32 targets with SSE2
33407 instruction set when @option{-ffast-math} is enabled
33409 2017-02-24 Jeff Law <law@redhat.com>
33411 PR rtl-optimizatoin/79286
33412 * ira.c (update_equiv_regs): Drop may_trap_p exception to
33415 2017-02-24 Richard Biener <rguenther@suse.de>
33417 PR tree-optimization/79389
33418 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
33421 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
33423 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
33424 function comment to reflect reality.
33425 (loop_exits_before_overflow): Fix typo in function description.
33427 2017-02-24 Richard Biener <rguenther@suse.de>
33429 PR tree-optimization/79389
33430 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
33431 properly that a threading opportunity exists. Detect conditional
33432 copy/constant propagation opportunities.
33434 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
33436 * config/visium/visium.md (type): Add trap.
33437 (b): New mode attribute.
33438 (*btst): Rename into...
33439 (*btst<mode>): ...this and adjust.
33440 (*cbranchsi4_btst_insn): Rename into...
33441 (*cbranch<mode>4_btst_insn): ...this and adjust.
33442 (trap): New define_insn.
33444 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33446 PR tree-optimization/79389
33447 * ifcvt.c (struct noce_if_info): Add rev_cond field.
33448 (noce_reversed_cond_code): New function.
33449 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
33450 reversed_comparison_code. Formatting fix.
33451 (noce_try_store_flag): Test rev_cond != NULL in addition to
33452 reversed_comparison_code.
33453 (noce_try_store_flag_constants): Likewise.
33454 (noce_try_store_flag_mask): Likewise.
33455 (noce_try_addcc): Use rev_cond if non-NULL instead of
33456 reversed_comparison_code.
33457 (noce_try_cmove_arith): Likewise. Formatting fixes.
33458 (noce_try_minmax, noce_try_abs): Clear rev_cond.
33459 (noce_find_if_block): Initialize rev_cond.
33460 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
33461 instead of false as last argument never attempt to reverse it
33464 2017-02-23 Bin Cheng <bin.cheng@arm.com>
33466 PR tree-optimization/79663
33467 * tree-predcom.c (combine_chains): Process refs in reverse order
33468 only for ZERO length chains, and add explaining comment.
33470 2017-02-23 Jeff Law <law@redhat.com>
33472 PR tree-optimization/79578
33473 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
33474 in call to operand_equal_p.
33476 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
33479 * config/i386/cpuid.h: Fix another undefined behavior.
33481 2017-02-23 Richard Biener <rguenther@suse.de>
33483 PR tree-optimization/79683
33484 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
33485 vector types for data-refs.
33487 2017-02-23 Martin Liska <mliska@suse.cz>
33489 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
33491 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33493 PR middle-end/79665
33494 * internal-fn.c (get_range_pos_neg): Moved to ...
33495 * tree.c (get_range_pos_neg): ... here. No longer static.
33496 * tree.h (get_range_pos_neg): New prototype.
33497 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
33498 are known to be in between 0 and signed maximum inclusive, try to
33499 expand both unsigned and signed divmod and use the cheaper one from
33502 2017-02-22 Jeff Law <law@redhat.com>
33504 PR tree-optimization/79578
33505 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
33506 to compare base operands.
33508 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
33511 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
33512 gpc_reg_operand instead of fpr_reg_operand.
33514 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
33516 * config/mips/mips.c (mips_return_in_memory): Force FP
33517 vector types to be returned in memory for o32 ABI.
33519 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33521 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
33522 instead of DW_TAG_member for static data member declarations and don't
33523 set no_linkage_name for static inline data members.
33524 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
33527 2017-02-22 Martin Liska <mliska@suse.cz>
33529 * doc/invoke.texi: Replace inequality signs with square brackets
33532 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
33535 * lra-constraints.c (simplify_operand_subreg): Handle
33536 WORD_REGISTER_OPERATIONS targets.
33538 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33541 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
33542 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
33543 elimination by swapping fld*.
33545 2017-02-22 Richard Biener <rguenther@suse.de>
33547 PR tree-optimization/79673
33548 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
33549 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
33550 irrelevant address-space qualifiers and avoiding a
33551 ADDR_SPACE_CONVERT_EXPR from fold_convert.
33553 2017-02-22 Richard Biener <rguenther@suse.de>
33555 PR tree-optimization/79666
33556 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
33557 to not symbolically negate if that may introduce undefined
33560 2017-02-22 Martin Liska <mliska@suse.cz>
33563 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
33564 * data-streamer-out.c (streamer_write_gcov_count_stream):
33566 * value-prof.c (stream_out_histogram_value): Make assert more
33567 precise based on type of counter.
33569 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
33572 * config/i386/i386.md (standard_x87sse_constant_load splitter):
33573 Use nonimmediate_operand instead of memory_operand for operand 1.
33574 (float-extend standard_x87sse_constant_load splitter): Ditto.
33576 2017-02-21 Jeff Law <law@redhat.com>
33578 PR tree-optimization/79621
33579 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
33580 blocks with edges to themselves.
33582 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33585 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
33586 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
33587 Use gimple_call_builtin_p.
33590 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
33591 on temporarily removed DEBUG_INSNs.
33593 PR tree-optimization/79649
33594 * tree-loop-distribution.c (classify_partition): Give up on
33595 non-generic address space loads/stores.
33597 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
33599 * doc/loop.texi (Loop manipulation): Remove nonexistent
33600 tree_ssa_loop_version from the documentation.
33601 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
33603 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33606 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
33607 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
33608 * config/rs6000/rs6000.c: Include except.h.
33609 (rs6000_expand_split_stack_prologue): Call
33610 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
33612 2017-02-21 Martin Jambor <mjambor@suse.cz>
33615 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
33616 have been analyzed.
33618 2017-02-21 Martin Jambor <mjambor@suse.cz>
33620 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
33621 for backward compatibility only.
33622 * doc/invoke.texi (Option Summary): Remove all references to
33623 -fipa-cp-alignment.
33625 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
33629 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33631 * lra-constraints.c (curr_insn_transform): Handle
33632 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33634 2017-02-21 Martin Liska <mliska@suse.cz>
33636 * config/i386/i386.opt: Replace -masm-dialect with -masm.
33638 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
33640 PR translation/79638
33641 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
33643 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
33646 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
33647 (arm_function_ok_for_sibcall): Return false for an indirect call by
33648 descriptor if all the argument registers are used.
33649 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
33650 alignment of the function.
33652 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33654 PR tree-optimization/61441
33655 * simplify-rtx.c (simplify_const_unary_operation): For
33656 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
33657 the sNaN unmodified.
33659 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33661 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33662 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33663 instead of SYSTEM_HEADER_DIR.
33665 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
33666 Martin Liška <mliska@suse.cz>
33668 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
33669 Fix typos and grammar, use active voice, and clarify.
33671 2017-02-20 Marek Polacek <polacek@redhat.com>
33673 PR middle-end/79537
33674 * gimplify.c (gimplify_expr): Handle unused *&&L;.
33677 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
33679 2017-02-20 Jakub Jelinek <jakub@redhat.com>
33682 * config/i386/i386.c (ix86_expand_builtin): Handle
33683 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
33684 ix86_builtins_isa[fcode].isa as a requirement of those
33685 flags and any other flag in the bitmask.
33686 (ix86_init_mmx_sse_builtins): Use 0 instead of
33687 ~OPTION_MASK_ISA_64BIT as mask.
33688 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
33689 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
33690 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
33691 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
33693 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33696 * lra-constraints.c (split_reg): Check requested split mode
33697 is supported by the register.
33699 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33701 * lra-constraints.c (simplify_operand_subreg): Remove early
33704 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33707 * lra-constraints.c (curr_insn_transform): Tighten condition
33708 for converting SUBREG reloads from OP_OUT to OP_INOUT.
33710 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33713 * lra-constraints.c (curr_insn_transform): Handle
33714 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33716 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
33719 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
33721 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
33723 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
33726 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
33729 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33731 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
33734 2017-02-18 Jakub Jelinek <jakub@redhat.com>
33736 * final.c (last_columnnum, override_columnnum): New variables.
33737 (final_start_function): Set last_columnnum, pass it to begin_prologue
33738 hook and pass 0 to dwarf2out_begin_prologue.
33739 (final_scan_insn): Update override_columnnum. Pass last_columnnum
33740 to source_line debug hook.
33741 (notice_source_line): Compute last_columnnum and for debug_column_info
33742 return true on column changes.
33743 * debug.h (struct gcc_debug_hooks): Add column argument to
33744 source_line and begin_prologue hooks.
33745 (debug_nothing_int_charstar_int_bool): Remove prototype.
33746 (debug_nothing_int_int_charstar,
33747 debug_nothing_int_int_charstar_int_bool): New prototypes.
33748 (dwarf2out_begin_prologue): Add column argument.
33749 * debug.c (do_nothing_debug_hooks): Adjust source_line and
33750 begin_prologue hooks.
33751 (debug_nothing_int_charstar_int_bool): Remove.
33752 (debug_nothing_int_int_charstar,
33753 debug_nothing_int_int_charstar_int_bool): New functions.
33754 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
33755 through to dwarf2out_source_line.
33756 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
33757 (dwarf2out_source_line): Add column argument, emit it if requested.
33758 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
33760 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33761 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33762 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
33763 through to dwarf2out_begin_prologue.
33764 (vmsdbgout_source_line): Add column argument, pass it through to
33765 dwarf2out_source_line.
33766 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
33767 dbxout_source_line caller.
33768 (dbxout_source_line): Add column argument.
33770 * common.opt (gno-column-info, gcolumn-info): New options.
33771 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
33772 (check_die): Also test for multiple DW_AT_decl_column attributes.
33773 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
33774 DW_AT_decl_column if requested.
33775 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
33777 (gen_variable_die): Likewise.
33778 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
33779 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
33782 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
33783 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
33784 (ix86_handle_option): Handle OPT_m3dnowa.
33785 * doc/invoke.texi (-m3dnowa): Document.
33786 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
33787 -m3dnowa instead of -m3dnow -march=athlon.
33790 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
33791 instead of gcc_assert for K, r and R code checks. Formatting fixes.
33793 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33796 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
33797 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
33798 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
33799 generator for vsx_xxpermdi_<mode>_be.
33800 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
33801 force big-endian semantics.
33802 (vsx_xxpermdi_<mode>_be): New define_expand with same
33803 implementation as previous version of vsx_xxpermdi_<mode>.
33805 2017-02-17 Jakub Jelinek <jakub@redhat.com>
33807 PR tree-optimization/79327
33808 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
33809 variable, its initialization and use.
33811 2017-02-17 Julia Koval <julia.koval@intel.com>
33813 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
33814 (OPTION_MASK_ISA_PKU_UNSET): New.
33815 (ix86_handle_option): Handle -mrdpid.
33816 * config/i386/cpuid.h (bit_RDPID): New.
33817 * config/i386/driver-i386.c (host_detect_local_cpu):
33818 Detect RDPID feature.
33819 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
33820 * config/i386/i386-c.c (ix86_target_macros_internal):
33822 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
33823 (ix86_valid_target_attribute_inner_p): Add "rdpid".
33824 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
33825 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
33826 * config/i386/i386.md (define_insn "rdpid"): New.
33827 * config/i386/i386.opt Add -mrdpid.
33828 * config/i386/immintrin.h (_rdpid_u32): New.
33830 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
33832 PR rtl-optimization/79541
33833 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
33834 instead of transforming it into USE.
33836 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
33838 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
33839 If HONOR_SNANS (SFmode) force the input to a register.
33840 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
33841 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
33842 an frsp or similar insn.
33844 2017-02-17 Martin Liska <mliska@suse.cz>
33846 PR rtl-optimization/79577
33847 * params.def (selsched-max-sched-times): Increase minimum to 1.
33849 2017-02-17 Martin Liska <mliska@suse.cz>
33851 PR rtl-optimization/79574
33852 * gcse.c (want_to_gcse_p): Prevent integer overflow.
33854 2017-02-17 Martin Liska <mliska@suse.cz>
33856 PR tree-optimization/79529
33857 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
33858 ssa_defined_default_def_p to handle cases which are implicitly
33860 * tree-ssa.c (ssa_defined_default_def_p): New function.
33861 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
33862 which are implicitly defined.
33863 * tree-ssa.h (ssa_defined_default_def_p): Declare.
33865 2017-02-17 Richard Biener <rguenther@suse.de>
33867 PR middle-end/79576
33868 * params.def (max-ssa-name-query-depth): Limit to 10.
33870 2017-02-17 Richard Biener <rguenther@suse.de>
33872 PR tree-optimization/79552
33873 * tree-ssa-structalias.c (visit_loadstore): Properly verify
33876 2017-02-17 Richard Biener <rguenther@suse.de>
33879 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
33881 2017-02-17 Marek Polacek <polacek@redhat.com>
33883 PR middle-end/79536
33884 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
33885 (fold_negate_expr): New wrapper.
33887 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
33889 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
33890 Correct terminology and de-emphasize pre-standard behavior.
33892 2017-02-16 Alan Modra <amodra@gmail.com>
33894 PR rtl-optimization/79286
33895 * ira.c (def_dominates_uses): New function.
33896 (update_equiv_regs): Don't create an equivalence for insns that
33897 may trap where the register def does not dominate the use.
33899 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
33901 PR rtl-optimization/78127
33902 * lra.c (lra): Call lra_eliminate before finish the loop after
33905 2017-02-16 Richard Biener <rguenther@suse.de>
33907 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
33909 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
33910 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
33911 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
33912 (isl_val_int_from_wi): New function.
33913 (extract_affine_gmp): Rename to ...
33914 (extract_affine_wi): ... this, take a widest_int.
33915 (extract_affine_int): Just wrap extract_affine_wi.
33916 (add_param_constraints): Use isl_val_int_from_wi.
33917 (add_loop_constraints): Likewise, and extract_affine_wi.
33919 2017-02-15 Jeff Law <law@redhat.com>
33921 PR middle-end/79521
33922 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
33923 ira_init_register_move_cost_if_necessary.
33925 2017-02-15 Martin Sebor <msebor@redhat.com>
33927 PR middle-end/32003
33928 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
33929 removed in a prior commit.
33931 2017-02-15 Bin Cheng <bin.cheng@arm.com>
33933 PR tree-optimization/79347
33934 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
33935 counters during peeling.
33937 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
33939 * Makefile.in (site.exp): Remove "set ISLVER".
33941 2017-02-15 Jakub Jelinek <jakub@redhat.com>
33944 * real.c (real_from_integer): Call real_convert even for decimal.
33946 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
33949 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
33951 2017-02-14 Andrew Pinski <apinski@cavium.com>
33953 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
33954 cores and change the partno/implementer to be correct.
33955 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
33956 the 'B" as the implementer.
33957 * config/aarch64/aarch64-tune.md: Regenerate.
33959 2017-02-14 Carl Love <cel@us.ibm.com>
33961 * config/rs6000/rs6000.c: Add case statement entry to make the
33962 xvcvuxdsp built-in argument unsigned.
33963 * config/rs6000/vsx.md: Fix the source and return operand types so they
33964 match the instruction definitions from the ISA document. Fix typo
33965 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
33968 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
33971 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
33972 member early_clobber_alts.
33973 * lra-lives.c (reg_early_clobber_p): New.
33974 (process_bb_lives): Use it.
33975 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
33976 (debug_operand_data): Initialize early_clobber_alts.
33977 (setup_operand_alternative): Set up early_clobber_alts.
33978 (collect_non_operand_hard_regs): Ditto. Pass early clobber
33979 alternatives to new_insn_reg.
33980 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
33982 (lra_update_insn_regno_info): Pass the new arg.
33984 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33986 PR middle-end/79505
33987 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
33988 (new_oacc_loop_raw): Don't clear already cleared fields.
33991 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
33992 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
33993 _mm512_prefetch_i64gather_ps): New inline functions and macros.
33995 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
33998 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
34000 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
34003 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
34004 the extra instruction to the right place to store 128-bit constant
34007 2017-02-14 Martin Sebor <msebor@redhat.com>
34009 PR middle-end/79448
34010 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
34011 warning for strings of unknown length.
34013 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
34015 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
34017 2017-02-14 Jeff Law <law@redhat.com>
34020 * ira-costs.c (scan_one_insn): Initialize register move costs
34021 for pseudos seen in USE/CLOBBER insns.
34023 PR tree-optimization/79095
34024 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
34025 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
34026 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
34027 if the operands are known to be not equal, then the resulting range
34029 (intersect_ranges): If the new range is ~[0,0] and the old range is
34030 wide, then prefer ~[0,0].
34031 * tree-vrp.c (overflow_comparison_p_1): New function.
34032 (overflow_comparison_p): New function.
34033 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
34034 if NAME is used in an overflow test.
34035 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
34036 overflow check that can be expressed as an equality test, then adjust
34037 ops to be that equality test.
34039 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34041 * config/s390/s390-builtin-types.def: Remove flags argument.
34042 * config/s390/s390.c (s390_init_builtins): Likewise.
34044 2017-02-14 Martin Liska <mliska@suse.cz>
34046 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
34047 vector. Fix trailing white spaces.
34049 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
34051 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34054 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34056 PR rtl-optimization/68664
34057 * config/arm/arm.c (arm_sched_can_speculate_insn):
34058 New function. Declare prototype.
34059 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34061 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34063 PR rtl-optimization/68664
34064 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34066 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34068 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34070 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34071 max skip bytes for function, loop and jump.
34073 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34075 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34076 ABS_EXPR for gimple dump.
34078 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34081 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34083 PR tree-optimization/79408
34084 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34085 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34086 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34087 also if rhs1 is INTEGER_CST.
34089 2017-02-14 Richard Biener <rguenther@suse.de>
34091 PR middle-end/79432
34092 * tree-into-ssa.c (insert_phi_nodes): When the function can
34093 have abnormal edges rewrite SSA names with broken use-def
34094 dominance out of SSA and register them for PHI insertion.
34096 2017-02-13 Martin Sebor <msebor@redhat.com>
34098 PR middle-end/79496
34099 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34100 clearing info.nowrite flag when snprintf size argument is a range.
34102 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34104 * cprop.c (cprop_jump): Add missing space in string literal.
34105 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34106 (get_constraint_for_component_ref): Likewise.
34107 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34108 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34109 * lra-constraints.c (process_alt_operands): Likewise.
34110 * ipa-inline.c (inline_small_functions): Likewise.
34111 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34112 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
34113 * trans-mem.c (diagnose_tm_1_op): Likewise.
34114 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
34115 (grid_parallel_clauses_gridifiable): Likewise.
34117 * config/nvptx/mkoffload.c (process): Add space in between
34120 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
34121 "MOD4_SSE_REGS" and "ALL_REGS".
34123 * spellcheck.c (test_data): Add , in between "foo" and "food".
34125 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34128 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
34129 boundary crossing check and subsequent code generation agree.
34131 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34133 * config/aarch64/aarch64.c (has_memory_op): Delete.
34134 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
34137 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34139 PR rtl-optimization/79388
34140 PR rtl-optimization/79450
34141 * combine.c (distribute_notes): When removing TEM_INSN for which
34142 corresponding dest has last value recorded, invalidate that last
34145 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34147 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
34148 of explicit '@'. Add missing assembly comment marker on branch costs
34151 2017-02-13 Nathan Sidwell <nathan@acm.org>
34153 * gengtype-lex.l (<in_struct>): Add '/'.
34155 2017-02-13 Martin Liska <mliska@suse.cz>
34158 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
34160 2017-02-13 Richard Biener <rguenther@suse.de>
34162 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
34164 * configure: Re-generate.
34165 * config.in: Likewise.
34166 * graphite-dependences.c: Simplify as if
34167 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
34168 * graphite-isl-ast-to-gimple.c: Likewise.
34169 * graphite-optimize-isl.c: Likewise.
34170 * graphite-poly.c: Likewise.
34171 * graphite-sese-to-poly.c: Likewise.
34172 * graphite.h: Likewise.
34173 * toplev.c: Include isl/version.h and use isl_version () for
34174 printing the ISL version.
34175 * doc/install.texi: Update ISL requirement.
34177 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34179 * doc/standards.texi (Standards): Update reference to
34182 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34184 * doc/extend.texi (Named Address Spaces): sourceware.org now
34186 * doc/install.texi (Binaries): Ditto.
34189 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34191 * doc/cpp.texi: Replace "stringify"/"stringification" with C
34192 standard terminology "stringize"/"stringizing" throughout.
34193 * doc/cppinternals.texi: Likewise.
34195 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34197 * doc/extend.texi: Fix some spelling mistakes and typos.
34198 * doc/invoke.texi: Likewise.
34200 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34203 * params.def (inline-min-speedup) Change from 10 to 8.
34205 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34207 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
34210 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34213 * ipa-inline-analysis.c (get_minimal_bb): New function.
34214 (record_modified): Use it.
34215 (remap_edge_change_prob): Handle also ancestor functions.
34217 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
34219 * doc/contrib.texi (Contributors): Remove broken link into
34220 the Mauve CVS repository.
34222 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34224 PR middle-end/79454
34225 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
34226 result computation whenever lhs doesn't have vector mode, not
34227 just when it has BLKmode.
34229 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34231 * doc/makefile.texi (profiledbootstrap): Refer to the
34232 installation instructions only in textual form.
34234 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34237 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
34239 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34241 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
34242 (Specific): Update mingw-w64 reference.
34244 (Specific): Remove broken link to Renesas RX processor.
34246 2017-02-10 Richard Biener <rguenther@suse.de>
34248 * toplev.c (process_options): Do not mention obsolete graphite
34249 options when printing sorry message about missing graphite support.
34250 Mention -floop-nest-optimize.
34252 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
34254 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
34255 (vtst_p16): Likewise.
34256 (vtstq_p8): Likewise.
34257 (vtstq_p16): Likewise.
34259 (vtstq_p64): Likewise.
34260 * config/arm/arm_neon.h (vgetq_lane_p64): New.
34261 (vset_lane_p64): New.
34262 (vsetq_lane_p64): New.
34264 2017-02-10 Jakub Jelinek <jakub@redhat.com>
34266 PR tree-optimization/79411
34267 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
34268 stmt operands are SSA_NAMEs used in abnormal phis.
34269 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
34272 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34275 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
34278 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34280 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
34282 2017-02-09 Jakub Jelinek <jakub@redhat.com>
34284 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
34288 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
34289 not arbitrary TREE_CONSTANT.
34292 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
34293 "omp declare target link" attribute unless is_global_var.
34294 * omp-offload.c (find_link_var_op): Likewise.
34296 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
34297 Chung-Lin Tang <cltang@codesourcery.com>
34299 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
34301 (gimplify_adjust_omp_clauses): Don't delete TILE.
34302 (gimplify_omp_for): Deal with TILE.
34303 * internal-fn.c (expand_GOACC_TILE): New function.
34304 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
34306 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
34307 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
34309 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
34310 avoid DIV for outermost collapse var.
34311 (expand_oacc_for): Insert tile element loop as needed. Adjust.
34312 Remove out of date comments, fix whitespace.
34313 * omp-general.c (omp_extract_for_data): Deal with tiling.
34314 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
34315 adjust OLF_DIM_BASE value.
34316 (struct omp_for_data): Add tiling field.
34317 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
34318 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
34319 for auto loops. Remove default auto determining, moved to
34320 oacc_loop_fixed_partitions.
34321 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
34322 stmts, add e_mask field.
34323 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
34324 (oacc_thread_numbers): Use oacc_dim_call.
34325 (oacc_xform_tile): New.
34326 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
34327 (finish_oacc_loop): Adjust for ifns vector.
34328 (oacc_loop_discover_walk): Append loop abstraction sites to list,
34329 add case for GOACC_TILE fns.
34330 (oacc_loop_xform_loop): Delete.
34331 (oacc_loop_process): Iterate over call list directly, and add
34332 handling for GOACC_TILE fns.
34333 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
34335 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
34336 vector partitioning to outer loops. Assign 2 partitions to loops
34337 when available. Add TILE handling.
34338 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
34339 (execite_oacc_device_lower): Process GOACC_TILE fns,
34340 ignore unknown specs.
34341 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
34342 * tree.c (omp_clause_num_ops): Adjust TILE ops.
34343 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
34345 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
34347 * configure.ac (ACX_BUGURL): Update.
34348 * configure: Regenerate.
34350 2017-02-09 Richard Biener <rguenther@suse.de>
34352 PR tree-optimization/69823
34353 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
34354 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
34356 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34358 * config/arc/arc-c.def: Add __NPS400__ definition.
34359 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
34360 (TARGET_NPS400): Define.
34362 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34364 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
34366 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
34367 pointer, arch_info.
34368 (arc_cpu_types): Fill the arch_info field with a pointer into the
34369 arc_arch_types table.
34370 (arc_selected_cpu): Declare.
34371 * config/arc/arc.c (arc_selected_cpu): Make global.
34372 (arc_selected_arch): Delete.
34373 (arc_base_cpu): Delete.
34374 (arc_override_options): Remove references to deleted variables,
34375 update access to arch information.
34376 (ARC_OPT): Update access to arch information.
34377 (ARC_OPTX): Likewise.
34378 * config/arc/arc.h (arc_base_cpu): Remove declaration.
34379 (TARGET_ARC600): Update access to arch information.
34380 (TARGET_ARC601): Likewise.
34381 (TARGET_ARC700): Likewise.
34382 (TARGET_EM): Likewise.
34383 (TARGET_HS): Likewise.
34384 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
34387 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
34390 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
34391 condition/operands for integer GE/LE/GEU/LEU operations.
34393 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
34395 PR translation/79397
34396 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
34399 2017-02-08 Martin Jambor <mjambor@suse.cz>
34402 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
34403 whether allocation happened.
34404 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
34405 nothing was allocated.
34407 2017-02-08 Jakub Jelinek <jakub@redhat.com>
34409 PR tree-optimization/79408
34410 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
34411 constant, but SSA_NAME with a known integer range, use the minimum
34412 of that range instead of op1 to determine if modulo can be replaced
34413 with its first operand.
34415 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34417 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
34419 2017-02-08 Richard Biener <rguenther@suse.de>
34421 PR tree-optimization/71824
34422 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34423 Check all loops contained in the merged region.
34425 2017-02-07 Andrew Pinski <apinski@cavium.com>
34427 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
34429 2017-02-07 Andrew Pinski <apinski@cavium.com>
34431 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
34432 (thunderxt88): Likewise.
34433 (thunderxt81): Disable LSE and change v8.1 to v8.
34434 (thunderxt83): Likewise.
34436 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34437 Richard Biener <rguenther@suse.de>
34439 PR middle-end/79399
34440 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
34441 type from int to size_t.
34442 * ira-costs.c (struct_costs_size): Change type from int to size_t.
34444 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34446 PR rtl-optimization/79386
34447 * cprop.c (bypass_conditional_jumps): Initialize
34448 bypass_last_basic_block already before splitting bbs after
34449 unconditional traps...
34450 (bypass_conditional_jumps): ... rather than here.
34453 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
34454 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
34455 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
34456 fix -masm=intel patterns.
34458 2017-02-07 Richard Biener <rguenther@suse.de>
34460 PR tree-optimization/79256
34461 PR middle-end/79278
34462 * builtins.c (get_object_alignment_2): Use min_align_of_type
34463 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
34464 and ADJUST_FIELD_ALIGN.
34466 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
34468 * doc/tm.texi: Regenerate.
34469 * stor-layout.c (layout_decl): Adjust.
34470 (update_alignment_for_field): Likewise.
34471 (place_field): Likewise.
34472 (min_align_of_type): Likewise.
34473 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
34474 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
34475 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
34476 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
34477 * config/frv/frv.c (frv_adjust_field_align): Likewise.
34478 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
34479 * config/i386/i386.c (x86_field_alignment): Likewise.
34480 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
34481 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
34482 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
34483 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
34484 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
34485 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
34489 2017-01-30 Richard Biener <rguenther@suse.de>
34491 PR tree-optimization/79256
34492 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34493 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34496 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
34498 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
34499 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
34500 builtins to SImode and emit a zero-extend, if necessary.
34502 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34504 * docs/invoke.texi (RISC-V Options): Alphabetize.
34506 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34508 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
34511 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34513 * config/riscv/riscv.c: New file.
34514 * common/config/riscv/riscv-common.c: Likewise.
34515 * config.gcc: Likewise.
34516 * config/riscv/constraints.md: Likewise.
34517 * config/riscv/elf.h: Likewise.
34518 * config/riscv/generic.md: Likewise.
34519 * config/riscv/linux.h: Likewise.
34520 * config/riscv/multilib-generator: Likewise.
34521 * config/riscv/peephole.md: Likewise.
34522 * config/riscv/pic.md: Likewise.
34523 * config/riscv/predicates.md: Likewise.
34524 * config/riscv/riscv-builtins.c: Likewise.
34525 * config/riscv/riscv-c.c: Likewise.
34526 * config/riscv/riscv-ftypes.def: Likewise.
34527 * config/riscv/riscv-modes.def: Likewise.
34528 * config/riscv/riscv-opts.h: Likewise.
34529 * config/riscv/riscv-protos.h: Likewise.
34530 * config/riscv/riscv.h: Likewise.
34531 * config/riscv/riscv.md: Likewise.
34532 * config/riscv/riscv.opt: Likewise.
34533 * config/riscv/sync.md: Likewise.
34534 * config/riscv/t-elf-multilib: Likewise.
34535 * config/riscv/t-linux: Likewise.
34536 * config/riscv/t-linux-multilib: Likewise.
34537 * config/riscv/t-riscv: Likewise.
34538 * configure.ac: Likewise.
34539 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
34540 Waterman as RISC-V maintainers.
34541 * doc/install.texi: Add RISC-V entries.
34542 * doc/invoke.texi: Add RISC-V options section.
34543 * doc/md.texi: Add RISC-V constraints section.
34544 * configure: Regenerated.
34546 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
34549 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
34550 false values to be constant vectors with all 0 or all 1 bits set.
34551 (vcondu<mode><mode>): Likewise.
34552 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
34554 (fpmask_comparison_operator): Update comment.
34555 (vecint_comparison_operator): New predicate.
34556 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
34557 vector conditionals when the true and false values are constant
34558 vectors with all 0 bits or all 1 bits set.
34560 2017-02-06 Martin Sebor <msebor@redhat.com>
34562 PR tree-optimization/79376
34563 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
34565 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
34567 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
34568 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
34569 to simplify split condition.
34571 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34573 * omp-expand.c (oxpand_omp_atomic_fetch_op,
34574 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
34577 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
34579 PR rtl-optimization/68664
34580 * target.def (can_speculate_insn): New hook.
34581 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
34582 * doc/tm.texi: Regenerate.
34583 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
34584 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
34585 (rs6000_sched_can_speculate_insn): New function.
34587 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34589 PR tree-optimization/79284
34590 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
34591 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
34592 vectorizable_mask_load_store, vectorizable_operation,
34593 vect_is_simple_cond, get_same_sized_vectype): Use it instead
34594 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
34595 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
34596 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
34597 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
34598 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
34599 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
34600 is_gimple_assign (stmt). Replace another such test with
34601 is_gimple_assign (stmt).
34603 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34606 * config/avr/avr.c (rtl-iter.h): Include it.
34607 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
34608 (avr_legitimate_combined_insn): ...and implementation.
34610 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
34612 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
34613 * config/s390/s390.c (s390_const_operand_ok)
34614 (s390_canonicalize_comparison, s390_extract_part)
34615 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
34616 (s390_contiguous_bitmask_p, s390_rtx_costs)
34617 (legitimize_pic_address): Likewise.
34618 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
34619 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
34620 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
34621 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
34622 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
34624 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34626 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
34627 REGNO($0) == REGNO($1).
34629 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34631 * config/s390/linux.h(SIZE_TYPE): Add comment.
34633 2017-02-06 Julian Brown <julian@codesourcery.com>
34634 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34635 Virendra Pathak <virendra.pathak@broadcom.com>
34637 * config/aarch64/aarch64-cores.def: Change the scheduler
34639 * config/aarch64/aarch64.md: Include thunderx2t99.md.
34640 * config/aarch64/thunderx2t99.md: New file.
34642 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34644 * doc/standards.texi (Go Language): Update link to language
34647 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
34649 * tree-eh.c (lower_resx): Sanitize profile.
34650 (cleanup_empty_eh_move_lp): Likewise.
34652 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
34655 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
34657 * cfgloopmanip.h (loop_version): Update prototype.
34658 * modulo-sched.c (sms_schedule): Update call of loop_version.
34659 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
34660 * tree-parloops.c (gen_parallel_loop): Likewise.
34661 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
34662 * tree-ssa-loop-split.c (split_loop): Likewise.
34663 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
34664 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
34666 2017-02-05 Martin Liska <mliska@suse.cz>
34669 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
34671 (print_operand_address): Initialize a struct to zero.
34673 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34675 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
34676 garbage collector only in textual form.
34678 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34680 * doc/extend.texi (x86 specific memory model extensions for
34681 transactional memory): Simplify a phrase.
34683 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
34686 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
34687 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
34688 (atomic_storedi_1): Likewise.
34690 2017-02-04 Jakub Jelinek <jakub@redhat.com>
34692 PR tree-optimization/79338
34693 * tree-parloops.c (gather_scalar_reductions): Don't call
34694 vect_analyze_loop_form for loop->inner before destroying loop's
34697 2017-02-03 Martin Sebor <msebor@redhat.com>
34699 PR tree-optimization/79327
34700 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
34701 when precision has resulted in leading zeros.
34702 (format_integer): Adjust the likely counter to assume an unknown
34703 argument that may be zero is non-zero.
34705 2017-02-03 Jason Merrill <jason@redhat.com>
34708 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
34709 avoid copying non-taken branch.
34711 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34713 PR tree-optimization/79340
34714 * tree-vect-loop.c (vectorizable_reduction): Release
34715 vec_defs elements after safe_splicing them into other vectors.
34718 PR tree-optimization/79327
34719 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
34720 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
34722 (format_integer): Use wide_int_to_tree instead of build_int_cst
34723 + to_?hwi. If argmin is NULL, just set argmin and argmax to
34724 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
34725 of shortest and longest sequence.
34727 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
34729 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
34730 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
34732 2017-02-03 Walter Lee <walt@tilera.com>
34735 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
34736 after initial stackframe link reg save.
34737 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
34739 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34742 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
34743 wu for stxssp alternative.
34745 2017-02-03 Martin Sebor <msebor@redhat.com>
34747 PR tree-optimization/79352
34748 * gimple-fold.c (get_range_strlen): Add argument.
34749 (get_range_strlen): Change return type to bool.
34750 (get_maxval_strlen): Pass in a dummy argument.
34751 * gimple-fold.h (get_range_strlen): Change return type to bool.
34752 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
34753 * tree.h (array_at_struct_end_p): Add argument.
34754 * tree.c (array_at_struct_end_p): Handle it.
34756 2017-02-03 Martin Liska <mliska@suse.cz>
34759 * multiple_target.c (create_dispatcher_calls): Redirect edge
34760 from a caller of a dispatcher.
34761 (expand_target_clones): Make the clones local.
34762 (ipa_target_clone): Do both target clones and resolvers.
34763 (ipa_dispatcher_calls): Remove the pass.
34764 (pass_dispatcher_calls::gate): Likewise.
34765 (make_pass_dispatcher_calls): Likewise.
34766 * passes.def (pass_target_clone): Put as very first IPA early
34769 2017-02-03 Martin Liska <mliska@suse.cz>
34771 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
34772 in case of a function with ifunc attribute.
34774 2017-02-03 Martin Liska <mliska@suse.cz>
34776 * cgraph.c (cgraph_node::dump): Dump function version info.
34777 * symtab.c (symtab_node::dump_base): Add missing new line.
34779 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34781 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
34782 (ifcombine_ifandif): Use it.
34784 2017-02-03 Martin Liska <mliska@suse.cz>
34786 * doc/invoke.texi: Document default value for
34787 use-after-scope-direct-emission-threshold.
34789 2017-02-03 Martin Liska <mliska@suse.cz>
34791 PR tree-optimization/79339
34792 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
34793 (format_floating): Likewise.
34795 2017-02-03 Martin Liska <mliska@suse.cz>
34798 * ipa-prop.c (ipa_node_params_t::insert): Remove current
34800 (ipa_node_params_t::remove): Likewise.
34801 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
34802 initialization from removed ipa_node_params_t::insert.
34803 (ipa_node_params::~ipa_node_params): Move from removed
34804 ipa_node_params_t::release.
34805 * symbol-summary.h (symbol_summary::m_released): New member.
34806 Do not release a summary twice. Do not allow to call finalizer
34807 for types of a summary that live in GGC memory.
34809 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34811 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
34814 2017-02-02 Martin Sebor <msebor@redhat.com>
34816 PR middle-end/79275
34817 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
34818 (format_string): Tighten up the range of output for non-constant
34819 strings and correct the expected range for wide non-constant strings.
34821 2017-02-02 Martin Sebor <msebor@redhat.com>
34823 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
34825 PR middle-end/32003
34826 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
34828 (-fdump-tree-@var): Add to index and document how to come up
34829 with pass-specific option and dump file names.
34830 (-fdump-passes): Clarify where to look for output.
34832 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34834 PR middle-end/77445
34835 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
34836 statistics of the analyzed path; allow threading for speed when
34837 any of BBs along the path are optimized for speed.
34839 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
34841 PR middle-end/78468
34842 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
34843 settings of the virtual registers.
34846 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
34848 * explow.c (get_dynamic_stack_size): Take known alignment of stack
34849 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
34852 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34854 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
34855 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
34857 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34859 * config/s390/s390.md: Add missing comments with the expanded
34861 * config/s390/vector.md: Likewise.
34862 * config/s390/vx-builtins.md: Likewise.
34864 2017-02-02 Jakub Jelinek <jakub@redhat.com>
34867 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
34868 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
34869 conditions on a single line.
34871 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34873 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34874 __S390_VX__ to __VX__.
34876 2017-02-01 Andrew Pinski <apinski@cavium.com>
34878 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
34879 stmt_info to record_stmt_cost.
34880 (vect_get_known_peeling_cost): Pass stmt_info if known to
34882 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
34883 cpu_vector_cost field into
34884 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
34885 field into vec_int_stmt_cost and vec_fp_stmt_cost.
34886 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
34887 splitting of scalar_stmt_cost and vec_stmt_cost.
34888 (thunderx_vector_cost): Likewise.
34889 (cortexa57_vector_cost): LIkewise.
34890 (exynosm1_vector_cost): Likewise.
34891 (xgene1_vector_cost): Likewise.
34892 (thunderx2t99_vector_cost): Improve after the splitting of the two
34894 (aarch64_builtin_vectorization_cost): Update for the splitting of
34895 scalar_stmt_cost and vec_stmt_cost.
34897 2017-02-01 Torvald Riegel <triegel@redhat.com>
34898 Richard Henderson <rth@redhat.com>
34900 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
34901 conditional on existance of a fast atomic load.
34902 * optabs-query.c (can_atomic_load_p): New function.
34903 * optabs-query.h (can_atomic_load_p): Declare it.
34904 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
34905 no fast atomic load is available for the particular size of access.
34906 (expand_atomic_compare_and_swap): Likewise.
34907 (expand_atomic_load): Likewise.
34908 (expand_atomic_store): Likewise.
34909 (expand_atomic_fetch_op): Likewise.
34910 * testsuite/lib/target-supports.exp
34911 (check_effective_target_sync_int_128): Remove x86 because it provides
34912 no fast atomic load.
34913 (check_effective_target_sync_int_128_runtime): Likewise.
34915 2017-02-01 Richard Biener <rguenther@suse.de>
34917 * graphite.c: Include tree-vectorizer.h for find_loop_location.
34918 (graphite_transform_loops): Provide opt-info for optimized nests.
34919 * tree-parloop.c (parallelize_loops): Provide opt-info for
34920 parallelized loops.
34922 2017-02-01 Richard Biener <rguenther@suse.de>
34924 PR middle-end/79315
34925 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
34926 was not set before.
34928 2017-02-01 Richard Biener <rguenther@suse.de>
34930 PR tree-optimization/71824
34931 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34932 Verify the loops are valid in the merged SESE region.
34933 (scop_detection::can_represent_loop_1): Check analyzing the
34934 evolution of the number of iterations in the region succeeds.
34936 2017-01-31 Ian Lance Taylor <iant@golang.org>
34938 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
34939 REG_ARGS_SIZE note to 32-bit push insns and call insn.
34941 2017-01-31 David Malcolm <dmalcolm@redhat.com>
34943 PR preprocessor/79210
34944 * input.c (get_substring_ranges_for_loc): Replace line_width
34945 assertion with error-handling.
34947 2017-01-31 Richard Biener <rguenther@suse.de>
34949 PR tree-optimization/77318
34950 * graphite-sese-to-poly.c (extract_affine): Fix assert.
34951 (create_pw_aff_from_tree): Take loop parameter.
34952 (add_condition_to_pbb): Pass loop of the condition to
34953 create_pw_aff_from_tree.
34955 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34957 * config/s390/s390.c (s390_asan_shadow_offset): New function.
34958 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
34960 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
34964 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
34966 (convert_int_to_float128): Likewise.
34967 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
34968 (convert_int_to_float128): Likewise.
34969 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
34970 (UNSPEC_IEEE128_CONVERT): Likewise.
34971 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
34972 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
34973 Use local variables for IBM extended format.
34974 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
34975 (fix_trunc<mode>si2_fprs): Likewise.
34976 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
34977 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
34978 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
34979 to know that we can now have integers of all sizes in vector
34981 (fix<uns>_<mode>di2_hw): Likewise.
34982 (float<uns>_<mode>si2_hw): Likewise.
34983 (fix_<mode>si2_hw): Likewise.
34984 (fixuns_<mode>si2_hw): Likewise.
34985 (float<uns>_<mode>di2_hw): Likewise.
34986 (float_<mode>di2_hw): Likewise.
34987 (float_<mode>si2_hw): Likewise.
34988 (floatuns_<mode>di2_hw): Likewise.
34989 (floatuns_<mode>si2_hw): Likewise.
34990 (xscvqp<su>wz_<mode>): Delete, no longer used.
34991 (xscvqp<su>dz_<mode>): Likewise.
34992 (xscv<su>dqp_<mode>): Likewise.
34993 (ieee128_mfvsrd_64bit): Likewise.
34994 (ieee128_mfvsrd_32bit): Likewise.
34995 (ieee128_mfvsrwz): Likewise.
34996 (ieee128_mtvsrw): Likewise.
34997 (ieee128_mtvsrd_64bit): Likewise.
34998 (ieee128_mtvsrd_32bit): Likewise.
35000 2017-01-31 Martin Liska <mliska@suse.cz>
35003 * ipa-prop.c (ipa_free_all_node_params): Call release method
35004 instead of ~sumbol_summary to not to trigger double times
35007 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
35009 PR tree-optimization/71691
35010 * bitmap.h (class auto_bitmap): New.
35011 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
35012 is_maybe_undefined instead of ssa_undefined_value_p.
35014 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35016 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
35017 __S390_ARCH_LEVEL__ to __ARCH__.
35019 2017-01-31 Jakub Jelinek <jakub@redhat.com>
35021 PR tree-optimization/79267
35022 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
35023 if should_remove_lhs_p is true.
35025 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
35028 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
35029 (add_alignment_attribute): New.
35030 (base_type_die): Add alignment attribute.
35031 (subrange_type_die): Likewise.
35032 (modified_type_die): Likewise.
35033 (gen_array_type_die): Likewise.
35034 (gen_descr_array_type_die: Likewise.
35035 (gen_enumeration_type_die): Likewise.
35036 (gen_subprogram_die): Likewise.
35037 (gen_variable_die): Likewise.
35038 (gen_field_die): Likewise.
35039 (gen_ptr_to_mbr_type_die): Likewise.
35040 (gen_struct_or_union_type_die): Likewise.
35041 (gen_subroutine_type_die): Likewise.
35042 (gen_typedef_die): Likewise.
35043 (base_type_cmp): Compare alignment attribute.
35045 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35048 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
35049 (setb_unsigned) New pattern for setb with CCUNS.
35050 * config/rs6000/rs6000.c (expand_block_compare): Use a different
35051 subfc./subfe sequence to avoid overflow problems. Generate a
35052 shorter sequence with cmpld/setb for power9.
35053 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35054 for generating subfc. instruction.
35055 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35056 now uses this instruction.
35058 2017-01-30 Ian Lance Taylor <iant@google.com>
35061 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35062 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35064 2017-01-30 Martin Sebor <msebor@redhat.com>
35066 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35067 Move constant to the right of a relational operator.
35068 (get_mpfr_format_length, format_character, format_string): Ditto.
35069 (should_warn_p, maybe_warn): Same.
35071 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35073 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35076 * asan.c (get_translation_unit_decl): Remove function.
35077 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35079 2017-01-30 Martin Liska <mliska@suse.cz>
35081 PR gcov-profile/79259
35082 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35083 -fprofile-generate.
35085 2017-01-30 Martin Liska <mliska@suse.cz>
35088 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35089 Initialize variables with NULL value.
35091 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35094 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35096 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35098 2017-01-30 Richard Biener <rguenther@suse.de>
35100 PR tree-optimization/79276
35101 * tree-vrp.c (process_assert_insertions): Properly adjust common
35102 when removing a duplicate.
35104 2017-01-30 Richard Biener <rguenther@suse.de>
35106 PR tree-optimization/79256
35107 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35108 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35110 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35112 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
35115 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
35116 ("*r<noxa>sbg_<mode>_sll_bitmask")
35117 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
35118 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
35119 Use contiguous_bitmask_nowrap_operand.
35121 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35124 * config/rs6000/altivec.h (vec_xl): Revise #define.
35125 (vec_xst): Likewise.
35127 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
35129 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
35131 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
35133 PR rtl-optimization/79194
35134 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
35135 traps before call to bypass_conditional_jumps.
35137 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35139 PR tree-optimization/71374
35140 * lra-constraints.c (check_conflict_input_operands): New.
35141 (match_reload): Use it.
35143 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35146 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
35147 account to calculate conflict_set.
35149 2017-01-27 Bin Cheng <bin.cheng@arm.com>
35151 PR rtl-optimization/78559
35152 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
35153 other_insn in combine.
35155 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
35157 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
35158 uint16_type_node for BT_UINT16.
35160 2017-01-27 David Malcolm <dmalcolm@redhat.com>
35162 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
35163 "RTL Tests" to menu.
35164 (GIMPLE Tests): New node.
35165 (RTL Tests): New node.
35167 2017-01-27 Richard Biener <rguenther@suse.de>
35169 PR tree-optimization/79245
35170 * tree-loop-distribution.c (distribute_loop): Apply cost
35171 modeling also to detected patterns.
35173 2017-01-27 Richard Biener <rguenther@suse.de>
35175 PR tree-optimization/71433
35176 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
35177 (compare_assert_loc): New function.
35178 (process_assert_insertions): Sort and optimize assert locations
35179 to remove duplicates and push down identical assertions on
35180 edges to their destination block.
35182 2017-01-27 Richard Biener <rguenther@suse.de>
35184 PR tree-optimization/79244
35185 * tree-vrp.c (remove_range_assertions): Forcefully propagate
35186 out SSA names even if abnormal.
35188 2017-01-27 Jakub Jelinek <jakub@redhat.com>
35190 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
35191 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
35192 instead of MPFR_RNDN.
35194 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
35197 * arm.c (arm_option_override): Don't call build_target_option_node
35198 until after doing all option overrides.
35199 (arm_valid_target_attribute_tree): Likewise.
35201 2017-01-27 Martin Liska <mliska@suse.cz>
35203 * doc/invoke.texi (-fprofile-arcs): Document profiling support
35204 for {cd}tors and C++ {cd}tors.
35206 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
35208 * config/s390/s390.md ("*setmem_long_and")
35209 ("*setmem_long_and_31z"): Use zero_extend instead of and.
35211 2017-01-26 Martin Sebor <msebor@redhat.com>
35213 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
35216 2017-01-26 Martin Sebor <msebor@redhat.com>
35218 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
35219 HAVE_DFmode before using XFmode or DFmode.
35220 (parse_directive): Avoid using the z length modifier to avoid
35221 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
35223 PR middle-end/78703
35224 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
35225 to accept adjustment as an array.
35226 (get_int_range): New function.
35227 (struct directive): Make width and prec arrays.
35228 (directive::set_width, directive::set_precision): Call get_int_range.
35229 (format_integer, format_floating): Handle width and precision ranges.
35230 (format_string, parse_directive): Same.
35232 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35235 * dwarf2out.c (generate_skeleton_bottom_up): For children with
35236 comdat_type_p set, just clone them, but keep the children in the
35240 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
35241 which have direct callers with -fvar-tracking-assignments enabled
35243 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
35244 inside of type units.
35246 2017-01-26 Martin Sebor <msebor@redhat.com>
35248 PR middle-end/78703
35249 * gimple-ssa-sprintf.c (struct result_range): Add likely and
35251 (struct format_result): Replace number_chars, number_chars_min,
35252 and number_chars_max with a single member of struct result_range.
35254 (format_result::operator+=): Adjust.
35255 (struct fmtresult): Remove bounded. Handle likely and unlikely
35257 (fmtresult::adjust_for_width_or_precision): New function.
35258 (fmtresult:type_max_digits): New function.
35259 (bytes_remaining): Handle likely and unlikely counters.
35260 (min_bytes_remaining): Remove.
35261 (format_percent): Simplify.
35262 (format_integer, format_floating): Set likely and unlikely counters.
35263 (get_string_length, format_character, format_string): Same.
35264 (format_plain, should_warn_p): New function.
35265 (maybe_warn): Call should_warn_p. Update diagnostic messages
35266 and handle those for all directives, including plain strings.
35267 (format_directive): Handle likely and unlikely counters.
35268 Remove unnecessary quoting from diagnostics. Add an informational
35270 (add_bytes): Remove.
35271 (pass_sprintf_length::compute_format_length): Simplify.
35272 (try_substitute_return_value): Handle likely and unlikely counters.
35274 2017-01-26 Carl Love <cel@us.ibm.com>
35276 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
35277 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
35279 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
35282 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
35283 endianess for subregs into account.
35284 * lra-constraints.c (lra_constraints): Do risky transformations
35285 always on the first iteration.
35286 * lra-lives.c (check_pseudos_live_through_calls): Add arg
35287 last_call_used_reg_set.
35288 (process_bb_lives): Define and use last_call_used_reg_set.
35289 * lra.c (lra): Always continue after lra_constraints on the first
35292 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
35294 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
35296 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
35298 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35300 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
35301 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
35302 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
35303 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
35304 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
35305 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
35306 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
35307 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
35308 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
35310 2017-01-26 Marek Polacek <polacek@redhat.com>
35313 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
35314 for the third operand.
35316 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35318 PR middle-end/79236
35319 * omp-low.c (struct omp_context): Add simt_stmt field.
35320 (scan_omp_for): Return omp_context *.
35321 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
35322 context to the _simt_ SIMD stmt.
35323 (lower_omp_for): For combined SIMD with sibling _simt_
35324 SIMD, make sure to use the same decls in _looptemp_
35325 clauses as in the sibling.
35327 2017-01-26 David Sherwood <david.sherwood@arm.com>
35329 PR middle-end/79212
35330 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
35333 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35336 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
35337 emit fld b; fld a; if possible.
35339 * brig-builtins.def: Update copyright years.
35340 * config/arm/arm_acle_builtins.def: Update copyright years.
35342 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
35345 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
35346 constraint instead of o for the stxsd instruction.
35348 2017-01-25 Carl Love <cel@us.ibm.com>
35350 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
35351 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
35353 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
35355 * doc/invoke.texi (C++ Dialect Options): Fix typo.
35357 2017-01-25 Richard Biener <rguenther@suse.de>
35359 PR tree-optimization/69264
35360 * target.def (vector_alignment_reachable): Improve documentation.
35361 * doc/tm.texi: Regenerate.
35362 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
35364 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
35365 earlier changes with respect to TYPE_USER_ALIGN.
35366 (vector_alignment_reachable_p): Likewise. Improve dumping.
35368 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35371 * config/arm/arm.md (xordi3): Force constant operand into a register
35374 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35376 * doc/invoke.texi (-fstore-merging): Correct default optimization
35377 levels at which it is enabled.
35378 (-O): Move -fstore-merging from list to...
35381 2017-01-25 Richard Biener <rguenther@suse.de>
35384 * omp-expand.c: Include debug.h.
35385 (expand_omp_taskreg): Make sure to generate early debug before
35386 outlining anything from a function.
35387 (expand_omp_target): Likewise.
35388 (grid_expand_target_grid_body): Likewise.
35390 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
35393 * asan.c (get_translation_unit_decl): New function.
35394 (asan_add_global): Extract modules file name from globals
35395 TRANSLATION_UNIT_DECL name.
35397 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
35400 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
35401 for long calls with APCS frame and VFP.
35403 2017-01-24 David Malcolm <dmalcolm@redhat.com>
35405 * cfg.c (original_copy_tables_initialized_p): New function.
35406 * cfg.h (original_copy_tables_initialized_p): New decl.
35407 * cfgrtl.c (relink_block_chain): Guard the call to
35408 free_original_copy_tables with a call to
35409 original_copy_tables_initialized_p.
35410 * cgraph.h (symtab_node::native_rtl_p): New decl.
35411 * cgraphunit.c (symtab_node::native_rtl_p): New function.
35412 (symtab_node::needed_p): Don't assert for early assembly output
35413 for __RTL functions.
35414 (cgraph_node::finalize_function): Set "force_output" for __RTL
35416 (cgraph_node::analyze): Bail out early for __RTL functions.
35417 (analyze_functions): Update assertion to support __RTL functions.
35418 (cgraph_node::expand): Bail out early for __RTL functions.
35419 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
35421 * function.h (struct function): Update comment for field
35423 * gimple-expr.c: Include "tree-pass.h".
35424 (gimple_has_body_p): Return false for __RTL functions.
35425 * Makefile.in (OBJS): Add run-rtl-passes.o.
35426 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
35428 (gcc::pass_manager::get_clean_slate): New accessor.
35429 * passes.c: Include "insn-addr.h".
35430 (should_skip_pass_p): Add logging. Update logic for running
35431 "expand" to be compatible with both __GIMPLE and __RTL. Guard
35432 property-provider override so it is only done for gimple passes.
35434 (skip_pass): New function.
35435 (execute_one_pass): Call skip_pass when skipping passes.
35436 * read-md.c (md_reader::read_char): Support filtering
35437 the input to a subset of line numbers.
35438 (md_reader::md_reader): Initialize fields
35439 m_first_line and m_last_line.
35440 (md_reader::read_file_fragment): New function.
35441 * read-md.h (md_reader::read_file_fragment): New decl.
35442 (md_reader::m_first_line): New field.
35443 (md_reader::m_last_line): New field.
35444 * read-rtl-function.c (function_reader::create_function): Only
35445 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
35446 curr_properties. Set DECL_INITIAL to a dummy block.
35447 (read_rtl_function_body_from_file_range): New function.
35448 * read-rtl-function.h (read_rtl_function_body_from_file_range):
35450 * run-rtl-passes.c: New file.
35451 * run-rtl-passes.h: New file.
35453 2017-01-24 Jeff Law <law@redhat.com>
35455 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
35458 2017-01-24 Bin Cheng <bin.cheng@arm.com>
35460 PR tree-optimization/79159
35461 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
35462 (record_nonwrapping_iv): Improve boundary using above function if no
35463 value range information.
35465 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
35466 Martin Jambor <mjambor@suse.cz>
35468 * brig-builtins.def: New file.
35469 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
35470 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
35471 (DEF_HSAIL_SAT_BUILTIN): Likewise.
35472 (DEF_HSAIL_INTR_BUILTIN): Likewise.
35473 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
35474 * builtin-types.def (BT_INT8): New.
35475 (BT_INT16): Likewise.
35476 (BT_UINT8): Likewise.
35477 (BT_UINT16): Likewise.
35478 (BT_FN_ULONG): Likewise.
35479 (BT_FN_UINT_INT): Likewise.
35480 (BT_FN_UINT_ULONG): Likewise.
35481 (BT_FN_UINT_LONG): Likewise.
35482 (BT_FN_UINT_PTR): Likewise.
35483 (BT_FN_ULONG_PTR): Likewise.
35484 (BT_FN_INT8_FLOAT): Likewise.
35485 (BT_FN_INT16_FLOAT): Likewise.
35486 (BT_FN_UINT32_FLOAT): Likewise.
35487 (BT_FN_UINT16_FLOAT): Likewise.
35488 (BT_FN_UINT8_FLOAT): Likewise.
35489 (BT_FN_UINT64_FLOAT): Likewise.
35490 (BT_FN_UINT16_UINT32): Likewise.
35491 (BT_FN_UINT32_UINT16): Likewise.
35492 (BT_FN_UINT16_UINT16_UINT16): Likewise.
35493 (BT_FN_INT_PTR_INT): Likewise.
35494 (BT_FN_UINT_PTR_UINT): Likewise.
35495 (BT_FN_LONG_PTR_LONG): Likewise.
35496 (BT_FN_ULONG_PTR_ULONG): Likewise.
35497 (BT_FN_VOID_UINT64_UINT64): Likewise.
35498 (BT_FN_UINT8_UINT8_UINT8): Likewise.
35499 (BT_FN_INT8_INT8_INT8): Likewise.
35500 (BT_FN_INT16_INT16_INT16): Likewise.
35501 (BT_FN_INT_INT_INT): Likewise.
35502 (BT_FN_UINT_FLOAT_UINT): Likewise.
35503 (BT_FN_FLOAT_UINT_UINT): Likewise.
35504 (BT_FN_ULONG_UINT_UINT): Likewise.
35505 (BT_FN_ULONG_UINT_PTR): Likewise.
35506 (BT_FN_ULONG_ULONG_ULONG): Likewise.
35507 (BT_FN_UINT_UINT_UINT): Likewise.
35508 (BT_FN_VOID_UINT_PTR): Likewise.
35509 (BT_FN_UINT_UINT_PTR: Likewise.
35510 (BT_FN_UINT32_UINT64_PTR): Likewise.
35511 (BT_FN_INT_INT_UINT_UINT): Likewise.
35512 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
35513 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
35514 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
35515 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
35516 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
35517 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
35518 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
35519 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
35520 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
35521 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
35522 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
35523 * doc/frontends.texi: List BRIG FE.
35524 * doc/install.texi (Testing): Add BRIG tesring requirements.
35525 * doc/invoke.texi (Overall Options): Mention BRIG.
35526 * doc/standards.texi (Standards): Doucment BRIG HSA version.
35528 2017-01-24 Richard Biener <rguenther@suse.de>
35530 PR translation/79208
35531 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
35533 2017-01-24 Martin Jambor <mjambor@suse.cz>
35536 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
35537 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
35538 and known_contexts.
35540 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
35542 PR middle-end/79123
35543 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
35544 casts from signed to unsigned really don't have a range.
35546 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
35548 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
35549 GMP_RNDx for compatiblity.
35551 2017-01-24 Martin Liska <mliska@suse.cz>
35554 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
35555 that would prevent us to call alloca with -1 as argument.
35557 2017-01-24 Jakub Jelinek <jakub@redhat.com>
35559 * dwarf2out.c (output_compilation_unit_header, output_file_names):
35560 Avoid -Wformat-security warning.
35562 2017-01-23 Andrew Pinski <apinski@cavium.com>
35564 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
35567 2017-01-23 Martin Sebor <msebor@redhat.com>
35569 PR middle-end/78703
35570 * gimple-ssa-sprintf.c (warn_level): New global.
35571 (format_integer): Use it here and throughout the rest of the file.
35572 Use the same switch to compute sign as base.
35573 (maybe_warn): New function.
35574 (format_directive): Factor out warnings into maybe_warn.
35575 Add debugging output. Use warn_level.
35576 (add_bytes): Use warn_level.
35577 (pass_sprintf_length::compute_format_length): Add debugging output.
35578 (try_substitute_return_value): Same.
35579 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
35581 PR middle-end/78703
35582 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
35583 (struct fmtresult, format_integer, format_floating): Adjust.
35584 (fmtresult::fmtresult): Set max correctly in two argument ctor.
35585 (get_string_length, format_string,format_directive): Same.
35586 (pass_sprintf_length::compute_format_length): Same.
35587 (try_substitute_return_value): Simplify slightly.
35589 PR middle-end/78703
35590 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
35591 (fmtresult::operator+=): Outlined.
35592 (struct fmtresult): Add ctors.
35593 (struct conversion_spec): Rename...
35594 (struct directive): ...to this. Add and remove data members.
35595 (directive::set_width, directive::set_precision): New functions.
35596 (format_percent): Use fmtresult ctor.
35597 (get_width_and_precision): Remove.
35598 (format_integer): Make naming changes. Avoid computing width and
35600 (format_floating): Same. Adjust indentation.
35601 (format_character, format_none): New functions.
35602 (format_string): Moved character handling to format_character.
35603 (format_directive): Remove arguments, change return type.
35604 (parse_directive): New function.
35605 (pass_sprintf_length::compute_format_length): Move directive
35606 parsing to parse_directive.
35608 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35610 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
35611 (assign_assembler_name_if_needed): ... this.
35612 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
35613 (assign_assembler_name_if_needed): ... this.
35614 (free_lang_data_in_cgraph): Adjust callers.
35615 * cgraphunit.c (cgraph_node::analyze): Likewise.
35616 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
35619 2017-01-23 Richard Biener <rguenther@suse.de>
35621 PR tree-optimization/79088
35622 PR tree-optimization/79188
35623 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
35624 resetting loop bounds after last path deletion. Reset loop
35625 bounds of the target loop, make code match the comments.
35626 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
35627 Make sure loops need no fixups.
35629 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35631 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
35632 exponent support with double type for first argument.
35633 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
35634 type returned by __builtin_vec_extract_sig,
35635 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
35636 functions from "vector int" to "vector unsigned int" or from
35637 "vector long long int" to "vector unsigned long long int".
35638 Changed type returned by __builtin_vec_extract_exp,
35639 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
35640 functions from "vector int" to "vector unsigned int" or from
35641 "vector long long int" to "vector unsigned long long int".
35642 Changed return type of __builtin_vec_test_data_class,
35643 __builtin_vec_test_data_class_sp, and
35644 __builtin_vec_test_data_class_dp from "vector int" to
35645 "vector bool int" or from "vector long long int" to "vector bool
35646 long long int" and changed second argument type from "unsigned
35647 int" to "int". Added new overloaded function forms "vector float
35648 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
35649 "vector float __builtin_vec_insert_exp_sp (vector float, vector
35650 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
35651 double, vector unsigned long long int)" and "vector double
35652 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
35653 long int)". Changed return type of
35654 __builtin_scalar_test_data_class and
35655 __builtin_scalar_test_data_class_sp and
35656 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
35657 int" and changed second argument from "unsigned int" to "int".
35658 Changed type returned by __builtin_scalar_test_neg,
35659 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
35660 from "int" to "bool int". Added new overloaded function form
35661 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
35662 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
35663 exponent double-precision with floating point first argument.
35664 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
35665 documentation of scalar_test_data_class, scalar_test_neg,
35666 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
35667 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
35668 vec_test_data_class built-in functions to reflect refinements in
35669 their type signatures.
35671 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
35673 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
35675 (aarch64_elf_asm_destructor): Likewise.
35677 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
35679 PR rtl-optimization/78634
35680 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
35681 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
35682 * ifcvt.c (noce_try_cmove): Add missing cost check.
35684 PR rtl-optimization/71724
35685 * combine.c (if_then_else_cond): Look for situations where it is
35686 beneficial to undo the work of one of the recursive calls.
35688 2017-01-23 Bin Cheng <bin.cheng@arm.com>
35690 PR tree-optimization/70754
35691 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
35692 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
35693 combined stmt before it if not NULL.
35694 (combine_chains): Process refs reversely and compute dominance point
35697 2017-01-23 Martin Liska <mliska@suse.cz>
35699 PR tree-optimization/79196
35700 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
35701 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
35703 (strlen_optimize_stmt): Call the renamed function.
35705 2017-01-23 Michael Matz <matz@suse.de>
35707 PR tree-optimization/78384
35708 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
35710 2017-01-23 Richard Biener <rguenther@suse.de>
35712 PR tree-optimization/79186
35713 * tree-vrp.c (register_new_assert_for): Make sure we've seen
35714 both incoming edges before moving an assert.
35716 2017-01-23 Martin Jambor <mjambor@suse.cz>
35718 * ipa-prop.c (load_from_param_1): Removed.
35719 (load_from_unmodified_param): Bits from load_from_param_1 put back
35721 (load_from_param): Removed.
35722 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
35723 with stmt. Reverted back to use of load_from_unmodified_param.
35725 2017-01-23 Martin Jambor <mjambor@suse.cz>
35728 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
35729 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
35730 field a pointer to garbage collected vector, mark lattices and
35731 ipcp_orig_node with GTY((skip)).
35732 (ipa_get_param_count): Adjust to descriptors being a pointer.
35733 (ipa_get_param): Likewise.
35734 (ipa_get_type): Likewise.
35735 (ipa_get_param_move_cost): Likewise.
35736 (ipa_set_param_used): Likewise.
35737 (ipa_get_controlled_uses): Likewise.
35738 (ipa_set_controlled_uses): Likewise.
35739 (ipa_is_param_used): Likewise.
35740 (ipa_node_params_t): Move into garbage collector. New methods insert
35742 (ipa_node_params_sum): Annotate wth GTY(()).
35743 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
35745 (ipa_load_from_parm_agg): Adjust declaration.
35746 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
35747 * ipa-profile.c (ipa_profile): Likewise.
35748 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
35749 (ipa_populate_param_decls): Make descriptors parameter garbage
35751 (ipa_dump_param): Adjust to descriptors being a pointer.
35752 (ipa_alloc_node_params): Likewise.
35753 (ipa_initialize_node_params): Likewise.
35754 (load_from_param_1): Make descriptors parameter garbage collected.
35755 (load_from_unmodified_param): Likewise.
35756 (load_from_param): Likewise.
35757 (ipa_load_from_parm_agg): Likewise.
35758 (ipa_node_params::~ipa_node_params): Removed.
35759 (ipa_free_all_node_params): Remove call to delete operator.
35760 (ipa_node_params_t::insert): New.
35761 (ipa_node_params_t::remove): Likewise.
35762 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
35763 copy known_csts and known_contexts vectors.
35764 (ipa_read_node_info): Adjust to descriptors being a pointer.
35765 (ipcp_modif_dom_walker): Make m_descriptors field garbage
35767 (ipcp_transform_function): Make descriptors variable garbage
35770 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
35772 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
35773 * config/i386/avx512dqintrin.h: Ditto.
35774 * config/i386/avx512fintrin.h: Ditto.
35775 * config/i386/i386.c: Handle new builtins.
35776 * config/i386/i386-builtin.def: Add new builtins.
35777 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
35778 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
35780 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35781 Martin Liska <mliska@suse.cz>
35783 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
35784 * asan.c (asan_expand_poison_ifn): Support stores and use
35785 appropriate ASAN report function.
35786 * internal-fn.c (expand_ASAN_POISON_USE): New function.
35787 * internal-fn.def (ASAN_POISON_USE): Declare.
35788 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
35789 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
35790 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
35791 ASAN_POISON calls w/o LHS.
35792 * tree-ssa.c (execute_update_addresses_taken): Create clobber
35793 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
35794 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
35795 * gimplify.c (asan_poison_variables): Add attribute
35796 use_after_scope_memory to variables that really needs to live
35798 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
35799 having the attribute.
35801 2017-01-23 Martin Liska <mliska@suse.cz>
35803 * asan.c (create_asan_shadow_var): New function.
35804 (asan_expand_poison_ifn): Likewise.
35805 * asan.h (asan_expand_poison_ifn): New declaration.
35806 * internal-fn.c (expand_ASAN_POISON): Likewise.
35807 * internal-fn.def (ASAN_POISON): New builtin.
35808 * sanopt.c (pass_sanopt::execute): Expand
35809 asan_expand_poison_ifn.
35810 * tree-inline.c (copy_decl_for_dup_finish): Make function
35812 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
35813 * tree-ssa.c (is_asan_mark_p): New function.
35814 (execute_update_addresses_taken): Rewrite local variables
35815 (identified just by use-after-scope as addressable) into SSA.
35817 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35819 * doc/install.texi (Specific): opensource.apple.com uses https
35820 now. Remove trailing slash.
35822 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35824 * README.Portability: Remove note on an Irix compatibility issue.
35826 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
35828 * gcov.c (INCLUDE_ALGORITHM): Define.
35829 (INCLUDE_VECTOR): Define.
35830 No longer include <vector> and <algorithm> directly.
35832 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
35834 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
35836 * doc/invoke.texi (Code Gen Options): Ditto.
35838 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
35841 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
35843 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
35845 rtl-optimization/79125
35846 * cprop.c (local_cprop_pass): Handle cases where we make an
35847 unconditional trap.
35849 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
35853 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
35854 read from, for big endian.
35856 2017-01-20 Jiong Wang <jiong.wang@arm.com>
35858 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
35859 register pauth builtins for LP64 only.
35861 2017-01-20 Marek Polacek <polacek@redhat.com>
35864 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
35867 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35869 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
35871 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
35872 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
35873 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
35875 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35878 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
35879 in big-endian mode when they are not a single duplicated value.
35881 2017-01-20 Richard Biener <rguenther@suse.de>
35883 * BASE-VER: Bump to 7.0.1.
35885 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35887 * omp-low.c (omplow_simd_context): New struct. Use it...
35888 (lower_rec_simd_input_clauses): ...here and...
35889 (lower_rec_input_clauses): ...here to hold common data. Adjust all
35890 references to idx, lane, max_vf, is_simt.
35892 2017-01-20 Graham Markall <graham.markall@embecosm.com>
35894 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
35897 2017-01-20 Martin Jambor <mjambor@suse.cz>
35899 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
35900 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
35902 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
35903 (GTFILES): Rename hsa.c to hsa-common.c.
35904 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
35905 * hsa-dump.c: Likewise.
35906 * hsa-gen.c: Likewise.
35907 * hsa-regalloc.c: Likewise.
35908 * ipa-hsa.c: Likewise.
35909 * omp-expand.c: Likewise.
35910 * omp-low.c: Likewise.
35911 * toplev.c: Likewise.
35913 2017-01-20 Marek Polacek <polacek@redhat.com>
35916 * doc/invoke.texi: Document -Wduplicated-branches.
35917 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
35918 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
35919 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
35920 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
35921 return 0 only when not OEP_LEXICOGRAPHIC.
35922 (fold_build_cleanup_point_expr): Use the expression
35923 location when building CLEANUP_POINT_EXPR.
35924 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
35925 * tree.c (add_expr): Handle error_mark_node.
35927 2017-01-20 Martin Liska <mliska@suse.cz>
35930 * tree-profile.c (init_ic_make_global_vars): Do not call
35932 (gimple_init_gcov_profiler): Likewise.
35934 2017-01-20 Martin Liska <mliska@suse.cz>
35937 * cgraph.h (maybe_create_reference): Remove argument and
35939 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
35941 * ipa-cp.c (create_specialized_node): Likewise.
35942 * symtab.c (symtab_node::maybe_create_reference): Handle
35943 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
35945 2017-01-20 Martin Liska <mliska@suse.cz>
35947 * read-rtl-function.c (function_reader::create_function): Use
35948 build_decl instread of build_decl_stat.
35950 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
35952 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
35953 * config/i386/avx512dqintrin.h: Ditto.
35954 * config/i386/avx512fintrin.h: Ditto.
35955 * config/i386/i386-builtin-types.def: Add new types.
35956 * config/i386/i386.c: Handle new types.
35957 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
35958 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
35959 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
35960 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
35961 (__builtin_ia32_kshiftridi): New.
35962 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
35964 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
35968 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
35969 define to rs6000_init_stack_protect_guard.
35970 (rs6000_init_stack_protect_guard): New function.
35972 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35973 Yunqiang Su <yunqiang.su@imgtec.com>
35975 * config.gcc (supported_defaults): Add madd4.
35976 (with_madd4): Add validation.
35977 (all_defaults): Add madd4.
35978 * config/mips/mips.opt (mmadd4): New option.
35979 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35981 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35983 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
35984 (ISA_HAS_FUSED_MADD4): Likewise.
35985 * doc/invoke.texi (-mmadd4): Document the new option.
35986 * doc/install.texi (--with-madd4): Document the new option.
35988 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35990 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
35991 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
35992 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
35993 (aarch64_init_pauth_hint_builtins): New.
35994 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
35995 (aarch64_expand_builtin): Expand new builtins.
35997 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35999 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
36000 * combine-stack-adj.c (no_unhandled_cfa): Handle
36001 REG_CFA_TOGGLE_RA_MANGLE.
36002 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
36003 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
36004 info for return address signing.
36005 (aarch64_expand_epilogue): Likewise.
36007 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36009 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
36010 * config/aarch64/aarch64-protos.h
36011 (aarch64_return_address_signing_enabled): New declaration.
36012 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
36014 (aarch64_expand_prologue): Sign return address before it's pushed onto
36016 (aarch64_expand_epilogue): Authenticate return address fetched from
36018 (aarch64_override_options): Sanity check for ILP32 and ISA level.
36019 (aarch64_attributes): New function attributes for "sign-return-address".
36020 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
36021 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
36022 ("*do_return"): Generate combined instructions according to key index.
36023 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
36024 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
36026 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
36027 * config/aarch64/aarch64.opt (msign-return-address=): New.
36028 * doc/extend.texi (AArch64 Function Attributes): Documents
36029 "sign-return-address=".
36030 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
36032 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36034 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
36035 overall option summary.
36037 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36039 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
36040 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
36041 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
36042 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
36044 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
36046 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
36047 -mpower9-minmax by default for -mcpu=power9.
36048 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
36049 128-bit floating point.
36051 2017-01-20 Alan Modra <amodra@gmail.com>
36053 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36054 optimizing for size.
36056 2017-01-20 Alan Modra <amodra@gmail.com>
36059 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36060 for strcmp and strncmp from corresponding builtin decl.
36062 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36064 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36065 instead of i386/rtems-64.h.
36066 * config/i386/rtems-64.h: Remove.
36068 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36072 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36074 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36076 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36078 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36079 Change int to HOST_WIDE_INT.
36080 * config/aarch64/aarch64-protos.h
36081 (aarch64_simd_gen_const_vector_dup): Likewise.
36082 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36084 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36086 * langhooks-def.h (lhd_type_for_size): New decl.
36087 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36088 * langhooks.c (lhd_type_for_size): New function, taken from
36091 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36093 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36094 define_bypass for CR latency.
36095 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36096 (power9-alu2): Add define_bypass for CR latency.
36098 (power9-mul): Update insn latency.
36099 (power9-mul-compare): Update insn latency, bypass latency and remove
36102 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36104 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36106 * config/aarch64/aarch64.md
36107 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36108 aarch64_nopcrelative_literal_loads.
36109 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36111 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
36113 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
36114 TARGET_LOONGSON_3A.
36115 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
36117 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
36120 * config.gcc (supported_defaults): Add lxc1-sxc1.
36121 (with_lxc1_sxc1): Add validation.
36122 (all_defaults): Add lxc1-sxc1.
36123 * config/mips/mips.opt (mlxc1-sxc1): New option.
36124 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36126 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36127 __mips_no_lxc1_sxc1.
36128 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
36129 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
36130 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
36132 2017-01-19 Richard Biener <rguenther@suse.de>
36134 PR tree-optimization/72488
36135 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
36136 sure to restore SSA info.
36137 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
36139 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
36141 PR rtl-optimization/79121
36142 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
36143 of the inner type when shifting an extended value.
36145 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36148 * symtab.c (symtab_node::equal_address_to): Fix comparing of
36149 interposable aliases.
36151 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
36154 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
36155 Use the evmergelohi instruction.
36156 (mov_si<mode>_e500_subreg4_2_le): Likewise.
36157 (mov_sitf_e500_subreg8_2_be): Likewise.
36158 (mov_sitf_e500_subreg12_2_le): Likewise.
36159 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
36160 (mov_si<mode>_e500_subreg4_2_be): Likewise.
36161 (mov_sitf_e500_subreg8_2_le): Likewise.
36162 (mov_sitf_e500_subreg12_2_be): Likewise.
36164 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36166 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
36167 attribute from vecsimple to vecperm.
36168 (altivec_vbpermq2): Likewise.
36170 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36173 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
36175 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36176 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
36177 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
36178 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
36179 case where N arg is SIZE_MAX.
36180 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
36181 (cmpstrsi): Add pattern.
36183 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
36185 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
36186 __builtin_vec_revb builtins.
36187 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
36188 built-in functions to support generation of the ISA 3.0 XXBR<x>
36189 vector byte reverse instructions.
36190 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
36191 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
36192 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
36193 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
36194 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
36195 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
36196 (P9V_BUILTIN_VEC_REVB): Likewise.
36197 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
36198 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
36199 (p9_xxbrq_v16qi): Likewise.
36200 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
36201 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
36202 (p9_xxbrh_v8hi): Likewise.
36203 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
36204 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
36205 vec_revb built-in functions.
36207 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
36209 PR rtl-optimization/78952
36210 * config/i386/i386.md (any_extract): New code iterator.
36211 (*insvqi_2): Use any_extract for source operand.
36212 (*insvqi_3): Use any_shiftrt for source operand.
36214 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
36216 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
36218 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
36220 2017-01-18 Matthias Klose <doko@ubuntu.com>
36222 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
36224 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36226 * config/rs6000/altivec.h (vec_bperm): Change #define.
36227 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
36228 (altivec_vbpermq2): New define_insn.
36229 (altivec_vbpermd): Likewise.
36230 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
36231 function interface.
36232 (VBPERMD): Likewise.
36233 (VBPERM): New polymorphic function interface.
36234 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
36235 Add entries for P9V_BUILTIN_VEC_VBPERM.
36236 * doc/extend.texi: Add interfaces for vec_bperm.
36238 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36240 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
36241 first letter of error messages.
36242 (s390_resolve_overloaded_builtin): Likewise.
36243 * config/s390/s390.c (s390_expand_builtin): Likewise.
36244 (s390_invalid_arg_for_unprototyped_fn): Likewise.
36245 (s390_valid_target_attribute_inner_p): Likewise.
36246 * config/s390/s390.md ("tabort"): Likewise.
36248 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
36250 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
36251 (ISA_AVOID_DIV_HILO): New macro.
36252 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
36253 (ISA_HAS_DDIV): Likewise.
36255 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36257 * doc/invoke.texi (fabi-version): Correct number of occurrences.
36259 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36261 * doc/invoke.texi (fabi-version): Spelling fix.
36263 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36266 * doc/invoke.texi (fabi-version): Mention mangling fix for
36269 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36272 * doc/invoke.texi (fabi-version): Document discriminator mangling.
36274 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
36277 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
36278 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
36280 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
36282 (stack_protect_test): Ditto.
36283 * config/rs6000/rs6000.opt (mstack-protector-guard=,
36284 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
36286 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
36287 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
36288 -mstack-protector-guard-offset=.
36289 (RS/6000 and PowerPC Options): Ditto.
36291 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36293 * config/i386/i386.h (MASK_CLASS_P): New define.
36294 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
36295 there are no registers from different register sets also when
36296 mask registers are used. Update function comment.
36297 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
36298 to (*k/*r) and (*k/*km) alternatives.
36300 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
36302 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
36303 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
36304 (EH_RETURN_HANDLER_RTX): New define.
36305 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
36306 Force frame pointer in EH return functions.
36307 (aarch64_expand_epilogue): Add barrier for eh_return.
36308 (aarch64_final_eh_return_addr): Remove.
36309 (aarch64_eh_return_handler_rtx): New function.
36310 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
36312 (aarch64_eh_return_handler_rtx): New prototype.
36314 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36316 * config/rs6000/altivec.h (vec_rlmi): New #define.
36317 (vec_vrlnm): Likewise.
36318 (vec_rlnm): Likewise.
36319 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
36320 (UNSPEC_VRLNM): Likewise.
36321 (VIlong): New mode iterator.
36322 (altivec_vrl<VI_char>mi): New define_insn.
36323 (altivec_vrl<VI_char>nm): Likewise.
36324 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
36326 (VRLDNM): Likewise.
36327 (RLNM): New polymorphic function entry.
36328 (VRLWMI): New monomorphic function entry.
36329 (VRLDMI): Likewise.
36330 (RLMI): New polymorphic function entry.
36331 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
36332 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
36333 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
36336 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36339 * dwarf2out.c (field_byte_offset): Restore the
36340 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
36341 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
36342 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
36345 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36348 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
36350 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36353 * dwarf2out.c (add_data_member_location_attribute): For constant
36354 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
36355 instead of DW_AT_data_member_location, DW_AT_bit_offset and
36356 DW_AT_byte_size attributes.
36358 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36360 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
36361 after forcing to constant memory when the code model is medium.
36363 2017-01-17 Julia Koval <julia.koval@intel.com>
36366 * config/i386/avx512fintrin.h
36367 (_mm512_i32gather_ps): Change __addr type to void const*.
36368 (_mm512_mask_i32gather_ps): Ditto.
36369 (_mm512_i32gather_pd): Ditto.
36370 (_mm512_mask_i32gather_pd): Ditto.
36371 (_mm512_i64gather_ps): Ditto.
36372 (_mm512_mask_i64gather_ps): Ditto.
36373 (_mm512_i64gather_pd): Ditto.
36374 (_mm512_mask_i64gather_pd): Ditto.
36375 (_mm512_i32gather_epi32): Ditto.
36376 (_mm512_mask_i32gather_epi32): Ditto.
36377 (_mm512_i32gather_epi64): Ditto.
36378 (_mm512_mask_i32gather_epi64): Ditto.
36379 (_mm512_i64gather_epi32): Ditto.
36380 (_mm512_mask_i64gather_epi32): Ditto.
36381 (_mm512_i64gather_epi64): Ditto.
36382 (_mm512_mask_i64gather_epi64): Ditto.
36383 (_mm512_i32scatter_ps): Change __addr type to void*.
36384 (_mm512_mask_i32scatter_ps): Ditto.
36385 (_mm512_i32scatter_pd): Ditto.
36386 (_mm512_mask_i32scatter_pd): Ditto.
36387 (_mm512_i64scatter_ps): Ditto.
36388 (_mm512_mask_i64scatter_ps): Ditto.
36389 (_mm512_i64scatter_pd): Ditto.
36390 (_mm512_mask_i64scatter_pd): Ditto.
36391 (_mm512_i32scatter_epi32): Ditto.
36392 (_mm512_mask_i32scatter_epi32): Ditto.
36393 (_mm512_i32scatter_epi64): Ditto.
36394 (_mm512_mask_i32scatter_epi64): Ditto.
36395 (_mm512_i64scatter_epi32): Ditto.
36396 (_mm512_mask_i64scatter_epi32): Ditto.
36397 (_mm512_i64scatter_epi64): Ditto.
36398 (_mm512_mask_i64scatter_epi64): Ditto.
36399 * config/i386/avx512pfintrin.h
36400 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
36401 (_mm512_mask_prefetch_i32gather_ps): Ditto.
36402 (_mm512_mask_prefetch_i64gather_pd): Ditto.
36403 (_mm512_mask_prefetch_i64gather_ps): Ditto.
36404 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
36405 (_mm512_prefetch_i32scatter_ps): Ditto.
36406 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
36407 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
36408 (_mm512_prefetch_i64scatter_pd): Ditto.
36409 (_mm512_prefetch_i64scatter_ps): Ditto.
36410 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
36411 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
36412 * config/i386/avx512vlintrin.h
36413 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
36414 (_mm_mmask_i32gather_ps): Ditto.
36415 (_mm256_mmask_i32gather_pd): Ditto.
36416 (_mm_mmask_i32gather_pd): Ditto.
36417 (_mm256_mmask_i64gather_ps): Ditto.
36418 (_mm_mmask_i64gather_ps): Ditto.
36419 (_mm256_mmask_i64gather_pd): Ditto.
36420 (_mm_mmask_i64gather_pd): Ditto.
36421 (_mm256_mmask_i32gather_epi32): Ditto.
36422 (_mm_mmask_i32gather_epi32): Ditto.
36423 (_mm256_mmask_i32gather_epi64): Ditto.
36424 (_mm_mmask_i32gather_epi64): Ditto.
36425 (_mm256_mmask_i64gather_epi32): Ditto.
36426 (_mm_mmask_i64gather_epi32): Ditto.
36427 (_mm256_mmask_i64gather_epi64): Ditto.
36428 (_mm_mmask_i64gather_epi64): Ditto.
36429 (_mm256_i32scatter_ps): Change __addr type to void*.
36430 (_mm256_mask_i32scatter_ps): Ditto.
36431 (_mm_i32scatter_ps): Ditto.
36432 (_mm_mask_i32scatter_ps): Ditto.
36433 (_mm256_i32scatter_pd): Ditto.
36434 (_mm256_mask_i32scatter_pd): Ditto.
36435 (_mm_i32scatter_pd): Ditto.
36436 (_mm_mask_i32scatter_pd): Ditto.
36437 (_mm256_i64scatter_ps): Ditto.
36438 (_mm256_mask_i64scatter_ps): Ditto.
36439 (_mm_i64scatter_ps): Ditto.
36440 (_mm_mask_i64scatter_ps): Ditto.
36441 (_mm256_i64scatter_pd): Ditto.
36442 (_mm256_mask_i64scatter_pd): Ditto.
36443 (_mm_i64scatter_pd): Ditto.
36444 (_mm_mask_i64scatter_pd): Ditto.
36445 (_mm256_i32scatter_epi32): Ditto.
36446 (_mm256_mask_i32scatter_epi32): Ditto.
36447 (_mm_i32scatter_epi32): Ditto.
36448 (_mm_mask_i32scatter_epi32): Ditto.
36449 (_mm256_i32scatter_epi64): Ditto.
36450 (_mm256_mask_i32scatter_epi64): Ditto.
36451 (_mm_i32scatter_epi64): Ditto.
36452 (_mm_mask_i32scatter_epi64): Ditto.
36453 (_mm256_i64scatter_epi32): Ditto.
36454 (_mm256_mask_i64scatter_epi32): Ditto.
36455 (_mm_i64scatter_epi32): Ditto.
36456 (_mm_mask_i64scatter_epi32): Ditto.
36457 (_mm256_i64scatter_epi64): Ditto.
36458 (_mm256_mask_i64scatter_epi64): Ditto.
36459 (_mm_i64scatter_epi64): Ditto.
36460 (_mm_mask_i64scatter_epi64): Ditto.
36461 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
36462 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
36463 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
36464 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
36465 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
36466 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
36467 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
36468 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
36469 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
36470 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
36471 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
36472 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
36473 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
36474 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
36475 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
36476 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
36477 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
36478 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
36479 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
36480 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
36481 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
36482 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
36483 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
36484 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
36485 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
36486 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
36487 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
36488 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
36489 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
36490 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
36491 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
36492 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
36493 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
36494 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
36495 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
36496 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
36497 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
36498 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
36499 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
36500 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
36501 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
36502 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
36503 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
36504 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
36505 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
36506 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
36507 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
36508 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
36509 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
36510 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
36511 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
36512 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
36513 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
36514 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
36515 definitions accordingly.
36517 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
36518 Kuan-Lin Chen <kuanlinchentw@gmail.com>
36521 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
36524 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
36527 * ira-conflicts.c (ira_build_conflicts): Update total conflict
36528 hard regs for inner regno.
36530 2017-01-17 Martin Liska <mliska@suse.cz>
36533 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
36534 assumption and add comment.
36536 2017-01-17 Nathan Sidwell <nathan@acm.org>
36538 * ipa-visibility.c (localize_node): New function, broken out of ...
36539 (function_and_variable_visibility): ... here. Call it.
36541 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36543 PR middle-end/77445
36544 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
36545 correctly set frequency of oudgoing edge.
36546 (duplicate_thread_path): Fix profile updating.
36548 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36551 * configure.ac: Add GCC_BASE_VER.
36552 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
36553 version from BASE-VER file.
36554 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
36555 (gcc.o): Depend on $(BASEVER).
36556 * common.opt (dumpfullversion): New option.
36557 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
36558 * doc/invoke.texi: Document -dumpfullversion.
36559 * doc/install.texi: Document --with-gcc-major-version-only.
36560 * configure: Regenerated.
36562 2017-01-17 Richard Biener <rguenther@suse.de>
36564 PR tree-optimization/71433
36565 * tree-vrp.c (register_new_assert_for): Merge same asserts
36566 on all incoming edges.
36567 (process_assert_insertions_for): Handle insertions at the
36570 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
36572 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
36573 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
36575 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
36578 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
36581 2017-01-17 Alan Modra <amodra@gmail.com>
36584 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
36585 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
36586 symbolic stack limit when pic.
36588 2017-01-16 Martin Sebor <msebor@redhat.com>
36590 PR tree-optimization/78608
36591 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
36593 2017-01-16 Jeff Law <law@redhat.com>
36596 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
36597 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
36598 for several include directories that may be relative to sysroot.
36599 * config/i386/x-mingw32 (gplus_includedir): Define.
36600 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
36601 (native_system_includedir): Likewise.
36602 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
36603 override if TARGET_SYSTEM_ROOT is defined.
36604 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
36606 PR tree-optimization/79090
36607 PR tree-optimization/33562
36608 PR tree-optimization/61912
36609 PR tree-optimization/77485
36610 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
36611 and computed trims into the dump file.
36613 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36615 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
36617 2017-01-16 Jakub Jelinek <jakub@redhat.com>
36620 * gimplify.c (gimplify_init_constructor): If want_value and
36621 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
36625 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
36626 sequence. Formatting fixes.
36627 (doloop_optimize): Formatting fixes.
36630 * gcc.c (debug_level_greater_than_spec_func): New function.
36631 (static_spec_functions): Add debug-level-gt spec function.
36632 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
36634 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
36635 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
36636 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
36637 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
36638 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
36639 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
36641 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
36643 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
36644 QImode fixups to general and mask registers only.
36646 2017-01-16 Carl Love <cel@us.ibm.com>
36648 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
36649 for built-in functions
36650 vector signed char vec_nabs (vector signed char)
36651 vector signed short vec_nabs (vector signed short)
36652 vector signed int vec_nabs (vector signed int)
36653 vector signed long long vec_nabs (vector signed long long)
36654 vector float vec_nabs (vector float)
36655 vector double vec_nabs (vector double)
36656 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
36658 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
36659 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
36660 * doc/extend.texi: Update the documentation file for the new built-in
36663 2017-01-16 Martin Sebor <msebor@redhat.com>
36665 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
36668 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36670 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
36671 UNSPEC_VSX__XXSPLTD to require special splat handling.
36673 2017-01-16 David Malcolm <dmalcolm@redhat.com>
36676 * system.h: Poison strndup.
36678 2017-01-16 Alan Modra <amodra@gmail.com>
36681 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
36684 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
36686 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
36688 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
36690 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
36691 call recog here. Assert that INSN_CODE (insn) is non-negative.
36693 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36696 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
36698 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
36699 in the currently scheduled RTL fragment.
36701 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36703 PR rtl-optimization/78751
36704 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
36707 2017-01-14 Jeff Law <law@redhat.com>
36709 PR tree-optimization/79090
36710 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
36711 variable length stores.
36712 (compute_trims): Delete dead assignment to *trim_tail.
36713 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
36716 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
36718 PR rtl-optimization/78626
36719 PR rtl-optimization/78727
36720 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
36721 of a block, and split such blocks after everything else is finished.
36723 2017-01-14 Alan Modra <amodra@gmail.com>
36726 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
36727 target legitimate_combined_insn.
36728 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
36729 (rs6000_legitimate_combined_insn): New function.
36730 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
36732 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
36733 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
36734 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
36736 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
36738 * doc/frontends.texi (G++ and GCC): Remove references to Java.
36740 2017-01-13 Jeff Law <law@redhat.com>
36742 PR tree-optimization/33562
36743 PR tree-optimization/61912
36744 PR tree-optimization/77485
36745 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
36747 (delete_dead_assignment): Likewise.
36748 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
36749 statement to delete_dead_call and delete_dead_assignment.
36751 2017-01-13 David Malcolm <dmalcolm@redhat.com>
36754 * substring-locations.c (format_warning_va): Strengthen case 1 so
36755 that both endpoints of the substring must be within the format
36756 range for just the substring to be printed.
36758 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
36760 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
36761 * config/i386/i386.c (ix86_target_string): Add missing options
36762 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
36763 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
36764 flags_other and ix86_target_other to flags2_other. Display unknown
36766 (ix86_valid_target_attribute_inner_p): Add missing options and
36767 reorder options by implied ISAs, as in ix86_target_string.
36769 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36771 * hash-table.h (hash_table::too_empty_p): New function.
36772 (hash_table::expand): Use it.
36773 (hash_table::traverse): Likewise.
36774 (hash_table::empty_slot): Use sizeof (value_type) instead of
36775 sizeof (PTR) to convert bytes to elements. Shrink the table
36776 if the current size is excessive for the current number of
36779 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36781 * ira-costs.c (record_reg_classes): Break from the inner loop
36782 early once alt_fail is known to be true. Update outer loop
36783 handling accordingly.
36785 2017-01-13 Jeff Law <law@redhat.com>
36787 * tree-ssa-dse.c (decrement_count): New function.
36788 (increment_start_addr, maybe_trim_memstar_call): Likewise.
36789 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
36790 when we know the partially dead statement is a mem* function.
36792 PR tree-optimization/61912
36793 PR tree-optimization/77485
36794 * tree-ssa-dse.c: Include expr.h.
36795 (maybe_trim_constructor_store): New function.
36796 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
36798 PR tree-optimization/33562
36799 PR tree-optimization/61912
36800 PR tree-optimization/77485
36801 * doc/invoke.texi: Document new dse-max-object-size param.
36802 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
36803 * tree-ssa-dse.c: Include params.h.
36804 (dse_store_status): New enum.
36805 (initialize_ao_ref_for_dse): New, partially extracted from
36807 (valid_ao_ref_for_dse, normalize_ref): New.
36808 (setup_live_bytes_from_ref, compute_trims): Likewise.
36809 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
36810 (maybe_trim_partially_dead_store): Likewise.
36811 (maybe_trim_complex_store): Likewise.
36812 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
36813 Track what bytes live from the original store. Return tri-state
36814 for dead, partially dead or live.
36815 (dse_dom_walker): Add constructor, destructor and new private members.
36816 (delete_dead_call, delete_dead_assignment): New extracted from
36818 (dse_optimize_stmt): Make a member of dse_dom_walker.
36819 Use initialize_ao_ref_for_dse.
36821 PR tree-optimization/33562
36822 PR tree-optimization/61912
36823 PR tree-optimization/77485
36824 * sbitmap.h (bitmap_count_bits): Prototype.
36825 (bitmap_clear_range, bitmap_set_range): Likewise.
36826 * sbitmap.c (bitmap_clear_range): New function.
36827 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
36829 2017-01-13 Martin Liska <mliska@suse.cz>
36832 * function.c (set_cfun): Add new argument force.
36833 * function.h (set_cfun): Likewise.
36834 * ipa-inline-transform.c (inline_call): Use the function when
36835 strict alising from is dropped for function we inline to.
36837 2017-01-13 Richard Biener <rguenther@suse.de>
36839 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
36840 for dumping GIMPLE INTEGER_CSTs.
36842 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
36844 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
36845 to 201112L since C++17.
36847 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
36850 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
36851 if -fsanitize=kernel-address is present.
36853 2017-01-13 Richard Biener <rguenther@suse.de>
36855 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
36856 as _Literal ( type ) number in case usual suffixes do not
36857 preserve all information.
36859 2017-01-13 Richard Biener <rguenther@suse.de>
36861 PR tree-optimization/77283
36862 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
36863 and ssa-iterators.h.
36864 (is_feasible_trace): Implement a cost model based on joiner
36867 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
36870 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
36871 char or short to __float128/_Float128 directly.
36873 2017-01-12 Martin Sebor <msebor@redhat.com>
36875 to -Wformat-overflow.
36876 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
36877 (min_bytes_remaining): Same.
36878 (get_string_length): Same.
36879 (format_string): Same.
36880 (format_directive): Same.
36882 (pass_sprintf_length::handle_gimple_call): Same.
36884 2017-01-12 Jakub Jelinek <jakub@redhat.com>
36886 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
36887 info.nowrite calls with no lhs that can't throw. Return bool
36888 whether gsi_remove has been called or not.
36889 (pass_sprintf_length::handle_gimple_call): Return bool whether
36890 try_substitute_return_value called gsi_remove. Formatting fix.
36891 (pass_sprintf_length::execute): Don't use gsi_remove if
36892 handle_gimple_call returned true.
36895 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
36896 be removed due to side-effects, don't remove following barrier nor
36897 turn the successor edge into fallthru edge.
36899 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36902 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
36903 element-reversing loads and stores as not swappable.
36905 2017-01-12 Nathan Sidwell <nathan@acm.org>
36906 Nicolai Stange <nicstange@gmail.com>
36908 * combine.c (try_combine): Don't ignore result of overlap checking
36909 loop. Combine overlap & asm check into single loop.
36911 2017-01-12 Richard Biener <rguenther@suse.de>
36913 * tree-pretty-print.c (dump_generic_node): Provide -gimple
36914 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
36916 2017-01-12 Richard Biener <rguenther@suse.de>
36918 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
36919 and TS_TARGET_OPTION directly derive from TS_BASE.
36920 * tree-core.h (tree_optimization_option): Derive from tree_base.
36921 (tree_target_option): Likewise.
36923 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
36925 * config/i386/i386.c (memory_address_length): Increase len
36926 only when rip_relative_addr_p returns false.
36928 2017-01-11 Julia Koval <julia.koval@intel.com>
36930 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
36931 (OPTION_MASK_ISA_SGX_SET): New.
36932 (ix86_handle_option): Handle OPT_msgx.
36933 * config.gcc: Added sgxintrin.h.
36934 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
36935 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
36936 * config/i386/i386.c (ix86_target_string): Add -msgx.
36938 (ix86_option_override_internal): Handle new options.
36939 (ix86_valid_target_attribute_inner_p): Add sgx.
36940 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
36941 * config/i386/i386.opt: Add msgx.
36942 * config/i386/sgxintrin.h: New file.
36943 * config/i386/x86intrin.h: Add sgxintrin.h.
36945 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36948 * fold-const.c (maybe_nonzero_address): Return 1 for function
36950 (tree_single_nonzero_warnv_p): Don't handle function local objects
36954 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
36957 2017-01-11 David Malcolm <dmalcolm@redhat.com>
36960 * opts.c: Include "spellcheck.h"
36961 (struct string_fragment): New struct.
36962 (struct edit_distance_traits<const string_fragment &>): New
36964 (get_closest_sanitizer_option): New function.
36965 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
36967 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36969 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
36971 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
36972 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
36973 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
36974 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
36975 for initial die_offset if dwarf_split_debug_info.
36976 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
36977 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
36979 (output_skeleton_debug_sections): Formatting fix. Use
36980 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
36981 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
36983 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
36985 * config/arm/cortex-a53.md: Add bypasses for
36986 cortex_a53_r2f_cvt.
36987 (cortex_a53_r2f): Only use for transfers.
36988 (cortex_a53_f2r): Likewise.
36989 (cortex_a53_r2f_cvt): Add reservation for conversions.
36990 (cortex_a53_f2r_cvt): Likewise.
36992 2017-01-11 Tamar Christina <tamar.christina@arm.com>
36994 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
36995 to all inlined functions, change static to extern.
36997 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
37000 * config/arm/arm.c (legitimize_pic_address): Handle reference to
37002 (arm_assemble_integer): Likewise.
37004 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
37006 * config.gcc: Use new awk script to check CPU, FPU and architecture
37007 parameters for --with-... options.
37008 * config/arm/parsecpu.awk: New file
37009 * config/arm/arm-cpus.in: New file.
37010 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
37012 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
37014 * config/arm/t-arm: Update dependency rules.
37015 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
37016 of processing .def files.
37017 * config/arm/genopt.sh: Deleted.
37018 * config/arm/gentune.sh: Deleted.
37019 * config/arm/arm-cores.def: Deleted.
37020 * config/arm/arm-arches.def: Deleted.
37021 * config/arm/arm-fpus.def: Deleted.
37022 * config/arm/arm-tune.md: Regenerated.
37023 * config/arm/arm-tables.opt: Regenerated.
37024 * config/arm/arm-cpu.h: New generated file.
37025 * config/arm/arm-cpu-data.h: New generated file.
37026 * config/arm/arm-cpu-cdata.h: New generated file.
37028 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
37031 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
37033 (input_varpool_node): Unpack dynamically_initialized bit.
37035 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
37037 PR rtl-optimization/79032
37038 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
37039 the alignment of the adjusted memory reference against that of MODE,
37040 instead of the alignment of the original memory reference.
37042 2017-01-11 Martin Jambor <mjambor@suse.cz>
37044 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
37046 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
37047 decorated functions.
37049 2017-01-11 Richard Biener <rguenther@suse.de>
37051 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
37052 set range/nonnull info for PHI results. Do not set it on
37053 stmts marked for removal.
37055 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37057 * expr.c (store_field): In the bitfield case, fetch the return value
37058 from the registers before applying a single big-endian adjustment.
37059 Always do a final load for a BLKmode value not larger than a word.
37061 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37064 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37065 that we correctly handle column numbers greater than
37066 LINE_MAP_MAX_COLUMN_NUMBER.
37068 2017-01-10 Martin Sebor <msebor@redhat.com>
37070 PR middle-end/78245
37071 * gimple-ssa-sprintf.c (get_destination_size): Call
37072 {init,fini}object_sizes.
37073 * tree-object-size.c (addr_object_size): Adjust.
37074 (pass_through_call): Adjust.
37075 (pass_object_sizes::execute): Adjust.
37076 * tree-object-size.h (fini_object_sizes): Declare.
37078 2017-01-10 Martin Sebor <msebor@redhat.com>
37080 PR tree-optimization/78775
37081 * builtins.c (get_size_range): Move...
37082 * calls.c: ...to here.
37083 (alloc_max_size): Accept zero argument.
37084 (operand_signed_p): Remove.
37085 (maybe_warn_alloc_args_overflow): Call get_size_range.
37086 * calls.h (get_size_range): Declare.
37088 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37090 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37091 from TI's devices.csv file as of September 2016.
37092 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37094 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37096 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37097 * doc/invoke.texi: Likewise.
37098 * doc/md.texi: Likewise.
37099 * doc/objc.texi: Likewise.
37101 2017-01-10 Joshua Conner <joshconner@google.com>
37103 * config/arm/fuchsia-elf.h: New file.
37104 * config/fuchsia.h: New file.
37105 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37106 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37108 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37110 2016-01-10 Richard Biener <rguenther@suse.de>
37112 PR tree-optimization/79034
37113 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
37114 Propagate out degenerate PHIs in the joiner.
37116 2017-01-10 Martin Liska <mliska@suse.cz>
37118 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
37119 (sort_congruence_classes_by_decl_uid): Likewise.
37120 (sort_congruence_class_groups_by_decl_uid): Likewise.
37121 (sem_item_optimizer::merge_classes): Sort class, groups in these
37122 classes and members in the groups by DECL_UID of declarations.
37123 This would make merge operations stable.
37125 2017-01-10 Martin Liska <mliska@suse.cz>
37127 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
37128 usage of m_classes_vec.
37129 (sem_item_optimizer::~sem_item_optimizer): Likewise.
37130 (sem_item_optimizer::get_group_by_hash): Likewise.
37131 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
37132 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
37133 (sem_item_optimizer::verify_classes): Likewise.
37134 (sem_item_optimizer::process_cong_reduction): Likewise.
37135 (sem_item_optimizer::dump_cong_classes): Likewise.
37136 (sem_item_optimizer::merge_classes): Likewise.
37137 * ipa-icf.h (congruence_class_hash): Rename from
37138 congruence_class_group_hash. Remove declaration of m_classes_vec.
37140 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
37142 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
37143 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
37144 * config.gcc: Add avx512vpopcntdqintrin.h.
37145 * config/i386/avx512vpopcntdqintrin.h: New.
37146 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
37147 * config/i386/i386-builtin-types.def: Add new types.
37148 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
37149 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
37150 __builtin_ia32_vpopcountq_v8di_mask): New.
37151 * config/i386/i386-c.c (ix86_target_macros_internal): Define
37152 __AVX512VPOPCNTDQ__.
37153 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
37154 (PTA_AVX512VPOPCNTDQ): Define.
37155 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
37156 TARGET_AVX512VPOPCNTDQ_P): Define.
37157 * config/i386/i386.opt: Add mavx512vpopcntdq.
37158 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
37159 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
37161 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37163 PR middle-end/77484
37164 * predict.def (PRED_CALL): Set to 67.
37166 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
37168 * expr.c (store_field): In the bitfield case, if the value comes from
37169 a function call and is of an aggregate type returned in registers, do
37170 not modify the field mode; extract the value in all cases if the mode
37171 is BLKmode and the size is not larger than a word.
37173 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
37176 * config/i386/cpuid.h: Fix undefined behavior.
37178 2017-01-04 Jeff Law <law@redhat.com>
37180 PR tree-optimization/79007
37181 PR tree-optimization/67955
37182 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
37183 conservative for pt.null when flag_non_call_exceptions is on.
37185 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37187 PR translation/79019
37188 PR translation/79020
37189 * params.def (PARAM_INLINE_MIN_SPEEDUP,
37190 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
37191 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
37193 * config/avr/avr.opt (maccumulate-args): Likewise.
37194 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
37195 * common.opt (freport-bug): Likewise.
37196 * cif-code.def (CIF_FINAL_ERROR): Likewise.
37197 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
37198 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
37199 translatable string.
37200 * config/i386/i386.c (function_value_32): Likewise.
37201 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
37202 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
37204 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
37205 * common/config/msp430/msp430-common.c (msp430_handle_option):
37207 * symtab.c (symtab_node::verify_base): Likewise.
37208 * opts.c (set_debug_level): Likewise.
37209 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
37210 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
37211 missing whitespace to translatable strings.
37212 * config/avr/avr.md (bswapsi2): Fix typo in comment.
37213 * config/sh/superh.h: Likewise.
37214 * config/i386/xopintrin.h: Likewise.
37215 * config/i386/znver1.md: Likewise.
37216 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
37217 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
37218 * double-int.h (struct double_int): Likewise.
37219 * double-int.c (div_and_round_double): Likewise.
37220 * wide-int.cc: Likewise.
37221 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
37222 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
37223 * cfgcleanup.c (crossjumps_occured): Renamed to ...
37224 (crossjumps_occurred): ... this.
37225 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
37228 PR tree-optimization/78899
37229 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
37230 returning bool return struct loop *, NULL for failure and the new
37232 (versionable_outer_loop_p): Don't version outer loop if it has
37233 dont_vectorized bit set.
37234 (tree_if_conversion): When versioning outer loop, ensure
37235 tree_if_conversion is performed also on the inner loop of the
37236 non-vectorizable outer loop copy.
37237 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
37238 LOOP_VECTORIZED in inner loop of the scalar outer loop and
37239 prevent vectorization of it.
37240 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
37241 the outer loop vectorization of the non-scalar version is attempted
37242 before vectorization of the inner loop in scalar version. If
37243 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
37244 vectorization of its inner loop.
37245 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
37246 has 2 inner loops, rename also on edges from bb whose single pred
37247 is outer_loop->header. Fix typo in function comment.
37249 2017-01-09 Martin Sebor <msebor@redhat.com>
37252 * asan.c (asan_emit_stack_protection): Increase local buffer size
37253 to avoid snprintf truncation warning.
37255 2017-01-09 Andrew Pinski <apinski@cavium.com>
37257 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
37258 to reference thunderx2t99 for the tuning structure
37259 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
37261 (thunderx2t99_extra_costs): This.
37262 * config/aarch64/aarch64-tune.md: Regenerate.
37263 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
37264 (vulcan_addrcost_table): This.
37265 (vulcan_regmove_cost): Rename to ...
37266 (thunderx2t99_regmove_cost): This.
37267 (vulcan_vector_cost): Rename to ...
37268 (thunderx2t99_vector_cost): this.
37269 (vulcan_branch_cost): Rename to ...
37270 (thunderx2t99_branch_cost): This.
37271 (vulcan_tunings): Rename to ...
37272 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
37273 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
37275 2017-01-09 Martin Jambor <mjambor@suse.cz>
37279 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
37280 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
37281 (propagate_vr_accross_jump_function): Use the above function for all
37282 value range computations for pass-through jump functions and type
37283 converasion from explicit value range values.
37284 (ipcp_propagate_stage): Do not attempt to deduce types of formal
37285 parameters from TYPE_ARG_TYPES.
37286 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
37287 (ipa_write_node_info): Stream type of the actual argument.
37288 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
37290 2017-01-09 Martin Liska <mliska@suse.cz>
37293 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
37294 (lookup_compiler): Do not show error message with have_E.
37296 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37298 PR tree-optimization/78938
37299 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
37300 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
37301 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
37302 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
37305 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37307 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
37310 2017-01-09 Richard Biener <rguenther@suse.de>
37312 PR tree-optimization/78997
37313 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
37314 name condition properly.
37316 2017-01-09 Richard Biener <rguenther@suse.de>
37319 * dwarf2out.c (is_cxx): New overload with context.
37320 (is_naming_typedef_decl): Use it.
37322 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37324 * invoke.texi (Option Summary): Correct spacing in option lists
37325 and add line breaks to fix over-long lines.
37327 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37329 PR middle-end/17660
37331 * extend.texi (Common Variable Attributes): Add xref to GCC
37332 Internals manual to explain mode attribute keywords.
37334 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37337 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
37338 and Preprocessor Options.
37339 (Options for Linking): Document -pthread here....
37340 (RS/6000 and PowerPC Options): ...not here.
37341 (Solaris 2 Options): ...or here.
37342 * doc/cppopts.texi: Document -pthread.
37344 2017-01-08 Martin Sebor <msebor@redhat.com>
37346 PR middle-end/77708
37347 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
37348 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
37349 New member functions.
37350 (format_directive): Used them.
37352 (pass_sprintf_length::handle_gimple_call): Same.
37353 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
37354 to avoid truncation for any argument.
37355 (extract_affine_mul): Same.
37356 * tree.c (get_file_function_name): Same.
37358 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37360 PR middle-end/77484
37361 * predict.def (PRED_INDIR_CALL): Set to 86.
37363 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37365 PR preprocessor/54124
37366 * doc/cppopts.texi: Reformat -d subtable to list the full name
37367 of the options. Add cross-reference to the docs for the general
37368 compiler -d options.
37369 * doc/invoke.texi (Developer Options): Add cross-reference to the
37370 preprocessor-specific -d option documentation.
37372 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37374 PR preprocessor/13498
37375 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
37376 redudant material, and reflect new command-line options.
37377 (System Headers): Likewise.
37379 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37381 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
37382 -isystem, and -idirafter. Copy-edit.
37383 * doc/cppopts.texi: Copy-edit. Remove contradiction about
37384 default for -ftrack-macro-expansion. Delete obsolete and
37385 badly-formatted implementation details about -fdebug-cpp output.
37386 * doc/cppwarnopts.texi: Copy-edit.
37388 2017-01-07 David Malcolm <dmalcolm@redhat.com>
37391 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37392 that the transition from a max line width >= 1<<10 to narrower
37393 lines works correctly.
37395 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
37397 * doc/options.texi (PerFunction): New.
37398 * opt-functions.awk (switch_flags): Map both Optimization and
37399 PerFunction to CL_OPTIMIZATION.
37400 * opth-gen.awk: Test for PerFunction flag along with
37402 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
37403 it only when the latter is present. Skip those that don't in
37404 the hash function generator.
37405 * common.opt (fvar-tracking): Mark as PerFunction instead of
37407 (fvar-tracking-assignments): Likewise.
37408 (fvar-tracking-assignments-toggle): Likewise.
37409 (fvar-tracking-uninit): Likewise.
37411 2017-01-07 Jakub Jelinek <jakub@redhat.com>
37413 PR translation/79018
37414 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
37417 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
37420 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
37421 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
37422 TARGET_LONG_JUMP_TABLE_OFFSETS.
37423 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
37424 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
37425 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
37426 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
37427 * config/m68k/m68k.md (tablejump expander): Likewise.
37428 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
37429 TARGET_LONG_JUMP_TABLE_OFFSETS.
37430 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
37431 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
37433 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
37434 David Holsgrove <david.holsgrove@xilinx.com>
37436 * common/config/microblaze/microblaze-common.c
37437 (TARGET_EXCEPT_UNWIND_INFO): Remove.
37438 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
37440 * config/microblaze/microblaze.c (microblaze_must_save_register)
37441 (microblaze_expand_epilogue, microblaze_return_addr): Handle
37443 (microblaze_eh_return): New function.
37444 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
37445 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
37446 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
37447 * config/microblaze/microblaze.md (eh_return): New pattern.
37449 2017-01-06 Jakub Jelinek <jakub@redhat.com>
37451 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
37452 GCC_DIAGNOSTIC_STRINGIFY): Define.
37454 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
37456 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37458 * config/arm/arm.md (<mcrr>): New.
37460 * config/arm/arm.c (arm_arch5te): New.
37461 (arm_option_override): Set arm_arch5te.
37462 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
37464 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
37465 (arm_mcrr_qualifiers): ... this. New.
37466 (MRRC_QUALIFIERS): Define to...
37467 (arm_mrrc_qualifiers): ... this. New.
37468 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
37470 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
37471 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
37472 (MRRCI, mrrc, MRRC): New.
37473 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
37474 VUNSPEC_MRRC2): New.
37476 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37478 * config/arm/arm.md (<mcr>): New.
37480 * config/arm/arm.c (arm_coproc_builtin_available): Add
37481 support for mcr, mrc, mcr2 and mrc2.
37482 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
37483 (arm_mcr_qualifiers): ... this. New.
37484 (MRC_QUALIFIERS): Define to ...
37485 (arm_mrc_qualifiers): ... this. New.
37486 (MCR_QUALIFIERS): Define to ...
37487 (arm_mcr_qualifiers): ... this. New.
37488 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
37490 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
37491 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
37492 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
37493 VUNSPEC_MRC2): New.
37495 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37497 * config/arm/arm.md (*ldc): New.
37501 * config/arm/arm.c (arm_coproc_builtin_available): Add
37502 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
37503 (arm_coproc_ldc_stc_legitimate_address): New.
37504 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
37505 'qualifier_const_pointer'.
37506 (LDC_QUALIFIERS): Define to...
37507 (arm_ldc_qualifiers): ... this. New.
37508 (STC_QUALIFIERS): Define to...
37509 (arm_stc_qualifiers): ... this. New.
37510 * config/arm/arm-protos.h
37511 (arm_coproc_ldc_stc_legitimate_address): New.
37512 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
37513 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
37514 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
37515 stc2, stcl, stc2l): New.
37516 * config/arm/constraints.md (Uz): New.
37517 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
37518 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
37519 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
37520 VUNSPEC_STC2L): New.
37522 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37524 * config/arm/arm.md (<cdp>): New.
37525 * config/arm/arm.c (neon_const_bounds): Rename this ...
37526 (arm_const_bounds): ... this.
37527 (arm_coproc_builtin_available): New.
37528 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
37529 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
37530 (CDP_QUALIFIERS): Define to...
37531 (arm_cdp_qualifiers): ... this. New.
37533 (arm_expand_builtin_args): Add case for 6 arguments.
37534 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
37535 (arm_const_bounds): ... this.
37536 (arm_coproc_builtin_available): New.
37537 * config/arm/arm_acle.h (__arm_cdp): New.
37539 * config/arm/arm_acle_builtins.def (cdp): New.
37541 * config/arm/iterators.md (CDPI,CDP,cdp): New.
37542 * config/arm/neon.md: Rename all 'neon_const_bounds' to
37543 'arm_const_bounds'.
37544 * config/arm/types.md (coproc): New.
37545 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
37546 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
37547 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
37548 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
37550 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37552 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
37553 (UBINOP_QUALIFIERS): New.
37555 (acle_builtin_data): New. Change comment.
37556 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
37557 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
37558 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
37559 arm_acle_builtins.def.
37560 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
37561 (arm_init_acle_builtins): New.
37562 (CRC32_BUILTIN): Remove.
37563 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
37564 crc32cb, crc32ch and crc32cw.
37565 (arm_init_crc32_builtins): Remove.
37566 (arm_init_builtins): Use arm_init_acle_builtins rather
37567 than arm_init_crc32_builtins.
37568 (arm_expand_acle_builtin): New.
37569 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
37570 * config/arm/arm_acle_builtins.def: New.
37572 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37574 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
37575 (arm_builtin_datum): ... this.
37576 (arm_init_neon_builtin): Rename to ...
37577 (arm_init_builtin): ... this. Add a new parameters PREFIX
37578 and USE_SIG_IN_NAME.
37579 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
37580 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
37581 'arm_builtin_datum'.
37582 (arm_init_vfp_builtins): Likewise.
37583 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
37584 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
37585 (arm_expand_neon_args): Rename to ...
37586 (arm_expand_builtin_args): ... this. Rename builtin_arg
37587 enum values and differentiate between ARG_BUILTIN_MEMORY
37588 and ARG_BUILTIN_NEON_MEMORY.
37589 (arm_expand_neon_builtin_1): Rename to ...
37590 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
37591 values, arm_expand_builtin_args and add bool parameter NEON.
37592 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
37593 (arm_expand_vfp_builtin): Likewise.
37594 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
37596 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37598 PR middle-end/77484
37599 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
37600 * predict.c (tree_estimate_probability_bb): Reverse direction of
37601 polymorphic call predictor.
37603 2017-01-06 David Malcolm <dmalcolm@redhat.com>
37605 * passes.c (execute_one_pass): Split out pass-skipping logic into...
37606 (determine_pass_name_match): ...this new function and...
37607 (should_skip_pass_p): ...this new function.
37609 2017-01-06 Nathan Sidwell <nathan@acm.org>
37611 * ipa-visibility.c (function_and_variable_visibility): Reformat
37612 comments and long lines. Remove extrneous if.
37613 * symtab.c (symtab_node::make_decl_local): Fix code format.
37614 (symtab_node::set_section_for_node): Fix comment typo.
37616 2017-01-06 Martin Liska <mliska@suse.cz>
37619 * lra-constraints.c: Rename invariant to lra_invariant.
37620 * predict.c (set_even_probabilities): Initialize e to NULL.
37622 2017-01-05 Martin Sebor <msebor@redhat.com>
37624 PR tree-optimization/78910
37625 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
37626 (format_integer): Correct off-by-one error in the handling
37627 of precision with negative numbers in signed conversions..
37629 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
37631 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
37633 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37635 PR tree-optimization/71016
37636 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
37637 factor_out_conditional_conversion. Formatting fix.
37638 (factor_out_conditional_conversion): Add cond_stmt argument.
37639 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
37640 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
37643 2017-01-05 David Malcolm <dmalcolm@redhat.com>
37645 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
37646 read-rtl-function.o, and selftest-rtl.o.
37647 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
37648 (selftest::aarch64_test_loading_full_dump): New function.
37649 (selftest::aarch64_run_selftests): New function.
37650 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
37651 selftest::aarch64_run_selftests.
37652 * config/i386/i386.c
37653 (selftest::ix86_test_loading_dump_fragment_1): New function.
37654 (selftest::ix86_test_loading_call_insn): New function.
37655 (selftest::ix86_test_loading_full_dump): New function.
37656 (selftest::ix86_test_loading_unspec): New function.
37657 (selftest::ix86_run_selftests): Call the new functions.
37658 * emit-rtl.c (maybe_set_max_label_num): New function.
37659 * emit-rtl.h (maybe_set_max_label_num): New decl.
37660 * function.c (instantiate_decls): Guard call to
37661 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
37662 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
37664 * gensupport.c (gen_reader::gen_reader): Pass "false"
37665 for new "compact" param of rtx_reader.
37666 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
37667 rather than an empty string for NULL strings.
37668 * read-md.c: Potentially include config.h rather than bconfig.h.
37669 Wrap include of errors.h with #ifdef GENERATOR_FILE.
37670 (have_error): New global, copied from errors.c.
37671 (md_reader::read_name): Rename to...
37672 (md_reader::read_name_1): ...this, adding "out_loc" param,
37673 and converting "missing name or number" to returning false, rather
37675 (md_reader::read_name): Reimplement in terms of read_name_1.
37676 (md_reader::read_name_or_nil): New function.
37677 (md_reader::read_string): Handle "(nil)" by returning NULL.
37678 (md_reader::md_reader): Add new param "compact".
37679 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
37680 (md_reader::read_file): New method.
37681 * read-md.h (md_reader::md_reader): Add new param "compact".
37682 (md_reader::read_file): New method.
37683 (md_reader::is_compact): New accessor.
37684 (md_reader::read_name): Convert return type from void to file_location.
37685 (md_reader::read_name_or_nil): New decl.
37686 (md_reader::read_name_1): New decl.
37687 (md_reader::m_compact): New field.
37688 (noop_reader::noop_reader): Pass "false" for new "compact" param
37690 (rtx_reader::rtx_reader): Add new "compact" param.
37691 (rtx_reader::read_rtx_operand): Make virtual and convert return
37692 type from void to rtx.
37693 (rtx_reader::read_until): New decl.
37694 (rtx_reader::handle_any_trailing_information): New virtual function.
37695 (rtx_reader::postprocess): New virtual function.
37696 (rtx_reader::finalize_string): New virtual function.
37697 (rtx_reader::m_in_call_function_usage): New field.
37698 (rtx_reader::m_reuse_rtx_by_id): New field.
37699 * read-rtl-function.c: New file.
37700 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
37701 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
37702 (selftest::verify_three_block_rtl_cfg): New decl.
37703 * read-rtl-function.h: New file.
37704 * read-rtl.c: Potentially include config.h rather than bconfig.h.
37705 For host, include function.h, memmodel.h, and emit-rtl.h.
37706 (one_time_initialization): New function.
37707 (struct compact_insn_name): New struct.
37708 (compact_insn_names): New array.
37709 (find_code): Handle insn codes in compact dumps.
37710 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
37711 (bind_subst_iter_and_attr): Likewise.
37712 (add_condition_to_string): Likewise.
37713 (add_condition_to_rtx): Likewise.
37714 (apply_attribute_uses): Likewise.
37715 (add_current_iterators): Likewise.
37716 (apply_iterators): Likewise.
37717 (initialize_iterators): Guard usage of apply_subst_iterator with
37718 #ifdef GENERATOR_FILE.
37719 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
37720 (md_reader::read_mapping): Likewise.
37721 (add_define_attr_for_define_subst): Likewise.
37722 (add_define_subst_attr): Likewise.
37723 (read_subst_mapping): Likewise.
37724 (check_code_iterator): Likewise.
37725 (rtx_reader::read_rtx): Likewise. Move one-time initialization
37727 (one_time_initialization): New function.
37728 (rtx_reader::read_until): New method.
37729 (read_flags): New function.
37730 (parse_reg_note_name): New function.
37731 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
37732 Handle reuse_rtx ids.
37733 Wrap iterator lookup within #ifdef GENERATOR_FILE.
37734 Add parsing support for RTL dumps, mirroring the special-cases in
37735 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
37736 values, and calling handle_any_trailing_information.
37737 (rtx_reader::read_rtx_operand): Convert return type from void
37738 to rtx, returning return_rtx. Handle case 'e'. Call
37739 finalize_string on XSTR and XTMPL fields.
37740 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
37741 "(nil)" values were omitted. Call the postprocess vfunc on the
37743 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
37744 class ctor. Initialize m_in_call_function_usage. Call
37745 one_time_initialization.
37746 * rtl-tests.c (selftest::test_uncond_jump): Call
37747 set_new_first_and_last_insn.
37748 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
37749 * selftest-rtl.c: New file.
37750 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
37751 (selftest::get_insn_by_uid): New decl.
37752 * selftest-run-tests.c (selftest::run_tests): Call
37753 read_rtl_function_c_tests.
37754 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
37755 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
37758 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
37760 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
37761 operands in a special way. Assert that pos+len <= mode precision.
37763 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37765 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
37766 3 argument Alias with unlimited for the negative form.
37767 (fno-vect-cost-model): Removed.
37769 2017-01-05 Martin Liska <mliska@suse.cz>
37771 * hsa-gen.c (gen_hsa_divmod): New function.
37772 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
37774 2017-01-05 Martin Liska <mliska@suse.cz>
37777 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
37780 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37782 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
37783 small constant length operands.
37785 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37787 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
37788 between loop iterations.
37790 2017-01-05 Martin Liska <mliska@suse.cz>
37793 * gimplify.c (gimplify_decl_expr): Compare to
37794 asan_poisoned_variables instread of checking flags.
37795 (gimplify_target_expr): Likewise.
37796 (gimplify_expr): Likewise.
37797 (gimplify_function_tree): Conditionally initialize
37798 asan_poisoned_variables.
37800 2017-01-04 Jeff Law <law@redhat.com>
37802 PR tree-optimizatin/78812
37803 * rtl.h (contains_mem_rtx_p): Prototype.
37804 * ifcvt.c (containts_mem_rtx_p): Move from here to...
37805 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
37806 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
37807 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
37808 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
37810 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37812 * input.c (assert_char_at_range): Default-initialize actual_range.
37814 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37816 * df-scan.c (df_ref_create_structure): Make regno unsigned,
37817 to match the caller.
37819 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37821 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
37822 insns after final jump in test to emit dummy move.
37824 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37826 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
37827 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
37829 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37831 * multiple_target.c (create_dispatcher_calls): Init e_next.
37832 * tree-ssa-loop-split.c (split_loop): Init border.
37833 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
37836 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
37841 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
37842 (altivec_register_operand): Do not return true if the operand
37843 contains a SUBREG mixing SImode and SFmode.
37844 (vsx_register_operand): Likewise.
37845 (vsx_reg_sfsubreg_ok): New predicate.
37846 (vfloat_operand): Do not return true if the operand contains a
37847 SUBREG mixing SImode and SFmode.
37848 (vint_operand): Likewise.
37849 (vlogical_operand): Likewise.
37850 (gpc_reg_operand): Likewise.
37851 (int_reg_operand): Likewise.
37852 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
37853 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
37854 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
37856 (rs6000_emit_move_si_sf_subreg): New helper function.
37857 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
37858 fixup SUBREGs involving SImode and SFmode.
37859 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
37860 numbers for the new peephole2 optimization.
37861 (peephole2 for SFmode unions): New peephole2 to optimize cases in
37862 the GLIBC math library that do AND/IOR/XOR operations on single
37863 precision floating point.
37864 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
37865 target macros to say whether we need to avoid SUBREGs mixing
37867 (TARGET_ALLOW_SF_SUBREG): Likewise.
37868 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
37869 (UNSPEC_SI_FROM_SF): Likewise.
37870 (iorxor): Change spacing.
37871 (and_ior_xor): New iterator for AND, IOR, and XOR.
37872 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
37873 (movdi_from_sf_zero_ext): Likewise.
37874 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
37875 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
37876 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
37877 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
37878 (fms<mode>4): Likewise.
37879 (fnma<mode>4): Likewise.
37880 (fnms<mode>4): Likewise.
37881 (nfma<mode>4): Likewise.
37882 (nfms<mode>4): Likewise.
37884 2017-01-04 Marek Polacek <polacek@redhat.com>
37887 * doc/invoke.texi: Document -Wpointer-compare.
37889 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37891 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
37894 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
37895 descriptions for -gdwarf-5 and emit them as uleb128 instead of
37898 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
37901 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
37902 documentation of the powerpc_popcntb_ok attribute.
37903 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
37904 code to issue warning messages if a requested CPU configuration is
37905 not supported by the binary (assembler and loader) toolchain.
37906 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
37907 made to define a built-in function that has been disabled.
37908 (paired_init_builtins): Add assertion to prevent ICE if attempt is
37909 made to define a built-in function that has been disabled.
37910 (altivec_init_builtins): Add comment explaining why definition
37911 of the DST built-in functions is not preceded by an assertion
37912 check. Add assertions to prevent ICE if attempts are made to
37913 define an altivec predicate or an abs* built-in function that has
37915 (htm_init_builtins): Add comment explaining why definition of the
37916 htm built-in functions is not preceded by an assertion check.
37918 2017-01-04 Jeff Law <law@redhat.com>
37920 PR tree-optimizatin/67955
37921 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
37922 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
37923 the points-to solution does not include pt_null. Use DECL_PT_UID
37926 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
37928 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
37929 Use gen_int_mode instead of gen_lopwart for const_int operands.
37931 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37933 PR tree-optimization/71563
37934 * match.pd: Simplify X << Y into X if Y is known to be 0 or
37935 out of range value - has low bits known to be zero.
37937 2017-01-04 Alan Modra <amodra@gmail.com>
37939 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
37940 * configure: Regenerate.
37941 * config.in: Regenerate.
37943 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37946 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
37947 a substring of the message, but strcmp with the whole message. Ifdef
37948 ENABLE_NLS, translate the message first using dgettext.
37950 2017-01-03 Jeff Law <law@redhat.com>
37952 PR tree-optimizatin/78856
37953 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
37954 (mark_threaded_blocks): Remove code to truncate thread paths that
37955 cross multiple loop headers. Instead invalidate the cached loop
37956 iteration information and handle case of a thread path walking
37957 into an irreducible region.
37959 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
37962 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
37963 assertions. Add support for doing the signbit if the IEEE 128-bit
37964 floating point value is in a GPR.
37965 * config/rs6000/rs6000.md (Fsignbit): Delete.
37966 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
37967 Update the length attribute if the value is in a GPR.
37968 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
37969 the sign or zero extension instruction, since the value is always 0/1.
37970 (signbit<mode>2_dm2): Delete using <Fsignbit>.
37973 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
37974 extracting SImode to a GPR register so that we can generate a
37975 store, limit the vector to be in a traditional Altivec register
37976 for the vextuwrx instruction.
37978 2017-01-03 Ian Lance Taylor <iant@google.com>
37980 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
37982 2017-01-03 Martin Sebor <msebor@redhat.com>
37984 PR tree-optimization/78696
37985 * gimple-ssa-sprintf.c (format_floating): Correct handling of
37986 precision. Use MPFR for %f for greater fidelity. Correct handling
37988 (pass_sprintf_length::compute_format_length): Set width and precision
37989 specified by asrerisk to void_node for vararg functions.
37990 (try_substitute_return_value): Adjust dump output.
37992 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
37994 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
37996 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
37998 * doc/invoke.texi (SPARC options): Document -mlra as the default.
37999 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
38000 -mlra/-mno-lra was passed to the compiler.
38002 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
38004 PR rtl-optimization/65618
38005 * emit-rtl.c (try_split): Move initialization of "before" and
38006 "after" to just before the call to emit_insn_after_setloc.
38008 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
38010 * doc/md.texi (Standard Names): Remove reference to Java frontend.
38012 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
38014 * dwarf2out.c (gen_enumeration_type_die): When
38015 -gno-strict-dwarf, add a DW_AT_encoding attribute.
38017 2017-01-03 Jakub Jelinek <jakub@redhat.com>
38019 PR tree-optimization/78965
38020 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
38021 Change first argument from const call_info & to call_info &. For %n
38022 set info.nowrite to false.
38024 PR middle-end/78901
38025 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
38026 possibly throwing calls.
38028 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
38029 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
38030 and fns handling, rather than in a separate case SSA_NAME.
38032 2017-01-02 Jeff Law <law@redhat.com>
38034 * config/darwin-driver.c (darwin_driver_init): Const-correctness
38035 fixes for first_period and second_period variables.
38037 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
38040 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
38041 (*insvqi_1): New insn pattern.
38042 (*insvqi_1_mem_rex64): Ditto.
38043 (*insvqi_2): Ditto.
38044 (*insvqi_3): Rename from *insvqi.
38046 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
38048 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
38050 * doc/cfg.texi (Edges): Remove reference to Java.
38051 (Maintaining the CFG): Ditto.
38053 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38055 PR middle-end/77674
38056 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38057 transparent aliases.
38059 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38061 PR middle-end/77484
38062 * predict.def (PRED_CALL): Update hitrate.
38063 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38064 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38065 into direct/indirect/polymorphic variants.
38067 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38069 Update copyright years.
38071 * gcc.c (process_command): Update copyright notice dates.
38072 * gcov-dump.c (print_version): Ditto.
38073 * gcov.c (print_version): Ditto.
38074 * gcov-tool.c (print_version): Ditto.
38075 * gengtype.c (create_file): Ditto.
38076 * doc/cpp.texi: Bump @copying's copyright year.
38077 * doc/cppinternals.texi: Ditto.
38078 * doc/gcc.texi: Ditto.
38079 * doc/gccint.texi: Ditto.
38080 * doc/gcov.texi: Ditto.
38081 * doc/install.texi: Ditto.
38082 * doc/invoke.texi: Ditto.
38084 Copyright (C) 2017 Free Software Foundation, Inc.
38086 Copying and distribution of this file, with or without modification,
38087 are permitted in any medium without royalty provided the copyright
38088 notice and this notice are preserved.