1 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3 * tree-vector-builder.h
4 (tree_vector_builder::new_binary_operation): Declare.
5 * tree-vector-builder.c
6 (tree_vector_builder::new_binary_operation): New function.
7 * fold-const.c (fold_relational_const): Use it.
8 (const_binop): Likewise. Check that both input vectors have
9 the same number of elements, thus excluding things like WIDEN_SUM.
10 Check whether it is possible to operate directly on the encodings
13 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
15 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
16 new_unary_operation, operating only on the encoded elements.
17 (const_unop): Likewise.
18 (exact_inverse): Likewise.
19 (distributes_over_addition_p): New function.
20 (const_binop): Use tree_vector_builder and new_unary_operation
21 for combinations of VECTOR_CST and INTEGER_CST. Operate only
22 on the encoded elements unless the encoding is strided and the
23 operation does not distribute over addition.
24 (fold_convert_const): Use tree_vector_builder and
25 new_unary_operation. Operate only on the encoded elements
26 for truncating integer conversions, or for non-stepped encodings.
28 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
30 * config/sparc/sparc.c: Include tree-vector-builder.h.
31 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
32 * expmed.c: Include tree-vector-builder.h.
33 (make_tree): Use tree_vector_builder instead of build_vector.
34 * fold-const.c: Include tree-vector-builder.h.
35 (const_binop): Use tree_vector_builder instead of build_vector.
36 (const_unop): Likewise.
37 (native_interpret_vector): Likewise.
38 (fold_vec_perm): Likewise.
39 (fold_ternary_loc): Likewise.
40 * gimple-fold.c: Include tree-vector-builder.h.
41 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
43 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
44 (simplify_vector_constructor): Use tree_vector_builder instead
46 * tree-vect-generic.c: Include tree-vector-builder.h.
47 (add_rshift): Use tree_vector_builder instead of build_vector.
48 (expand_vector_divmod): Likewise.
49 (optimize_vector_constructor): Likewise.
50 * tree-vect-loop.c: Include tree-vector-builder.h.
51 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
52 of build_vector. Explicitly use a stepped encoding for
54 * tree-vect-slp.c: Include tree-vector-builder.h.
55 (vect_get_constant_vectors): Use tree_vector_builder instead
57 (vect_transform_slp_perm_load): Likewise.
58 (vect_schedule_slp_instance): Likewise.
59 * tree-vect-stmts.c: Include tree-vector-builder.h.
60 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
61 (vect_gen_perm_mask_any): Likewise.
62 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
63 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
65 (build_vector_from_val): Likewise. Explicitly use a duplicate
68 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
70 * doc/generic.texi (VECTOR_CST): Describe new representation of
72 * vector-builder.h: New file.
73 * tree-vector-builder.h: Likewise.
74 * tree-vector-builder.c: Likewise.
75 * Makefile.in (OBJS): Add tree-vector-builder.o.
76 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
77 * tree-core.h (tree_base): Add a vector_cst field to the u union.
78 (tree_vector): Change the number of elements to
79 vector_cst_encoded_nelts.
80 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
81 (VECTOR_CST_ELTS): Delete.
82 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
83 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
84 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
85 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
86 (VECTOR_CST_ENCODED_ELT): Likewise.
87 (vector_cst_encoded_nelts): New function.
88 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
89 VECTOR_CST_NELTS_PER_PATTERN as arguments.
90 (vector_cst_int_elt, vector_cst_elt): Declare.
91 * tree.c: Include tree-vector-builder.h.
92 (tree_code_size): Abort if passed VECTOR_CST.
93 (tree_size): Update for new VECTOR_CST layout.
94 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
95 VECTOR_CST_NELTS_PER_PATTERN as arguments.
96 (build_vector): Use tree_vector_builder.
97 (vector_cst_int_elt, vector_cst_elt): New functions.
98 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
99 encoded elements and then create the vector in the canonical form.
100 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
101 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
102 (tree_c_tests): Call test_vector_cst_patterns.
103 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
105 (hash_tree): Likewise.
106 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
107 (streamer_write_tree_header): Likewise.
108 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
109 (streamer_alloc_tree): Likewise. Update call to make_vector.
110 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
112 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
114 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
115 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
116 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
118 2017-12-07 Bin Cheng <bin.cheng@arm.com>
119 Richard Biener <rguenther@suse.de>
121 PR tree-optimization/81303
122 * Makefile.in (gimple-loop-interchange.o): New object file.
123 * common.opt (floop-interchange): Reuse the option from graphite.
124 * doc/invoke.texi (-floop-interchange): Ditto. New document for
125 -floop-interchange and mention it for -O3.
126 * opts.c (default_options_table): Enable -floop-interchange at -O3.
127 * gimple-loop-interchange.cc: New file.
128 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
129 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
130 * passes.def (pass_linterchange): New pass.
131 * timevar.def (TV_LINTERCHANGE): New time var.
132 * tree-pass.h (make_pass_linterchange): New declaration.
133 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
134 interchange. Record IV before/after increment in new parameters.
135 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
136 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
138 (check_reduction_path): ...New function here.
139 * tree-vectorizer.h (check_reduction_path): New declaration.
141 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
144 PR rtl-optimization/80818
145 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
146 always early clobbered.
147 * lra-lives.c (process_bb_lives): Check input hard regs for early
148 clobbered non-operand hard reg.
150 2017-12-07 Jakub Jelinek <jakub@redhat.com>
153 * tree-cfg.c (verify_gimple_assign_binary): Don't require
154 types_compatible_p, just that TYPE_MODE is the same.
156 2017-12-07 Martin Sebor <msebor@redhat.com>
159 * attribs.c (empty_attribute_table): Initialize new member of
160 struct attribute_spec.
161 (decl_attributes): Add argument. Handle mutually exclusive
162 combinations of attributes.
163 (selftests::test_attribute_exclusions): New function.
164 (selftests::attribute_c_tests): Ditto.
165 * attribs.h (decl_attributes): Add default argument.
166 * selftest.h (attribute_c_tests): Declare.
167 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
168 * tree-core.h (attribute_spec::exclusions, exclude): New type and
170 * doc/extend.texi (Common Function Attributes): Update const and pure.
171 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
172 of struct attribute_spec.
173 * config/arc/arc.c (arc_attribute_table): Same.
174 * config/arm/arm.c (arm_attribute_table): Same.
175 * config/avr/avr.c ( avr_attribute_table): Same.
176 * config/bfin/bfin.c (bfin_attribute_table): Same.
177 * config/cr16/cr16.c (cr16_attribute_table): Same.
178 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
179 * config/h8300/h8300.c (h8300_attribute_table): Same.
180 * config/i386/i386.c (ix86_attribute_table): Same.
181 * config/ia64/ia64.c (ia64_attribute_table): Same.
182 * config/m32c/m32c.c (m32c_attribute_table): Same.
183 * config/m32r/m32r.c (m32r_attribute_table): Same.
184 * config/m68k/m68k.c (m68k_attribute_table): Same.
185 * config/mcore/mcore.c (mcore_attribute_table): Same.
186 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
187 * config/mips/mips.c (mips_attribute_table): Same.
188 * config/msp430/msp430.c (msp430_attribute_table): Same.
189 * config/nds32/nds32.c (nds32_attribute_table): Same.
190 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
191 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
192 * config/rl78/rl78.c (rl78__attribute_table): Same.
193 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
194 * onfig/rx/rx.c (rx_attribute_table): Same.
195 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
196 * config/sh/sh.c (sh_attribute_table): Same.
197 * config/sparc/sparc.c (sparc_attribute_table): Same.
198 * config/spu/spu.c (spu_attribute_table): Same.
199 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
200 * config/v850/v850.c (v850_attribute_table): Same.
201 * config/visium/visium.c (visium_attribute_table): Same.
203 2017-12-07 Tamar Christina <tamar.christina@arm.com>
206 * config/arm/arm.c (INCLUDE_STRING): Define.
207 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
208 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
211 2017-12-07 Michael Matz <matz@suse.de>
213 Add unroll and jam pass
215 * gimple-loop-jam.c: New file.
216 * Makefile.in (OBJS): Add gimple-loop-jam.o.
217 * common.opt (funroll-and-jam): New option.
218 * opts.c (default_options_table): Add unroll-and-jam at -O3.
219 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
220 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
221 * passes.def: Add pass_loop_jam.
222 * timevar.def (TV_LOOP_JAM): Add.
223 * tree-pass.h (make_pass_loop_jam): Declare.
224 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
225 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
226 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
227 to flow_loop_tree_node_add.
228 (duplicate_subloops, copy_loops_to): Append to sibling list.
229 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
230 * doc/invoke.texi (-funroll-and-jam): Document new option.
231 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
233 2017-12-07 Richard Biener <rguenther@suse.de>
235 PR tree-optimization/83296
236 PR tree-optimization/67769
237 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
238 flow sensitive info in an unrelated BB.
239 (value_replacement): Use reset_flow_sensitive_info.
240 (minmax_replacement): Reset flow sensitive info on the def
241 we move. Do not reset flow sensitive info in the whole BB
243 (abs_replacement): Likewise.
245 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
248 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
249 rs6000_cpu to the given -mcpu=, or to the default processor.
251 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
253 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
254 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
255 instead of rs6000_cpu_attr.
256 (rs6000_adjust_cost): Ditto.
257 (is_microcoded_insn): Ditto.
258 (rs6000_adjust_priority): Ditto.
259 (rs6000_issue_rate): Ditto.
260 (rs6000_use_sched_lookahead): Ditto.
261 (rs6000_use_sched_lookahead_guard): Ditto.
262 (rs6000_sched_reorder): Ditto.
263 (force_new_group): Ditto.
264 * config/rs6000/rs6000.md (cpu attribute): Ditto.
265 (group_ending_nop): Ditto.
267 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
269 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
270 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
271 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
272 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
274 (rs6000_reassociation_width): Ditto.
275 (rs6000_emit_epilogue): Ditto.
276 (rs6000_adjust_cost): Ditto.
277 (is_microcoded_insn): Ditto.
278 (is_cracked_insn): Ditto.
279 (rs6000_adjust_priority): Ditto.
280 (rs6000_sched_reorder): Ditto.
281 (rs6000_sched_reorder2): Ditto.
282 (insn_must_be_first_in_group): Ditto.
283 (insn_must_be_last_in_group): Ditto.
284 (rs6000_register_move_cost): Ditto.
285 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
288 2017-12-07 Julia Koval <julia.koval@intel.com>
290 * config.gcc: Add vaesintrin.h.
291 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI): New type.
292 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
293 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi): New builtins.
294 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
295 * config/i386/immintrin.h: Include vaesintrin.h.
296 * config/i386/sse.md (vaesdec_<mode>): New pattern.
297 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
298 _mm_aesdec_epi128): New intrinsics.
300 2017-12-06 David Malcolm <dmalcolm@redhat.com>
302 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
303 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
304 c-family/c-spellcheck.cc.
305 (best_macro_match::best_macro_match): Likewise.
307 (struct edit_distance_traits<cpp_hashnode *>): Move to
308 c-family/c-spellcheck.h.
309 (class best_macro_match): Likewise.
311 2017-12-06 Jakub Jelinek <jakub@redhat.com>
313 PR tree-optimization/83293
314 * gimple-ssa-strength-reduction.c (insert_initializers): Use
315 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
316 might insert into empty bb.
319 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
320 simplify for plus with :c added, and pointer_plus without that.
321 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
322 with undefined overflow and the conversion is not widening,
323 perform negation in utype and only convert to type afterwards.
324 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
325 simplify for plus with :c added, and pointer_plus without that.
326 If type is integral with undefined overflow and the conversion is
327 not widening, perform minus in utype and only convert to type
328 afterwards. Move the last pointer_diff_expr simplify into the
331 2017-12-06 Martin Sebor <msebor@redhat.com>
333 PR tree-optimization/82646
334 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
337 2017-12-06 Martin Sebor <msebor@redhat.com>
339 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
342 PR tree-optimization/83075
343 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
344 strncat/strncpy don't change length of source string.
346 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
349 2017-11-29 Martin Aberg <maberg@gaisler.com>
351 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
352 to prevent b2bst errata sequence.
353 (sqrtdf2_fix): Likewise.
355 2017-12-06 Jakub Jelinek <jakub@redhat.com>
357 PR tree-optimization/81945
358 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
359 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
360 to dest_cfun has orig_loop_num set, either remap it to the new
361 loop number if the loop got moved too, or clear it.
363 2017-12-05 Steve Ellcey <sellcey@cavium.com>
365 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
367 (thunderx2t99_nothing): New insn reservation.
368 (thunderx2t99_mrs): New insn reservation.
369 (thunderx2t99_multiple): New insn reservation.
370 (thunderx2t99_alu_basi): Add bfx to reservation.
371 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
373 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
376 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
378 2017-12-05 Bin Cheng <bin.cheng@arm.com>
380 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
381 tree-ssa-pre.c::remove_dead_inserted_code.
382 * tree-ssa-dce.h: New file.
383 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
384 (remove_dead_inserted_code): Move and rename to function
385 tree-ssa-dce.c::simple_dce_from_worklist.
386 (pass_pre::execute): Update use.
388 2017-12-05 Richard Biener <rguenther@suse.de>
390 PR tree-optimization/83277
391 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
392 to code-gen liveout vars.
394 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
396 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
397 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
398 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
401 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
403 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
405 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
406 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
409 2017-12-05 Richard Biener <rguenther@suse.de>
411 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL):
413 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
414 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
415 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
417 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
419 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
421 2017-12-05 Martin Liska <mliska@suse.cz>
422 Jakub Jelinek <jakub@redhat.com>
424 * doc/invoke.texi: Document the options.
425 * flag-types.h (enum sanitize_code): Add
426 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
427 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
428 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
429 * opts.c: Define new sanitizer options.
430 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
431 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
433 2017-12-05 Julia Koval <julia.koval@intel.com>
435 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
436 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
437 (ix86_handle_option): Handle -mavx512vnni.
438 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
439 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
440 * config/i386/i386-c (__AVX512VNNI__): New.
441 * config/i386/i386.c (ix86_target_string): Handle new option.
442 (ix86_valid_target_attribute_inner_p): Handle new option.
443 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
444 * config/i386/i386.opt (mavx512vnni): New option.
446 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
449 * athlon.md: Disable for generic.
450 * haswell.md: Enable for generic.
451 * i386.c (ix86_sched_init_global): Add core hooks for generic.
452 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
454 (ix86_adjust_cost): Move generic to haswell path.
456 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
458 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
459 instead of MEM_P in a couple more places. Fix formatting issues.
461 2017-12-04 Jim Wilson <jimw@sifive.com>
463 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
464 instead of GP_REG_LAST-1.
465 (riscv_adjust_libcall_cfi_prologue): Likewise.
466 (riscv_adjust_libcall_cri_epilogue): Likewise.
467 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
470 2017-12-04 Luis Machado <luis.machado@linaro.org>
472 * ipa-pure-const.c (check_decl): Add missing newline.
473 (state_from_flags): Likewise.
475 2017-12-04 Jeff Law <law@redhat.com>
477 PR tree-optimizatin/78496
478 * gimple-ssa-evrp-analyze.h
479 (evrp_range_analyzer::get_vr_values): Simplify.
480 * gimple-ssa-evrp-analyze.c: Corresponding changes.
481 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
482 and gimple-ssa-evrp-analyze.h.
483 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
484 (simplify_stmt_for_jump_threading): Copy a blob of code from
485 tree-vrp.c to use ranges to simplify statements.
486 (dom_opt_dom_walker::before_dom_children): Call
487 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
488 (dom_opt_dom_walker::after_dom_children): Similarly for
489 evrp_range_analyzer::leave.
490 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
493 * gimple-ssa-evrp-analyze.c
494 (evrp_range_analyzer::extract_range_from_stmt): Always use
495 vr_values::update_value_range so preexisting range info is
496 medged with new range info, even if the new range is VR_VARYING.
498 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
500 * combine.c: Adjust comment.
501 (use_crosses_set_p): Delete.
502 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
503 (try_combine): Ditto.
505 2017-12-04 Richard Biener <rguenther@suse.de>
507 PR tree-optimization/83255
508 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
509 Re-add zero-iteration check.
511 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
513 PR rtl-optimization/83245
514 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
515 hard registers as earlyclobber, also if not in an asm.
517 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
521 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
524 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
525 rs6000_cpu based on cpu_index, not tune_index.
527 2017-12-04 Richard Biener <rguenther@suse.de>
529 PR tree-optimization/83238
530 * graphite-scop-detection.c (scop_detection::merge_sese): Make
531 code match comment, rejecting invalid SESE regions.
533 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
535 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
536 require base operand is a REG_POINTER prior to reload on targets
537 with non-equivalent space registers.
539 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
541 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
542 (update_specialized_profile): Fix updating of counts.
543 (perhaps_add_new_callers): Likewise.
545 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
548 * x86-tune.def: Remove obsolette FIXMEs.
549 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
550 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
551 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
553 (X86_TUNE_PAD_RETURNS): Disable for generic.
555 2017-12-02 Jakub Jelinek <jakub@redhat.com>
557 PR tree-optimization/83170
558 PR tree-optimization/83241
559 * gimple-ssa-store-merging.c
560 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
561 gimple_vuse (ins_stmt) in case it has changed.
562 (imm_store_chain_info::output_merged_store): Likewise.
564 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
568 * tree-cfg.c (pass_warn_function_return::execute): Handle
569 __builtin_ubsan_handle_missing_return like __builtin_unreachable
570 with BUILTINS_LOCATION.
574 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
575 is BLKmode for vector field with vector raw mode, use TYPE_MODE
576 instead of DECL_MODE.
578 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
579 last argument to rtx pointer.
580 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
581 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
582 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
583 depending on the chosen ISAs.
584 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
585 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
586 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
588 * config/i386/sse.md (mov<mode>_internal): Likewise.
589 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
591 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
593 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
595 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
597 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
598 parameters from prototype.
599 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
600 parameters. Don't print a comment.
601 (emit_fusion_gpr_load): Adjust.
602 (emit_fusion_load_store): Adjust.
603 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
604 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
605 comment on the second line.
607 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
610 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
611 rs6000_cpu based on cpu_index, not tune_index.
613 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
615 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
616 which_alternative instead of which_alternative + 1.
617 (output_asm_insn): Print an extra tab if the template is short.
619 2017-12-01 Jim Wilson <jimw@sifive.com>
621 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
623 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
624 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
626 * doc/tm.texi: Regenerate.
628 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
631 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
632 whether we can allocate pseudos before trying to fix an address.
633 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
634 memory address is indexed or indirect.
635 (floatuns_<mode>si2_hw2): Likewise.
637 2017-12-01 Jason Merrill <jason@redhat.com>
639 * Makefile.in (TAGS): Add c-family/*.cc.
641 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
643 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
644 (call_value_insn): Likewise.
645 (sibcall_insn): Likewise.
646 (sibcall_value_insn): Likewise.
647 (movsi_aarch64): Likewise.
648 (movdi_aarch64): Likewise.
649 (add_losym_): Likewise.
650 (ldr_got_small_): Likewise.
651 (ldr_got_small_sidi): Likewise.
652 (ldr_got_small_28k_): Likewise.
653 (ldr_got_small_28k_sidi): Likewise.
654 * config/aarch64/aarch64.c (aarch64_print_address_internal):
655 Move output_addr_const to symbolic case. Add error check.
657 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
659 * config/s390/predicates.md (plus16_Q_operand): New predicate.
660 * config/s390/s390.md: Disable MVC merging peephole if it would
661 disable operand forwarding.
662 (new peephole2): Split MVCs if it would turn them into up to 2
665 2017-12-01 Richard Biener <rguenther@suse.de>
667 PR tree-optimization/83232
668 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
669 detection of same access. Instead of breaking the group here
670 do not consider the duplicate. Add comment explaining real fix.
672 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
674 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
676 2017-12-01 Sudakshina Das <sudi.das@arm.com>
678 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
680 2017-12-01 Jakub Jelinek <jakub@redhat.com>
682 * function.h (struct function): Remove cilk_frame_decl,
683 is_cilk_function and calls_cilk_spawn fields.
684 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
686 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
687 cilk_elemental field.
688 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
689 * target.def: Adjust comment.
690 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
691 Don't test cilk_elemental.
693 PR tree-optimization/83233
694 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
695 bswap_stat name for the struct.
698 * tree.h (SWITCH_BREAK_LABEL_P): Define.
699 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
700 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
701 SWITCH_BREAK_LABEL_P set on the label.
702 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
703 added for default case if it was missing and not all cases covered.
704 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
705 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
707 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
708 to avoid -Wimplicit-fallthrough warning.
709 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
710 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
712 PR tree-optimization/83221
713 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
715 (init_reassoc): Formatting fix.
718 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
719 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
723 * tree-cfg.c: Include asan.h.
724 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
726 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
728 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
730 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
732 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
733 vpcmpeqd instruction.
735 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
737 * config/i386/i386.c (standard_sse_constant_opcode): Fix
738 registers type for 128bit mode.
740 2017-11-30 Jakub Jelinek <jakub@redhat.com>
742 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
744 * gengtype-state.c (read_state_pair): Likewise.
745 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
746 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
747 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
748 * ifcvt.c (noce_try_store_flag_constants): Likewise.
749 * tree-ssa-ccp.c (ccp_finalize): Likewise.
750 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
751 * builtins.c (fold_builtin_3): Likewise.
752 * graphite-scop-detection.c
753 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
754 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
757 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
760 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
761 parameter. Return true if ignore_decl_rtl_set_p is true and other
762 conditions are satisfied.
763 * asan.h (asan_protect_global): Add new parameter.
764 * varasm.c (categorize_decl_for_section): Pass true as second parameter
765 to asan_protect_global calls.
767 2017-11-30 Jim Wilson <jimw@sifive.com>
769 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
770 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
771 -mstrict-align, add info on default value. Delete redundant lines for
772 -mabi. Add missing -mexplicit-relocs docs.
774 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
776 * config/arc/arc.md (trap): New pattern.
778 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
780 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
781 instruction to end into a delay slot.
782 * config/arc/arc.md (cond_delay_insn): Check if the instruction
783 can be placed into a delay slot against reg_note.
785 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
787 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
788 labels number of usages.
790 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
792 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
794 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
796 2017-11-30 Jakub Jelinek <jakub@redhat.com>
799 * internal-fn.c (expand_mul_overflow): Optimize unsigned
800 multiplication by power of 2 constant into two shifts + comparison.
802 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
805 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
807 2017-11-30 Richard Biener <rguenther@suse.de>
809 PR tree-optimization/83202
810 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
811 allow_peel argument and guard peeling.
812 (canonicalize_loop_induction_variables): Likewise.
813 (canonicalize_induction_variables): Pass false.
814 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
815 peeling from cunrolli.
817 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
819 * combine.c (try_combine): Print a message to dump file whenever
820 I0, I1, or I2 cannot be combined into I3.
822 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
824 PR rtl-optimization/83156
825 PR rtl-optimization/82621
826 * combine.c (try_combine): Don't split an I2 if one of the dests is
827 set again before I3. Allow unused dests.
829 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
831 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
833 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
835 PR rtl-optimization/80818
836 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
837 recursively. Use insn code for clobber.
838 (lra_set_insn_recog_data): Pass the new arg to
839 collect_non_operand_hard_regs.
840 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
842 (lra_update_insn_regno_info): Pass insn to
843 add_regs_to_insn_regno_info.
845 2017-11-29 Jim Wilson <jimw@sifive.com>
846 Andrew Waterman <andrew@sifive.com>
848 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
849 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
850 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
851 use SINGLE_SHIFT_COST.
852 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
853 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
855 2017-11-29 Julia Koval <julia.koval@intel.com>
857 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
858 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
859 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
860 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
861 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
862 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
863 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
864 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
865 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
866 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
867 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
868 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
869 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
870 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
871 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
872 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
873 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
874 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
875 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
876 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
877 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
878 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
879 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
880 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
881 __builtin_ia32_vpshldv_v2di_mask,
882 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
883 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
884 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
886 2017-11-29 Julia Koval <julia.koval@intel.com>
888 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
889 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
890 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
891 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
892 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
893 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
894 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
895 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
896 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
897 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
898 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
899 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
900 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
901 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
902 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
903 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
904 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
905 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
906 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
907 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
908 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
910 2017-11-29 Daniel Cederman <cederman@gaisler.com>
912 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
913 movsi_pic_gotdata_op instruction as a load for the UT699 errata
916 2017-11-29 Martin Aberg <maberg@gaisler.com>
918 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
919 to prevent b2bst errata sequence.
920 (sqrtdf2_fix): Likewise.
922 2017-11-29 Daniel Cederman <cederman@gaisler.com>
924 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
925 (div_sqrt_insn_p): New function.
926 (sparc_do_work_around_errata): Insert NOP instructions to
927 prevent sequences that could trigger the TN-0013 errata for
928 certain LEON3 processors.
929 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
930 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
931 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
932 (in_branch_delay): Prevent div and sqrt in delay slot if
934 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
936 2017-11-29 Daniel Cederman <cederman@gaisler.com>
938 * config/sparc/sparc.c (atomic_insn_p): New function.
939 (sparc_do_work_around_errata): Insert NOP instructions to
940 prevent sequences that could trigger the TN-0010 errata for
942 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
943 instruction referable in atomic_insns_p.
945 2017-11-29 Daniel Cederman <cederman@gaisler.com>
947 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
948 (atomic_compare_and_swap_leon3_1): Likewise.
951 2017-11-29 Daniel Cederman <cederman@gaisler.com>
953 * config/sparc/sparc.c (fpop_insn_p): New function.
954 (sparc_do_work_around_errata): Insert NOP instructions to
955 prevent sequences that could trigger the TN-0012 errata for
957 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
958 * config/sparc/sparc.md (fix_gr712rc): New attribute.
959 (in_branch_annul_delay): Prevent floating-point instructions
960 in delay slot of annulled integer branch.
962 2017-11-29 Richard Biener <rguenther@suse.de>
964 PR tree-optimization/83202
965 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
967 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
968 nodes vectorized to the same stmts multiple times.
969 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
970 (vect_analyze_slp_instance): Adjust.
971 (scalar_stmts_to_slp_tree_map_t): New typedef.
972 (vect_schedule_slp_instance): Add a map recording the SLP node
973 representing the vectorized stmts for a set of scalar stmts.
974 Avoid code-generating redundancies.
975 (vect_schedule_slp): Allocate map and pass it down.
977 2017-11-29 Nathan Sidwell <nathan@acm.org>
980 * tree.c (build_complex_type): Fix canonicalization. Only fill in
983 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
985 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
986 cases for printing LDP/STP memory addresses.
987 (aarch64_print_address_internal): Renamed from
988 aarch64_print_operand_address, added parameter, add Pmode check.
989 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
990 (aarch64_print_operand_address): Indirect to
991 aarch64_print_address_internal.
992 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
993 'y' operand output specifier.
995 2017-11-29 Jakub Jelinek <jakub@redhat.com>
998 * tree.c (build_simple_mem_ref_loc): Handle
999 get_addr_base_and_unit_offset returning a MEM_REF.
1002 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1005 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1008 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1010 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1012 (layout::move_to_column): Likewise.
1014 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1017 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1018 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1020 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1021 Martin Jambor <mjambor@suse.cz>
1024 * tree.h (expr_type_first_operand_type_p): Declare
1025 * tree.c (expr_type_first_operand_type_p): New function.
1026 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1027 (ipa_value_from_jfunc): Adjust declaration.
1028 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1029 Use it as result type for arithmetics, unless it is NULL in which case
1030 be more conservative.
1031 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1032 ipa_get_jf_pass_through_result.
1033 (propagate_vals_across_pass_through): Likewise.
1034 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1035 is to propagate_vals_across_pass_through.
1036 (propagate_constants_across_call): Pass PARM_TYPE to
1037 propagate_scalar_across_jump_function.
1038 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1039 ipa_value_from_jfunc.
1040 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1041 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1042 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1043 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1044 target_type, pass it to ipa_value_from_jfunc.
1045 (update_indirect_edges_after_inlining): Pass parameter type to
1046 try_make_edge_direct_simple_call.
1048 2017-11-28 Jeff Law <law@redhat.com>
1050 * gimple-ssa-evrp-analyze.c
1051 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1052 refine ranges if scev_initialized_p returns true.
1053 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1055 2017-11-28 Julia Koval <julia.koval@intel.com>
1057 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1058 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
1059 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
1060 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
1061 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
1062 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
1063 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
1064 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
1065 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
1066 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
1067 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
1068 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
1069 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
1070 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
1071 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
1072 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
1073 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
1074 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
1075 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
1076 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
1077 __builtin_ia32_vpshrd_v2di_mask): New builtins.
1078 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
1080 2017-11-28 Julia Koval <julia.koval@intel.com>
1082 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
1083 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
1084 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
1085 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
1086 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
1087 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
1088 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
1089 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
1090 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
1091 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
1092 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
1093 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
1094 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
1095 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
1096 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
1097 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
1098 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
1099 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
1100 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
1101 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
1102 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
1103 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
1104 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
1105 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
1106 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
1107 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
1108 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
1109 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
1110 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
1111 __builtin_ia32_vpshld_v2di_mask): New builtins.
1112 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1113 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
1115 2017-11-28 Richard Biener <rguenther@suse.de>
1117 PR tree-optimization/80776
1118 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
1120 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
1122 (evrp_range_analyzer::record_ranges_from_incoming_edges):
1123 If the incoming edge is an effective fallthru because the other
1124 edge only reaches a __builtin_unreachable () then record ranges
1125 derived from the controlling condition in SSA info.
1126 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
1127 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1129 2017-11-28 Olivier Hainque <hainque@adacore.com>
1131 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
1132 on mingw build hosts.
1134 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1137 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
1139 2017-11-28 Richard Biener <rguenther@suse.de>
1142 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
1143 copies generated from memcpy use a character array as reference
1146 2017-11-28 Julia Koval <julia.koval@intel.com>
1147 Sebastian Peryt <sebastian.peryt@intel.com>
1149 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
1150 c-family/c-cilkplus.o, c-family/array-notation-common.o,
1151 cilk-common.o, cilk.h, cilk-common.c): Remove.
1153 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
1154 * builtins.c (is_builtin_name): Remove cilkplus condition.
1155 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
1156 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
1157 cilk-builtins.def, cilkplus.def): Remove.
1158 * cif-code.def (CILK_SPAWN): Remove.
1159 * cilk-builtins.def: Delete.
1160 * cilk-common.c: Ditto.
1162 * cilkplus.def: Ditto.
1163 * config/darwin.h (fcilkplus): Delete.
1164 * cppbuiltin.c: Ditto.
1165 * doc/extend.texi: Remove cilkplus doc.
1166 * doc/generic.texi: Ditto.
1167 * doc/invoke.texi: Ditto.
1168 * doc/passes.texi: Ditto.
1169 * gcc.c (fcilkplus): Remove.
1170 * gengtype.c (cilk.h): Remove.
1171 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
1173 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
1175 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
1177 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
1178 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
1179 cilkplus conditions.
1180 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
1181 inline_read_section): Ditto.
1182 * ipa-inline-analysis.c (cilk.h): Remove.
1183 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
1184 * lto-wrapper.c (merge_and_complain, append_compiler_options,
1185 append_linker_options): Remove condition for fcilkplus.
1186 * lto/lto-lang.c (cilk.h): Remove.
1187 (lto_init): Remove condition for fcilkplus.
1188 * omp-expand.c (expand_cilk_for_call): Delete.
1189 (expand_omp_taskreg, expand_omp_for_static_chunk,
1190 expand_omp_for): Remove cilkplus
1192 (expand_cilk_for): Delete.
1193 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
1194 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
1195 execute_lower_omp, diagnose_sb_0): Ditto.
1196 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
1197 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1198 * tree-nested.c: Ditto.
1199 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
1200 (dump_generic_node): Ditto.
1201 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1202 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
1204 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
1206 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1208 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
1209 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
1210 for vector mode and !TARGET_SIMD.
1212 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1214 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
1216 * tree.h (SWITCH_LABELS): Remove.
1217 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
1218 assert SWITCH_BODY is non-NULL.
1219 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
1221 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
1223 PR tree-optimization/80788
1224 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
1225 has TREE_OVERFLOW set, call drop_tree_overflow.
1227 2017-11-28 Richard Biener <rguenther@suse.de>
1229 PR tree-optimization/83158
1230 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more
1233 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
1236 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
1237 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
1239 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
1241 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
1242 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
1244 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1247 * varasm.c (compare_constant): Take the mode of the constants into
1248 account when comparing floating point constants.
1250 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
1252 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
1253 from explicit instantiation of debug_helper.
1254 * vec.h (DEFINE_DEBUG_VEC): Ditto.
1256 2017-11-27 Richard Biener <rguenther@suse.de>
1258 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
1261 2017-11-27 Richard Biener <rguenther@suse.de>
1263 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
1264 (type_hash_canon): Also clear int_cst_hash_table entry for
1266 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
1268 2017-11-27 Tamar Christina <tamar.christina@arm.com>
1270 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1271 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1272 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1274 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1276 * hash-map.h (gt_cleare_cache): Avoid UB.
1278 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
1280 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
1281 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
1282 (decide_unroll_runtime_iterations): Likewise.
1283 (decide_unroll_stupid): Likewise.
1285 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1288 * config/i386/i386.md: Add a loop with incssp.
1290 2017-11-27 Martin Jambor <mjambor@suse.cz>
1292 PR tree-optimization/81248
1293 * tree-sra.c (splice_param_accesses): Remove size check.
1294 (decide_one_param_reduction): Fix size check.
1295 * gimple-pretty-print.c (dump_profile): Silence warning.
1296 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
1298 2017-11-27 Jakub Jelinek <jakub@redhat.com>
1301 * dbxout.c (lastlineno): New variable.
1302 (dbx_debug_hooks): Use dbxout_switch_text_section as
1303 switch_text_section debug hook.
1304 (dbxout_function_end): Switch to current_function_section
1305 rather than function_section. If crtl->has_bb_partition,
1306 output just one N_FUN, depending on in_cold_section_p.
1307 (dbxout_source_line): Remember last lineno in lastlineno.
1308 (dbxout_switch_text_section): New function.
1309 (dbxout_function_decl): Adjust dbxout_block caller.
1310 (dbx_block_with_cold_children): New function.
1311 (dbxout_block): Return true if any LBRAC/RBRAC have been
1312 emitted. Use dbx_block_with_cold_children at depth == 0
1313 in second partition. Add PARENT_BLOCKNUM argument, pass
1314 it optionally adjusted to children. Output LBRAC/RBRAC
1315 around recursive call only if the block is in the current
1316 partition, if not and anything was output, emit empty
1318 * final.c (final_scan_insn): Compute cold_function_name
1319 before calling switch_text_section debug hook. Call
1320 that hook even if dwarf2out_do_frame if not emitting
1324 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
1325 TREE_READONLY decls.
1327 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1329 PR rtl-optimization/82488
1330 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
1332 2017-11-26 Julia Koval <julia.koval@intel.com>
1334 * config/i386/i386.c (processor_target_table): Add skylake_cost for
1336 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
1339 2017-11-26 Julia Koval <julia.koval@intel.com>
1341 * config/i386/driver-i386.c (host_detect_local_cpu):
1342 Detect skylake-avx512.
1344 2017-11-26 Julia Koval <julia.koval@intel.com>
1346 * config.gcc: Add -march=cannonlake.
1347 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
1348 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
1349 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
1350 (PTA_CANNONLAKE): New.
1351 (processor_target_table): Add cannonlake.
1352 (ix86_option_override_internal): Ditto.
1353 (fold_builtin_cpu): Ditto.
1354 (get_builtin_code_for_version): Handle cannonlake.
1355 (M_INTEL_COREI7_CANNONLAKE): New.
1356 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
1357 * doc/invoke.texi: Add -march=cannonlake.
1359 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
1361 * plugin.c (add_new_plugin): Use platform-specific library extensions.
1362 (try_init_one_plugin): Alternative implementation for MinGW.
1363 * Makefile.in (plugin_implib): New.
1364 (gengtype-lex.c): Fix broken AIX workaround.
1365 * configure: Regenerate.
1366 * doc/plugins.texi: Document support for MinGW.
1368 2017-11-25 Jakub Jelinek <jakub@redhat.com>
1370 PR rtl-optimization/81553
1371 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
1372 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
1373 is a shift where C1 has different mode than the whole shift, use C1's
1374 mode for MULT rather than the shift's mode.
1377 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
1378 builtins not enabled in the currently selected ISA.
1380 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
1382 PR tree-optimization/71026
1383 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
1384 (insert_reciprocals): Change to insert reciprocals before a division
1385 by a square and to insert the square of a reciprocal.
1386 (execute_cse_reciprocals_1): Change to consider division by a square.
1387 (register_division_in): Add importance parameter.
1389 2017-11-24 Richard Biener <rguenther@suse.de>
1391 PR tree-optimization/82402
1392 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
1393 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
1395 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
1397 * match.pd (0-ptr): New transformation.
1399 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
1402 * ipa-inline.c (inline_small_functions): Set current badnes correctly
1403 when skipping checking.
1405 2017-11-24 Richard Biener <rguenther@suse.de>
1407 PR tree-optimization/83128
1408 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
1409 (vn_reference_lookup_3): Likewise.
1411 2017-11-24 Jakub Jelinek <jakub@redhat.com>
1414 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
1415 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
1418 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
1419 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
1420 check for the latter with gimple_call_builtin_p. Do not handle
1421 BUILT_IN_STPNCPY_CHK which is not a pass through call.
1423 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
1425 * config/arm/arm_neon.h: Fix pragma GCC push_options before
1428 2017-11-23 Julia Koval <julia.koval@intel.com>
1430 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
1431 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
1432 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
1433 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
1434 _mm512_maskz_expandloadu_epi16): New intrinsics.
1435 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
1436 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
1437 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
1438 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
1439 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
1440 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
1441 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
1442 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
1443 _mm256_maskz_expandloadu_epi8): New intrinsics.
1444 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
1445 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
1446 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
1447 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
1448 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
1449 * config/i386/sse.md (VI248_VLBW): New iterator.
1450 (expand<mode>_mask, expand<mode>_maskz): New patterns.
1452 2017-11-23 Julia Koval <julia.koval@intel.com>
1454 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
1455 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
1456 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
1457 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
1458 _mm512_mask_compressstoreu_epi16): New.
1459 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
1460 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
1461 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
1462 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
1463 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
1464 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
1465 _mm256_mask_compressstoreu_epi8): New.
1466 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
1467 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
1468 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
1469 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
1470 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
1471 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
1472 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
1473 __builtin_ia32_compresshi128_mask,
1474 __builtin_ia32_compressstoreuqi512_mask,
1475 __builtin_ia32_compressstoreuhi512_mask,
1476 __builtin_ia32_compressstoreuqi256_mask,
1477 __builtin_ia32_compressstoreuqi128_mask,
1478 __builtin_ia32_compressstoreuhi256_mask,
1479 __builtin_ia32_compressstoreuhi128_mask): New builtins.
1480 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
1482 (ix86_expand_special_args_builtin): Handle new types.
1483 (s4fma_expand): Handle new builtin array.
1484 * config/i386/immintrin.h: Include new headers.
1485 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
1486 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
1488 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1490 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
1491 fixes. Declare temp and g variables at the top in order to avoid
1492 {} in most of the cases.
1494 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
1496 * match.pd (ptr-0): New transformation.
1498 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1500 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
1501 (struct addr_mode_cost_table): New.
1502 (struct tune_params): Add field addr_mode_costs.
1503 * config/arm/arm.c (generic_addr_mode_costs): New.
1504 (arm_slowmul_tune): Initialise addr_mode_costs field.
1505 (arm_fastmul_tune): Likewise.
1506 (arm_strongarm_tune): Likewise.
1507 (arm_xscale_tune): Likewise.
1508 (arm_9e_tune): Likewise.
1509 (arm_marvell_pj4_tune): Likewise.
1510 (arm_v6t2_tune): Likewise.
1511 (arm_cortex_tune): Likewise.
1512 (arm_cortex_a8_tune): Likewise.
1513 (arm_cortex_a7_tune): Likewise.
1514 (arm_cortex_a15_tune): Likewise.
1515 (arm_cortex_a35_tune): Likewise.
1516 (arm_cortex_a53_tune): Likewise.
1517 (arm_cortex_a57_tune): Likewise.
1518 (arm_exynosm1_tune): Likewise.
1519 (arm_xgene1_tune): Likewise.
1520 (arm_cortex_a5_tune): Likewise.
1521 (arm_cortex_a9_tune): Likewise.
1522 (arm_cortex_a12_tune): Likewise.
1523 (arm_cortex_a73_tune): Likewise.
1524 (arm_v7m_tune): Likewise.
1525 (arm_cortex_m7_tune): Likewise.
1526 (arm_v6m_tune): Likewise.
1527 (arm_fa726te_tune): Likewise.
1528 (arm_mem_costs): Use table lookup to calculate cost of addressing
1531 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1533 * config/arm/arm.c (arm_mem_costs): New function.
1534 (arm_rtx_costs_internal): Use arm_mem_costs.
1536 2017-11-23 Mark Wielaard <mark@klomp.org>
1538 * dwarf2out.c (init_sections_and_labels): Use generation to create
1539 unique ranges_section_label and ranges_base_label. Return generation.
1540 (output_rnglists): Add generation argument. Use generation to create
1541 unique ranges labels.
1542 (dwarf2out_finish): Get generation from init_sections_and_labels
1543 and pass generation to output_rnglists.
1545 2017-11-23 Mike Stump <mikestump@comcast.net>
1546 Eric Botcazou <ebotcazou@adacore.com>
1548 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
1549 * cfgloop.h (struct loop): Add unroll field.
1550 * function.h (struct function): Add has_unroll bitfield.
1551 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
1552 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
1553 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
1554 (pass_rtl_unroll_loops::gate): Likewise.
1555 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
1556 for which loop->unroll==1.
1557 (decide_unroll_constant_iterations): Use note for consistency and
1558 take loop->unroll into account. Return early if loop->unroll is set.
1559 Fix thinko in existing test.
1560 (decide_unroll_runtime_iterations): Use note for consistency and
1561 take loop->unroll into account.
1562 (decide_unroll_stupid): Likewise.
1563 * lto-streamer-in.c (input_cfg): Read loop->unroll.
1564 * lto-streamer-out.c (output_cfg): Write loop->unroll.
1565 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
1567 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
1568 (print_loop): Print loop->unroll if set.
1569 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
1570 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
1571 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
1573 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
1574 loop->unroll is set and smaller than the trip count. Otherwise bypass
1575 entirely the heuristics if loop->unroll is set. Remove dead note.
1576 Fix off-by-one bug in other note.
1577 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
1578 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
1580 (tree_unroll_loops_completely): Make static.
1581 (pass_complete_unroll::execute): Use correct type for variable.
1582 (pass_complete_unrolli::execute): Fix formatting.
1583 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
1585 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
1587 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
1588 enable when TARGET_PREFER_AVX128 is set.
1590 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1592 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
1594 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
1595 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
1597 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1599 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
1600 frequencies. Fix estimation of aggregate parameters.
1602 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1604 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
1607 2017-11-23 Tom de Vries <tom@codesourcery.com>
1609 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
1611 2017-11-23 Tom de Vries <tom@codesourcery.com>
1613 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
1616 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
1619 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
1620 sibcall_value_pcrel_fdpic): Use local variable instead of
1622 (calli_tbr_rel): Add missing operand 2.
1623 (call_valuei_tbr_rel): Add missing operand 3.
1625 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1628 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
1629 bitpos/bitsize covering the whole destination, use store_expr only if
1630 the complex mode is the same. Otherwise, use expand_normal and if
1631 it returns CONCAT, subreg each part separately instead of trying to
1632 subreg the whole result.
1634 2017-11-23 Richard Biener <rguenther@suse.de>
1636 PR tree-optimization/23094
1637 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
1638 come here from walking over backedges in the first iteration.
1639 (vn_reference_lookup_3): Skip clobbers that store the same value.
1641 2017-11-23 Richard Biener <rguenther@suse.de>
1643 PR tree-optimization/81403
1644 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
1645 a block we need a leader relative to.
1646 (phi_translate_1): For nary processing require a leader from
1647 get_representative_for given we run expression simplification
1648 using match-and-simplify. Remove previous fix.
1650 2017-11-22 Jeff Law <law@redhat.com>
1652 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
1653 Use new method allocate_value_range rather than accessing the
1654 vrp_value_range_pool data member directly.
1655 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
1656 to use extract_range_from_stmt method to avoid need for
1657 extract_range_from_assignment method.
1658 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
1659 method rather than setting values_propgated data member directly.
1660 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
1661 and values propagated data members and extract_range_from_assignment
1662 method. Reorder private data members to conform to standards.
1663 Add new methods set_lattice_propagation_complete and
1664 allocate_value_range.
1666 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
1668 PR rtl-optimization/83030
1669 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
1670 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
1671 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
1672 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
1673 CROSSING_JUMP_P flag is already set before setting it.
1674 * cfgrtl.c (fixup_partition_crossing): Likewise.
1675 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
1678 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1680 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
1681 Handle the case where both arguments are using gen_const_vec_series.
1683 2017-11-22 David Malcolm <dmalcolm@redhat.com>
1686 * pretty-print.c (pp_format): Move quoting implementation to
1687 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
1688 to pass address of "quote" local.
1689 (pp_begin_quote): New function.
1690 (pp_end_quote): New function.
1691 * pretty-print.h (printer_fn): Convert penultimate param from bool
1693 (pp_begin_quote): New decl.
1694 (pp_end_quote): New decl.
1695 * tree-diagnostic.c (default_tree_printer): Convert penultimate
1696 param from bool to bool *.
1697 * tree-diagnostic.h (default_tree_printer): Likewise.
1699 2017-11-22 Jeff Law <law@redhat.com>
1701 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
1702 Initialize vr_values.
1703 (evrp_range_analyzer::try_find_new_range): Call methods attached to
1704 vr_values via vr_values class instance rather than delegators.
1705 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
1706 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1707 (evrp_range_analyzer::push_value_range): Likewise.
1708 (evrp_range_analyzer::pop_value_range): Likewise.
1709 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
1710 most delegators. Those remaining are exposed as public interfaces.
1711 Make vr_values a pointer and private.
1712 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
1714 (evrp_range_analyzer::get_vr_value): New method.
1715 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
1716 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
1717 (class evrp_dom_walker): Attach evrp_folder class, initialize
1718 it in the ctor. Remove temporary delegators.
1719 (evrp_dom_walker::before_dom_children): Call methods in attached
1720 evrp_range_analyzer class via class instance pointer. Use
1721 free value_range_constant_singleton to remove need for
1722 op_with_constant_singleton_value delegator method. Do not
1723 create a vrp_prop class instance for every call! Narrow
1724 scope of a couple variables.
1725 (evrp_dom_walker::cleanup): Call methods in attached
1726 evrp_range_analyzer class via class instance pointer.
1727 * vr-values.h (class vr_values): Privatize many methods and
1730 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1732 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
1733 * cfgexpand.c (expand_debug_expr): Remove handling for them.
1734 * expr.c (expand_expr_real_2): Likewise.
1735 * fold-const.c (const_unop): Likewise.
1736 * optabs-tree.c (optab_for_tree_code): Likewise.
1737 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1738 * tree-inline.c (estimate_operator_cost): Likewise.
1739 * tree-pretty-print.c (dump_generic_node): Likewise.
1740 (op_code_prio): Likewise.
1741 (op_symbol_code): Likewise.
1742 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
1743 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
1744 * internal-fn.c (direct_internal_fn_optab): New function.
1745 (direct_internal_fn_array, direct_internal_fn_supported_p
1746 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
1747 * fold-const-call.c (fold_const_reduction): New function.
1748 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
1750 * tree-vect-loop.c: Include internal-fn.h.
1751 (reduction_code_for_scalar_code): Rename to...
1752 (reduction_fn_for_scalar_code): ...this and return an internal
1754 (vect_model_reduction_cost): Take an internal_fn rather than
1756 (vect_create_epilog_for_reduction): Likewise. Build calls rather
1758 (vectorizable_reduction): Use internal functions rather than tree
1759 codes for the reduction operation. Update calls to the functions
1761 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
1762 Use calls to internal functions rather than REDUC tree codes.
1763 * config/aarch64/aarch64-simd.md: Update comment accordingly.
1765 2017-11-22 Olivier Hainque <hainque@adacore.com>
1767 * config/vxworks.c (vxworks_override_options): Pick default
1768 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
1769 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
1770 DWARF_GNAT_ENCODINGS_DEFAULT.
1771 * config/vxworksae.h: Likewise.
1773 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
1775 PR tree-optimization/83104
1776 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
1779 2017-11-22 Marek Polacek <polacek@redhat.com>
1780 H.J. Lu <hongjiu.lu@intel.com>
1781 Jason Merrill <jason@redhat.com>
1786 * calls.c (initialize_argument_information): Call
1787 warn_parameter_passing_abi target hook.
1788 (store_one_arg): Use 0 for empty record size. Don't push 0 size
1789 argument onto stack.
1790 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
1791 * common.opt: Update -fabi-version description.
1792 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
1793 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
1795 (ix86_is_empty_record): New function.
1796 (ix86_warn_parameter_passing_abi): New function.
1797 (TARGET_EMPTY_RECORD_P): Redefine.
1798 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
1799 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
1800 * doc/tm.texi: Regenerated.
1801 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
1802 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
1803 * dwarf2out.c (get_ultimate_context): Move to tree.c.
1804 * explow.c (hard_function_value): Call arg_int_size_in_bytes
1805 instead of int_size_in_bytes.
1806 * expr.c (copy_blkmode_to_reg): Likewise.
1807 * function.c (aggregate_value_p): Return 0 for empty types.
1808 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
1809 (locate_and_pad_parm): Call arg size_in_bytes instead
1811 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
1812 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
1813 * target.def (empty_record_p, warn_parameter_passing_abi): New target
1815 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
1816 (std_gimplify_va_arg_expr): Skip empty records. Call
1817 arg_size_in_bytes instead size_in_bytes.
1818 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
1819 * tree-core.h (tree_type_common): Add empty_flag.
1820 (tree_decl_common): Update comments.
1821 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
1823 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1824 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
1826 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1827 * tree.c (default_is_empty_type): New function.
1828 (default_is_empty_record): New function.
1829 (arg_int_size_in_bytes): New function.
1830 (arg_size_in_bytes): New function.
1831 (get_ultimate_context): New function.
1832 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
1833 TRANSLATION_UNIT_WARN_EMPTY_P.
1834 (default_is_empty_record, arg_int_size_in_bytes,
1835 arg_size_in_bytes, get_ultimate_context): Declare.
1837 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1839 * config/arm/arm.c (cmse_clear_registers): New function.
1840 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
1841 code by call to cmse_clear_registers.
1842 (cmse_nonsecure_entry_clear_before_return): Likewise.
1844 2017-11-22 Tamar Christina <tamar.christina@arm.com>
1846 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
1847 (vdot_s32, vdotq_s32): New.
1848 (vdot_lane_u32, vdotq_lane_u32): New.
1849 (vdot_lane_s32, vdotq_lane_s32): New.
1852 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1855 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
1856 for unsigned values with fewer HWIs than the precision.
1857 (test_overflow): New function.
1858 (wide_int_cc_tests): Call it.
1860 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1862 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
1863 has the same mode class as Pmode.
1865 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1867 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
1868 padding_bits_to_clear_ptr.
1869 (cmse_nonsecure_entry_clear_before_return): Likewise.
1871 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1873 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
1874 auto_sbitap instead of integer bitfield to control register needing
1877 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1879 PR tree-optimization/83044
1880 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
1881 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
1882 that need the upper bound. Subtract offset from
1883 get_addr_base_and_unit_offset only if positive and subtract it
1884 before division by eltsize rather than after it.
1887 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
1888 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
1889 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
1893 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
1894 expand_binop with *mul_widen_optab, make sure at least one of the
1895 operands doesn't have VOIDmode.
1898 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
1900 PR rtl-optimization/82044
1901 PR tree-optimization/82042
1902 * dse.c (record_store): Check for overflow.
1903 (check_mem_read_rtx): Properly check for overflow if width == -1, call
1904 add_wild_read instead of clear_rhs_from_active_local_stores on
1905 overflow and log it into dump_file.
1907 2017-11-22 Richard Biener <rguenther@suse.de>
1909 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
1910 fake edges to exit when looking for a place to insert.
1911 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
1913 (insert_into_preds_of_block): Commit edge insertion immediately,
1914 assert that doesn't require new BBs.
1915 (fini_pre): Release expressions.
1916 (pass_pre::execute): Shuffle things around a bit, if the fn
1917 is too large do not compute AVAIL either as this is really the
1920 2017-11-22 Richard Biener <rguenther@suse.de>
1922 PR tree-optimization/83089
1923 * tree-if-conv.c (pass_if_conversion::execute): If anything
1924 changed reset SCEV and free the number of iteration estimates.
1926 2017-11-21 Martin Sebor <msebor@redhat.com>
1928 PR tree-optimization/82945
1929 * calls.h (warn_nonstring_bound): Remove unused function.
1931 2017-11-21 Martin Sebor <msebor@redhat.com>
1933 PR tree-optimization/82945
1934 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
1935 * calls.h (maybe_warn_nonstring_arg): Declare new function.
1936 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
1938 (initialize_argument_information): Call maybe_warn_nonstring_arg.
1939 * calls.h (get_attr_nonstring_decl): Declare new function.
1940 * doc/extend.texi (attribute nonstring): Update.
1941 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
1942 get_attr_nonstring_decl and handle it.
1943 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
1944 detection of nul-termination.
1945 (strlen_to_stridx): Change to a pointer.
1946 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
1947 (pass_strlen::execute): Same.
1949 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
1951 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
1952 for the new option -mprefer-vector-width=[none|128|256|512].
1953 * config/i386/i386.c (ix86_target_string): remove old style options
1954 -mprefer-avx256 and make -mprefer-avx128 as alias.
1955 (ix86_option_override_internal): Apply defaults for the
1956 -mprefer-vector-width=[128|256] option.
1957 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
1958 Implement macros to work with -mprefer-vector-width=.
1959 * config/i386/i386.opt: Implemented option
1960 -mprefer-vector-width=[none|128|256|512].
1961 * doc/invoke.texi: Documentation for
1962 -mprefer-vector-width=[none|128|256|512].
1964 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
1966 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
1968 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
1970 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
1972 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
1974 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
1976 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
1977 * expr.c (expand_expr_real_2): Likewise.
1978 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
1979 fold_binary_loc): Likewise.
1980 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
1981 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
1982 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
1983 MINUS_EXPR transformations.
1984 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
1985 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
1986 * tree-inline.c (estimate_operator_cost): Likewise.
1987 * tree-pretty-print.c (dump_generic_node, op_code_prio,
1988 op_symbol_code): Likewise.
1989 * tree-vect-stmts.c (vectorizable_operation): Likewise.
1990 * vr-values.c (extract_range_from_binary_expr): Likewise.
1991 * varasm.c (initializer_constant_valid_p_1): Likewise.
1992 * tree.def: New tree code POINTER_DIFF_EXPR.
1994 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
1996 * config/i386/i386.md (*bswap<mode>2_movbe): Add
1997 integer suffix to movbe mnemonic.
1998 (*bswaphi2_movbe): Ditto.
1999 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2001 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2004 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2005 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2006 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2008 2017-11-21 Martin Liska <mliska@suse.cz>
2010 * tree-inline.c (expand_call_inline): Remove not needed
2013 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2014 Jakub Jelinek <jakub@redhat.com>
2017 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2018 Remove static keyword from f variable.
2020 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2022 PR tree-optimization/83086
2023 * gimple-ssa-store-merging.c
2024 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2025 rather than n.base_addr.
2027 2017-11-21 Martin Liska <mliska@suse.cz>
2029 PR rtl-optimization/82044
2030 PR tree-optimization/82042
2031 * dse.c (check_mem_read_rtx): Check for overflow.
2033 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2035 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2038 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2040 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2041 and its corresponding call arg location note.
2043 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2044 Andrew Burgess <andrew.burgess@embecosm.com>
2046 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2048 (arc_return_slot_offset): Likewise.
2049 (arc_eh_return_address_location): New declaration.
2050 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2051 (MUST_SAVE_REGISTER): Add exception handler case.
2052 (MUST_SAVE_RETURN_ADDR): Likewise.
2053 (arc_frame_pointer_required): Likewise.
2054 (arc_frame_pointer_needed): New function.
2055 (arc_compute_frame_size): Changed.
2056 (arc_expand_prologue): Likewise.
2057 (arc_expand_epilogue): Likewise.
2058 (arc_initial_elimination_offset): Likewise.
2059 (arc_return_slot_offset): Delete.
2060 (arc_eh_return_address_location): New function.
2061 (arc_builtin_setjmp_frame_value): Likewise.
2062 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
2063 (EH_RETURN_STACKADJ_RTX): Define.
2064 (EH_RETURN_HANDLER_RTX): Likewise.
2065 * config/arc/arc.md (eh_return): Delete.
2067 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2069 * print-tree.h (debug_vec_tree): Remove prototype.
2070 * gdbinit.in (pvt): Remove macro.
2072 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2074 PR tree-optimization/83047
2075 * gimple-ssa-store-merging.c
2076 (imm_store_chain_info::output_merged_store): If the loads with the
2077 same vuse are in different basic blocks, for load_gsi pick a load
2078 location that is dominated by the other loads.
2081 * config/i386/i386.c (ix86_memmodel_check): Start
2082 -Winvalid-memory-model diagnostics with lowercase letter.
2085 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
2086 set in_cold_section_p to the partition containing loc_list->first.
2087 When seeing loc_list->last_before_switch node, update secname and
2088 perform range_across_switch second partition handling only after that.
2091 * run-rtl-passes.c: Include debug.h.
2092 (run_rtl_passes): Call debug_hooks->assembly_start.
2093 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
2097 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
2098 OPTAB_DIRECT in calls to expand_simple_binop.
2100 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2103 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
2105 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2108 * spellcheck.h (best_match::blithely_get_best_candidate): New
2111 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2113 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
2114 with lower case letter.
2116 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
2118 * config/i386/i386.md (bswaphi2): New expander.
2119 (*bswaphi2_movbe): New insn pattern.
2120 (bswaphi -> rorhi pepehole2): New peephole pattern.
2122 2017-11-20 Jeff Law <law@redhat.com>
2124 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
2125 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
2126 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
2127 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
2128 gimple-ssa-evrp-analyze.h.
2130 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
2131 set BB_VISITED here.
2132 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
2134 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2136 * tree-predcom.c: Add general comment on Store-Store chains.
2137 (split_data_refs_to_components): Postpone clearing eliminate_store_p
2139 (get_chain_last_ref_at): Rename into...
2140 (get_chain_last_write_at): ...this.
2141 (get_chain_last_write_before_load): New function.
2142 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
2143 CT_STORE_STORE when write reference is added.
2144 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
2145 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
2146 (initialize_root_vars_store_elim_1): Ditto.
2147 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
2148 definition is created.
2149 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
2150 chain by replacing it with dominant stored value.
2152 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2154 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
2156 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
2158 * vr-values.c (extract_range_from_binary_expr): Use a full range
2161 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
2163 * config/arm/arm.md (R4_REGNUM): Define constant.
2164 (nonsecure_call_internal): Remove r4 clobber.
2165 (nonsecure_call_value_internal): Likewise.
2166 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
2167 clobber and resequence match_operands.
2168 (nonsecure_call_value_reg_thumb1_v5): Likewise.
2169 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
2170 (nonsecure_call_value_reg_thumb2): Likewise.
2172 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2174 PR tree-optimization/78821
2175 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
2176 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
2177 to the address of the base rather than the base itself.
2178 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
2179 (find_bswap_or_nop_finalize): New function.
2180 (find_bswap_or_nop): Use it.
2181 (bswap_replace): Return a tree rather than bool, change first
2182 argument from gimple * to gimple_stmt_iterator, allow inserting
2183 into an empty sequence, allow ins_stmt to be NULL - then emit
2184 all stmts into gsi. Fix up MEM_REF address gimplification.
2185 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
2186 (struct store_immediate_info): Add N and INS_STMT non-static
2188 (store_immediate_info::store_immediate_info): Initialize them
2189 from newly added ctor args.
2190 (merged_store_group::apply_stores): Formatting fixes. Sort by
2192 (stmts_may_clobber_ref_p): For stores call also
2193 refs_anti_dependent_p.
2194 (gather_bswap_load_refs): New function.
2195 (imm_store_chain_info::try_coalesce_bswap): New method.
2196 (imm_store_chain_info::coalesce_immediate_stores): Use it.
2197 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
2198 (imm_store_chain_info::output_merged_store): Fail if number of
2199 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
2200 and NOP_EXPR rhs_code.
2201 (pass_store_merging::process_store): Compute n and ins_stmt, if
2202 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
2203 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
2205 (pass_store_merging::execute): Calculate dominators.
2207 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
2208 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
2209 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
2210 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2211 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
2212 class pass_optimize_bswap, bswap_replace,
2213 pass_optimize_bswap::execute): Moved to ...
2214 * gimple-ssa-store-merging.c: ... this file.
2215 Include optabs-tree.h.
2216 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
2217 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2218 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
2219 anonymous namespace, remove static keywords.
2220 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
2221 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
2223 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
2226 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
2228 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
2230 * vec.h (debug_helper): New function.
2231 (DEFINE_DEBUG_VEC): New macro.
2232 * hash-set.h (debug_helper): New function.
2233 (DEFINE_DEBUG_HASH_SET): New macro.
2234 * cfg.c (debug_slim (edge)): New function.
2235 Call DEFINE_DEBUG_VEC for edges.
2236 Call DEFINE_DEBUG_HASH_SET for edges.
2237 * cfghooks.c (debug_slim (basic_block)): New function.
2238 Call DEFINE_DEBUG_VEC for basic blocks.
2239 Call DEFINE_DEBUG_HASH_SET for basic blocks.
2240 * print-tree.c (debug_slim): New function to handle trees.
2241 Call DEFINE_DEBUG_VEC for trees.
2242 Call DEFINE_DEBUG_HASH_SET for trees.
2243 (debug (vec<tree, va_gc>) &): Remove.
2244 (debug (<vec<tree, va_gc>) *): Remove.
2245 * print-rtl.c (debug_slim): New function to handle const_rtx.
2246 Call DEFINE_DEBUG_VEC for rtx_def.
2247 Call DEFINE_DEBUG_VEC for rtx_insn.
2248 Call DEFINE_DEBUG_HASH_SET for rtx_def.
2249 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
2250 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
2251 (debug (vec<rtx_insn *> *ptr): Remove.
2252 (debug_insn_vector): Remove.
2253 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
2255 2017-11-20 Tom de Vries <tom@codesourcery.com>
2257 PR rtl-optimization/82020
2258 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
2259 IF_THEN_ELSE condition.
2261 2017-11-19 Jeff Law <law@redhat.com>
2263 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2264 of degenerates resulting from ignoring an edge.
2266 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2269 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
2271 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2274 * profile-count.c (profile_count::to_sreal_scale): Fix return value
2275 for uninitialied counts.
2277 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2280 * tree-inline.c (estimate_num_insns): Set to 1 at least.
2282 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2285 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
2287 2017-11-19 Tom de Vries <tom@codesourcery.com>
2289 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
2292 2017-11-19 Tom de Vries <tom@codesourcery.com>
2294 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
2297 2017-11-19 Tom de Vries <tom@codesourcery.com>
2299 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2302 2017-11-19 Tom de Vries <tom@codesourcery.com>
2304 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
2306 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
2308 2017-11-19 Tom de Vries <tom@codesourcery.com>
2310 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
2312 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
2313 semicolon after MCORE_EXPORT_NAME call.
2315 2017-11-19 Tom de Vries <tom@codesourcery.com>
2318 * vmsdbgout.c (vmsdbgout_early_finish): New function.
2319 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
2321 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
2323 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
2324 (cgraph_edge::clone): Cleanup updating of profile.
2325 * ipa-cp.c (update_profiling_info): Likewise.
2326 * ipa-inline-transform.c (inline_transform): Likewise.
2327 * ipa-inline.c (inline_small_functions): Add missing space to dump.
2328 * ipa-split.c (execute_split_functions): Do not split when function
2330 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
2331 * profile-count.c (profile_count::dump): Add global0.
2332 (profile_count::to_cgraph_frequency): Do not ICE when entry is
2334 (profile_count::to_sreal_scale): Likewise.
2335 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
2336 (profile_count::combine_with_ipa_count): New function.
2337 * profile-count.h (profile_guessed_global0adjusted): New.
2338 (profile_count::adjusted_zero): New.
2339 (profile_count::global0adjusted): New.
2340 (profile_count::combine_with_ipa_count): New.
2341 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
2342 correct profile of return block of split functions.
2343 (copy_cfg_body): Remove unused profile_count.
2344 (copy_body): Likewise.
2345 (expand_call_inline): Update.
2346 (tree_function_versioning): Update.
2348 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
2350 * hash-set.h (hash_set::empty): New.
2351 * tree-ssa-threadbackward.h: Delete.
2352 * tree-ssa-threadbackward.c (class thread_jumps): New.
2353 Move max_threaded_paths into class.
2354 (fsm_find_thread_path): Remove arguments that are now in class.
2355 (profitable_jump_thread_path): Rename to...
2356 (thread_jumps::profitable_jump_thread_path): ...this.
2357 (convert_and_register_jump_thread_path): Rename to...
2358 (thread_jumps::convert_and_register_current_path): ...this.
2359 (check_subpath_and_update_thread_path): Rename to...
2360 (thread_jumps::check_subpath_and_update_thread_path): ...this.
2361 (register_jump_thread_path_if_profitable): Rename to...
2362 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
2363 (handle_phi): Rename to...
2364 (thread_jumps::handle_phi): ...this.
2365 (handle_assignment): Rename to...
2366 (thread_jumps::handle_assignment): ...this.
2367 (fsm_find_control_statement_thread_paths): Rename to...
2368 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
2369 (find_jump_threads_backwards): Rename to...
2370 (thread_jumps::find_jump_threads_backwards): ...this.
2371 Initialize path local data.
2372 (pass_thread_jumps::execute): Call find_jump_threads_backwards
2373 from within thread_jumps class.
2374 (pass_early_thread_jumps::execute): Same.
2376 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2378 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
2380 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2382 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
2383 to canonical location.
2385 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2388 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
2390 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
2392 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
2394 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
2395 rcrt1.o%s/grcrt1.o%s for -static-pie.
2397 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2399 * i386.c (ix86_multiplication_cost, ix86_division_cost,
2400 ix86_shift_rotate_cost): Break out from ...
2401 (ix86_rtx_costs): ... here.
2402 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
2405 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2407 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
2408 when entry block was promoted unlikely.
2409 (estimate_bb_frequencies): Increase frequency scale.
2410 * profile-count.h (profile_count): Export precision info.
2412 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2414 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
2415 disturbin profile of entry block.
2417 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2419 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
2422 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2424 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
2425 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
2427 2017-11-17 Jeff Law <law@redhat.com>
2429 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
2430 from evrp_dom_walker class. Various methods moved into new class.
2431 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
2432 (evrp_range_analyzer::enter): New method.
2433 (evrp_range_analyzer::leave): New method.
2434 (evrp_dom_walker): Remove delegators no longer needed by this class.
2435 Replace vr_values data member with evrp_range_analyzer
2437 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
2438 method extracted from evrp_dom_walker::before_dom_children.
2439 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
2440 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
2442 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
2443 Add private copy constructor and move assignment operators.
2444 Privatize methods and class data where trivially possible.
2445 (evrp_dom_walker::cleanup): New function, extracted from
2446 execute_early_vrp. Simplify access to class data.
2448 * vr-values.h (get_output_for_vrp): Prototype.
2449 * vr-values.c (get_output_for_vrp): New function extracted from
2450 vrp_visit_assignment_or_call and extract_range_from_stmt.
2451 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
2453 2017-11-17 Luis Machado <luis.machado@linaro.org>
2455 * config/aarch64/aarch64.c
2456 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
2457 (qdf24xx_tunings) <autoprefetcher_model>: Set to
2458 tune_params::AUTOPREFETCHER_WEAK.
2460 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2463 * config/arm/arm.c (arm_valid_target_attribute_rec):
2464 Parse "arch=" and "+<ext>".
2465 (arm_valid_target_attribute_tree): Re-init global options.
2466 (arm_option_override): Make non-static.
2467 (arm_options_perform_arch_sanity_checks): Make errors fatal.
2468 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
2469 (__ARM_FEATURE_CRC32): Support undef.
2470 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
2471 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
2473 2017-11-17 David Malcolm <dmalcolm@redhat.com>
2475 * gdbinit.in (break-on-diagnostic): New command.
2477 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2479 * config/i386/i386.c (ix86_expand_epilogue): Change simple
2480 return to indirect jump for EH return if control-flow
2481 protection is enabled. Change explicit 'false' argument in
2482 pro_epilogue_adjust_stack with a value of
2484 * config/i386/i386.md (simple_return_indirect_internal):
2485 Remove SImode restriction to support 64-bit.
2487 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2489 * combine.c (added_notes_insn): New.
2490 (try_combine): Handle added_notes_insn like added_links_insn.
2491 Rewrite return value code.
2492 (distribute_notes): Set added_notes_insn to the earliest insn we added
2495 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2497 PR rtl-optimization/82621
2498 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
2499 dest of one of those SETs is unused.
2501 2017-11-17 Richard Biener <rguenther@suse.de>
2504 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
2505 * tree-pretty-print.c (dump_generic_node): Handle
2506 annot_expr_parallel_kind.
2507 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
2508 * gimplify.c (gimple_boolify): Likewise.
2510 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2512 * config.gcc (extra_headers): Add cet.h for x86 targets.
2513 * config/i386/cet.h: New file.
2514 * doc/install.texi: Add --enable-cet/--disable-cet.
2516 2017-11-17 Richard Biener <rguenther@suse.de>
2518 PR tree-optimization/83017
2519 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
2520 (gen_parallel_loop): Properly count iterations.
2521 (parallelize_loops): Handle loop->can_be_parallel independent
2522 of flag_loop_parallelize_all. Make static profitability test match
2524 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
2525 * invoke.texi (parloops-min-per-thread): Document.
2527 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
2529 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
2530 upstreaming review comments.
2532 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2534 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
2535 with fast unaligned access.
2536 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
2538 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
2540 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
2541 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
2543 2017-11-17 Richard Biener <rguenther@suse.de>
2545 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
2546 folding of references.
2548 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
2551 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
2552 of replacing call to strncmp with corresponding call to strcmp when
2555 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
2557 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
2558 option prefer-avx256 for skylake-avx512 configuration.
2559 * config/i386/i386.c (ix86_option_override_internal): Ditto.
2560 (get_builtin_code_for_version): Ditto.
2562 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2563 Monk Chiang <sh.chiang04@gmail.com>
2565 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
2566 (FIXED_REGISTERS): Reserve more register numbers.
2567 (CALL_USED_REGISTERS): Likewise.
2568 (REG_ALLOC_ORDER): Likewise.
2569 (REG_CLASS_CONTENTS): Likewise.
2570 (REGISTER_NAMES): Likewise.
2572 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2573 Kito Cheng <kito.cheng@gmail.com>
2575 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
2577 * config/nds32/iterators.md: Add vector mode iterators and attributes.
2579 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
2581 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
2582 parameter list for vec_splats.
2584 2017-11-16 Joseph Myers <joseph@codesourcery.com>
2586 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
2588 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
2590 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
2593 * doc/install.texi: Document incompatibility of Perl >=5.6.26
2594 with the required version of automake 1.11.6.
2596 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
2598 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
2600 (DU_C2_3_power9): Correct reservation combinations.
2601 (FP_DIV_power9, VEC_DIV_power9): New.
2602 (power9-alu): Split out rotate/shift...
2603 (power9-rot): ...to here, correct dispatch resource.
2604 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
2606 (power9-fp): Correct latency.
2607 (power9-sdiv): Add div/sqrt resource.
2608 (power9-ddiv): Correct latency, add div/sqrt resource.
2609 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
2610 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
2612 (power9-qpdiv, power9-qpmul): Adjust resource usage.
2614 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2616 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
2617 switch statement mapping KF built-ins to TF built-ins if we don't
2618 have the proper ISA 3.0 assembler support.
2620 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2622 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
2623 (lower_emutls_function_body): Do not compute it.
2625 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2627 * ipa-split.c (split_bb_info): Turn time to sreal.
2628 (split_point): Likewise.
2629 (dump_split_point): Likewise.
2630 (fine_split_points): Likewise.
2631 (execute_split_functions): Only zero split_bbs; turn time to sreals.
2633 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2635 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
2638 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2640 * predict.c (combine_predictions_for_bb): Preserve zero predicted
2642 (expensive_function_p): Remove useless assert.
2643 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
2645 2017-11-16 Martin Sebor <msebor@redhat.com>
2647 PR tree-optimization/82588
2648 PR tree-optimization/82583
2649 * tree-vrp.c (check_array_ref): Handle flexible array members,
2650 string literals, and inner indices.
2651 (search_for_addr_array): Add detail to diagnostics.
2653 2017-11-16 Nathan Sidwell <nathan@acm.org>
2657 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
2658 (SET_DECL_ASSEMBLER_NAME): Forward to
2659 overwrite_decl_assembler_name.
2660 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
2661 (overwrite_decl_assembler_name): Declare.
2662 * tree.c (overwrite_decl_assembler_name): New.
2663 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
2664 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
2665 (LANG_HOOKS_INITIALIZER): Add it.
2666 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
2667 * langhooks.c (lhd_set_decl_assembler_name): Use
2668 SET_DECL_ASSEMBLER_NAME.
2669 (lhd_overwrite_decl_assembler_name): Default implementation.
2671 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2672 Jackson Woodruff <jackson.woodruff@arm.com>
2674 PR tree-optimization/71026
2675 * match.pd: Canonicalize constant multiplies in division.
2677 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2679 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
2680 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
2681 Remove OPT_fomit_frame_pointer entry.
2682 * common/config/arc/arc-common.c: Likewise.
2683 * common/config/arm/arm-common.c: Likewise.
2684 * common/config/avr/avr-common.c: Likewise.
2685 * common/config/c6x/c6x-common.c: Likewise.
2686 * common/config/cr16/cr16-common.c: Likewise.
2687 * common/config/cris/cris-common.c: Likewise.
2688 * common/config/epiphany/epiphany-common.c: Likewise.
2689 * common/config/fr30/fr30-common.c: Likewise.
2690 * common/config/frv/frv-common.c: Likewise.
2691 * common/config/ia64/ia64-common.c: Likewise.
2692 * common/config/iq2000/iq2000-common.c: Likewise.
2693 * common/config/lm32/lm32-common.c: Likewise.
2694 * common/config/m32r/m32r-common.c: Likewise.
2695 * common/config/mcore/mcore-common.c: Likewise.
2696 * common/config/microblaze/microblaze-common.c: Likewise.
2697 * common/config/mips/mips-common.c: Likewise.
2698 * common/config/mmix/mmix-common.c: Likewise.
2699 * common/config/mn10300/mn10300-common.c: Likewise.
2700 * common/config/nios2/nios2-common.c: Likewise.
2701 * common/config/pa/pa-common.c: Likewise.
2702 * common/config/pdp11/pdp11-common.c: Likewise.
2703 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
2704 * common/config/riscv/riscv-common.c: Likewise.
2705 * common/config/rs6000/rs6000-common.c: Likewise.
2706 * common/config/rx/rx-common.c: Likewise.
2707 * common/config/s390/s390-common.c: Likewise.
2708 * common/config/sh/sh-common.c: Likewise.
2709 * common/config/sparc/sparc-common.c: Likewise.
2710 * common/config/tilegx/tilegx-common.c: Likewise.
2711 * common/config/tilepro/tilepro-common.c: Likewise.
2712 * common/config/v850/v850-common.c: Likewise.
2713 * common/config/visium/visium-common.c: Likewise.
2714 * common/config/xstormy16/xstormy16-common.c: Likewise.
2715 * common/config/xtensa/xtensa-common.c: Likewise.
2716 * invoke.texi (-fomit-frame-pointer): Update documentation.
2718 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2720 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
2722 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2724 * cfg.c (scale_bbs_frequencies_int,
2725 cale_bbs_frequencies_gcov_type): Remove.
2726 * cfg.h (scale_bbs_frequencies_int,
2727 cale_bbs_frequencies_gcov_type): Remove.
2729 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2731 * tree-ssa-loop-manip.c
2732 (scale_dominated_blocks_in_loop): Update to profile counts.
2733 (tree_transform_and_unroll_loop): Likewise.
2735 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2737 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
2738 scale_bbs_frequencies_int.
2740 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2742 * final.c (compute_alignments): Use counts rather than frequencies.
2744 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2746 * cfgloopanal.c: Include sreal.h
2747 (average_num_loop_insns): Use counts and sreal for accounting.
2749 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2751 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
2754 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2755 Kito Cheng <kito.cheng@gmail.com>
2757 * config/nds32/constraints.md: Provide more constraints.
2758 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
2759 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
2760 support constraints usage.
2762 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2764 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
2766 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2767 Kito Cheng <kito.cheng@gmail.com>
2769 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
2770 * config/nds32/nds32.opt: Refine the layout.
2771 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
2772 TARGET_EXT_STRING): Support new options.
2773 * config/nds32/nds32.h: Likewise.
2774 * config/nds32/nds32.md: Likewise.
2775 * config/nds32/nds32-predicates.c: Likewise.
2776 * config/nds32/constraints.md: Likewise.
2777 * common/config/nds32/nds32-common.c: Likewise.
2779 2017-11-16 Julia Koval <julia.koval@intel.com>
2782 * config/i386/gfniintrin.h: Add sse check.
2783 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
2785 2017-11-16 Julia Koval <julia.koval@intel.com>
2787 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
2788 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
2789 (ix86_handle_option): Handle -mavx512vbmi2.
2790 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
2791 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2792 * config/i386/i386-c.c (__AVX512VBMI2__): New.
2793 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
2794 (ix86_valid_target_attribute_inner_p): Ditto.
2795 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
2796 * config/i386/i386.opt (mavx512vbmi2): New option.
2797 * doc/invoke.texi: Add new option.
2799 2017-11-16 Julia Koval <julia.koval@intel.com>
2801 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
2802 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
2803 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
2804 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
2805 _mm512_gf2p8mul_epi8): New intrinsics.
2806 * config/i386/i386-builtin-types.def
2807 (V64QI_FTYPE_V64QI_V64QI): New type.
2808 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
2809 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
2810 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
2811 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
2812 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
2813 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2815 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
2817 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
2818 explicitly as a stream of bytes.
2820 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2822 * config/rs6000/altivec.h (vec_xst_be): New #define.
2823 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
2824 and externalize from *altivec_vperm_<mode>_internal.
2825 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
2827 (XL_BE_V8HI): Likewise.
2828 (XL_BE_V4SI): Likewise.
2829 (XL_BE_V4SI): Likewise.
2830 (XL_BE_V2DI): Likewise.
2831 (XL_BE_V4SF): Likewise.
2832 (XL_BE_V2DF): Likewise.
2833 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
2834 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
2835 all array entries with these keys: VSX_BUILTIN_VEC_XL,
2836 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
2837 VSX_BUILTIN_VEC_XST_BE.
2838 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
2839 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
2841 (altivec_init_builtins): Replace conditional calls to def_builtin
2842 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
2843 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
2844 with unconditional calls. Remove calls to def_builtin for
2845 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
2846 __builtin_vec_xst_be.
2847 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
2848 to define_expand, and add alternate RTL generation for P8.
2849 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
2850 vsx_ld_elemrev_v8hi.
2851 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
2852 add alternate RTL generation for P8.
2853 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
2854 vsx_ld_elemrev_v16qi.
2855 (vsx_st_elemrev_v8hi): Convert define_insn
2856 to define_expand, and add alternate RTL generation for P8.
2857 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
2858 vsx_st_elemrev_v8hi.
2859 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
2860 add alternate RTL generation for P8.
2861 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
2862 vsx_st_elemrev_v16qi.
2864 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
2867 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
2868 TARGET_AVX512ER check.
2869 (ix86_option_override_internal): Set MASK_VZEROUPPER if
2870 neither -mzeroupper nor -mno-zeroupper is used and
2871 TARGET_EMIT_VZEROUPPER is set.
2872 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
2873 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
2875 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2877 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2878 folding of vector compares.
2879 (fold_build_vec_cmp): New helper function.
2880 (fold_compare_helper): New helper function.
2881 (builtin_function_type): Add compare builtins to the list of functions
2882 having unsigned arguments. Cosmetic updates to comment indentation.
2883 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
2884 the not+eq combination.
2886 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2888 PR tree-optimization/82726
2889 PR tree-optimization/70754
2890 * tree-predcom.c (order_drefs_by_pos): New function.
2891 (combine_chains): Move code setting has_max_use_after to...
2892 (try_combine_chains): ...here. New parameter. Sort combined chains
2893 according to position information.
2894 (tree_predictive_commoning_loop): Update call to above function.
2895 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
2897 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2899 PR tree-optimization/82726
2901 2017-01-23 Bin Cheng <bin.cheng@arm.com>
2903 PR tree-optimization/70754
2904 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
2905 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
2906 combined stmt before it if not NULL.
2907 (combine_chains): Process refs reversely and compute dominance point
2911 2017-02-23 Bin Cheng <bin.cheng@arm.com>
2913 PR tree-optimization/79663
2914 * tree-predcom.c (combine_chains): Process refs in reverse order
2915 only for ZERO length chains, and add explaining comment.
2917 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2919 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
2920 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
2921 * doc/invoke.texi (ARM Options): Add armv8.3-a.
2923 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2925 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
2927 2017-11-15 Martin Liska <mliska@suse.cz>
2929 * tree-cfg.c (pass_warn_function_return::execute):
2930 Compare warn_return_type for greater than zero.
2932 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
2936 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
2937 to return true on Xeon and not on Xeon Phi.
2938 (ix86_check_avx256_register): Changed to ...
2939 (ix86_check_avx_upper_register): ... this. Add extra check for
2940 VALID_AVX512F_REG_OR_XI_MODE.
2941 (ix86_avx_u128_mode_needed): Changed
2942 ix86_check_avx256_register to ix86_check_avx_upper_register.
2943 (ix86_check_avx256_stores): Changed to ...
2944 (ix86_check_avx_upper_stores): ... this. Changed
2945 ix86_check_avx256_register to ix86_check_avx_upper_register.
2946 (ix86_avx_u128_mode_after): Changed
2947 avx_reg256_found to avx_upper_reg_found. Changed
2948 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
2949 (ix86_avx_u128_mode_entry): Changed
2950 ix86_check_avx256_register to ix86_check_avx_upper_register.
2951 (ix86_avx_u128_mode_exit): Ditto.
2952 * config/i386/i386.h: (host_detect_local_cpu): New define.
2954 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2956 * config/arm/xgene1.md (xgene1): Split into automatons
2957 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
2958 (xgene1_f_load): Adjust reservations and/or types.
2959 (xgene1_f_store): Likewise.
2960 (xgene1_load_pair): Likewise.
2961 (xgene1_store_pair): Likewise.
2962 (xgene1_fp_load1): Likewise.
2963 (xgene1_load1): Likewise.
2964 (xgene1_store1): Likewise.
2965 (xgene1_move): Likewise.
2966 (xgene1_alu): Likewise.
2967 (xgene1_simd): Likewise.
2968 (xgene1_bfm): Likewise.
2969 (xgene1_neon_load1): Likewise.
2970 (xgene1_neon_store1): Likewise.
2971 (xgene1_neon_logic): Likewise.
2972 (xgene1_neon_st1): Likewise.
2973 (xgene1_neon_ld1r): Likewise.
2974 (xgene1_alu_cond): Added.
2975 (xgene1_shift_reg): Likwise.
2976 (xgene1_bfx): Likewise.
2977 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
2979 2017-11-15 Jakub Jelinek <jakub@redhat.com>
2982 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
2984 (can_widen_mult_without_libcall): New function.
2985 (expand_mul_overflow): If only checking unsigned mul overflow,
2986 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
2987 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
2988 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
2989 (expand_DIVMOD): Formatting fix.
2990 * expmed.h (expand_mult): Add NO_LIBCALL argument.
2991 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
2992 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
2994 PR tree-optimization/82977
2995 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
2996 constructed temporary to strlen_to_stridx.put.
2998 2017-11-15 Martin Liska <mliska@suse.cz>
3000 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3001 * configure: Regenerate.
3003 2017-11-15 Martin Liska <mliska@suse.cz>
3006 * config/sh/sh-mem.cc: Use proper probability for
3009 2017-11-14 Jeff Law <law@redhat.com>
3011 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3012 the red zone for stack_clash_protection_final_dynamic_probe targets
3013 when the total dynamic stack size is zero bytes.
3015 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3016 blocks is post order.
3018 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3020 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3021 * final.c (rest_of_clean_state): Set it for the
3022 -fcompare-debug dump.
3023 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3024 class when TDF_COMPARE_DEBUG is set.
3026 * dwarf2out.c (gen_producer_string): Discard
3029 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3032 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3033 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3034 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3035 (__TGMATH_REAL_2_3): Remove macros.
3036 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3037 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3039 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3041 (remquo): Define using __TGMATH_REAL_3.
3043 2017-11-14 Jeff Law <law@redhat.com>
3045 * vr-values.c: New file with contents extracted from tree-vrp.c.
3046 * Makefile.in (OBJS): Add vr-values.o
3047 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3048 (set_value_range, set_and_canonicalize_value_range): Likewise.
3049 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3050 (value_range_constant_singleton, symbolic_range_p): Likewise.
3051 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3052 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3053 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3054 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3055 (find_case_label_range, find_case_label_index): Likewise.
3056 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3057 (range_int_cst_singleton_p, value_inside_range): Likewise.
3058 (get_single_symbol): Likewise.
3059 (switch_update): Move structure definition here.
3060 (to_remove_edges, to_update_switch_stmts): Provide externs.
3061 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
3062 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
3063 (vrp_val_is_min, set_value_range): Likewise.
3064 (set_and_canonicalize_value_range, copy_value_range): Likewise.
3065 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
3066 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
3067 (range_is_nonnull, range_int_cst_p): Likewwise.
3068 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
3069 (get_single_symbol, operand_less_p): Likewise
3070 (compare_values_warnv, compare_values): Likewise.
3071 (value_inside_range, value_range_constant_singleton): Likewise.
3072 (zero_nonzero_bitgs_from_vr): Likewise.
3073 (extract_range_from_binary_expr_1): Likewise.
3074 (overflow_comparison_p): Likewise.
3075 (to_remove_edges, to_update_switch_stmts): Likewise.
3076 (find_case_label-index, find_case_label_range): Likewise.
3077 (switch_update, set_value_range_to_nonnegative): Remove.
3078 (set_value_range_to_truthvalue): Likewise.
3079 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
3080 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
3081 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
3082 (find_case_label_ranges, test_for_singularity): Likewise.
3083 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
3084 (x_vr_values): Move to its remaining use site.
3086 2017-11-10 Jeff Law <law@redhat.com>
3088 * vr-values.h (VR_INITIALIZER): Move #define here.
3089 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
3090 * Makefile.in (OBJS): Add tree-evrp.o
3091 * tree-vrp.h (assert_info): Move structure definition here.
3092 (set_value_range_to_varying): Prototype.
3093 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3094 (infer_value_range, register_edge_assert_for): Likewise.
3095 (stmt_interesting_for_vrp): Likewise.
3096 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
3097 (set_value_range_to_varying): No longer static.
3098 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3099 (infer_value_range, register_edge_assert_for): Likewise.
3101 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3103 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
3104 generate the XXBRD instruction.
3106 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
3107 (rs6000_builtin_type_compatible): Treat _Float128 and long double
3108 as being compatible if -mabi=ieeelongdouble.
3109 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
3110 to setup float128 built-ins with hardware support.
3111 (BU_FLOAT128_HW_2): Likewise.
3112 (BU_FLOAT128_HW_3): Likewise.
3113 (BU_FLOAT128_HW_VSX_1): Likewise.
3114 (BU_FLOAT128_HW_VSX_2): Likewise.
3115 (scalar_extract_expq): Change float128 built-in functions to
3116 accommodate having both KFmode and TFmode functions. Use the
3117 KFmode variant as the default.
3118 (scalar_extract_sigq): Likewise.
3119 (scalar_test_neg_qp): Likewise.
3120 (scalar_insert_exp_q): Likewise.
3121 (scalar_insert_exp_qp): Likewise.
3122 (scalar_test_data_class_qp): Likewise.
3123 (sqrtf128_round_to_odd): Delete processing the round to odd
3124 built-in functions as special built-in functions, and define them
3125 as float128 built-ins. Use the KFmode variant as the default.
3126 (truncf128_round_to_odd): Likewise.
3127 (addf128_round_to_odd): Likewise.
3128 (subf128_round_to_odd): Likewise.
3129 (mulf128_round_to_odd): Likewise.
3130 (divf128_round_to_odd): Likewise.
3131 (fmaf128_round_to_odd): Likewise.
3132 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
3133 support for KFmode and TFmode xststdcqp calls.
3134 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
3135 point, switch the built-in handlers for the get/set float128
3136 exponent, get float128 mantissa, float128 test built-ins, and the
3137 float128 round to odd built-in functions. Eliminate creating the
3138 float128 round to odd built-in functions as special built-ins.
3139 (rs6000_init_builtins): Eliminate special creation of the float128
3140 round to odd built-in functions.
3141 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
3142 function insns to support both TFmode and KFmode variants.
3143 (xsxsigqp_<mode>): Likewise.
3144 (xsiexpqpf_<mode>): Likewise.
3145 (xsiexpqp_<mode>): Likewise.
3146 (xststdcqp_<mode>): Likewise.
3147 (xststdcnegqp_<mode>): Likewise.
3148 (xststdcqp_<mode>): Likewise.
3150 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3152 * tree-ssa-threadupdate.c (compute_path_counts): Remove
3153 unused path_in_freq_ptr parameter.
3154 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
3156 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3158 * ipa-inline.c (edge_badness): Dump sreal frequency.
3159 (compute_inlined_call_time): Match natural implementaiton ...
3160 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
3161 forgotten division by CGRAPH_FREQ_BASE.
3163 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3165 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
3166 Solaris 11. Update comment.
3167 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
3169 * config/sol2.h (STARTFILE_SPEC): Likewise.
3170 * configure: Regenerate.
3172 2017-11-14 Carl Love <cel@us.ibm.com>
3174 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
3175 le_ and be_ prefixes to swap* variables. Remove
3176 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
3179 2017-11-14 Jason Merrill <jason@redhat.com>
3181 Support GTY((cache)) on hash_map.
3182 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
3183 (ggc_cache_remove): Override it instead of ggc_mx.
3184 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
3185 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
3186 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
3187 (simple_cache_map_traits): Override maybe_mx.
3188 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
3189 (hash_map): Friend gt_cleare_cache.
3190 (gt_cleare_cache): New.
3191 * tree.h (tree_cache_traits): New hash_map traits class.
3192 (tree_cache_map): New typedef.
3194 2017-11-14 Richard Biener <rguenther@suse.de>
3196 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
3197 paramter and handling.
3198 (cleanup_control_flow_bb): Likewise.
3199 (cleanup_control_flow_pre): New helper performing a DFS walk
3200 to call cleanup_control_flow_bb in PRE order.
3201 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
3202 via cleanup_control_flow_pre.
3204 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
3206 * config/aarch64/aarch64-simd.md
3207 (aarch64_simd_bsl<mode>_internal): Remove DImode.
3208 (*aarch64_simd_bsl<mode>_alt): Likewise.
3209 (aarch64_simd_bsldi_internal): New.
3210 (aarch64_simd_bsldi_alt): Likewise.
3212 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3214 * tracer.c (better_p): Do not compare frequencies.
3215 * reg-stack.c (better_edge): Likewise.
3216 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
3219 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3221 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
3222 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
3223 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
3224 * lto-streamer-in.c (input_function): Use update_max_bb_count.
3225 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
3226 * predict.c (maybe_hot_frequency_p): Inline to ...
3227 (maybe_hot_count_p): ... here; rewrite to counts.
3228 (counts_to_freqs): Rename to ...
3229 (update_max_bb_count): ... this one.
3230 (expensive_function_p): Use counts.
3231 (estimate_bb_frequencies): Update.
3232 (rebuild_frequencies): Update.
3233 * predict.h (counts_to_freqs): Rename to ...
3234 (update_max_bb_count): ... this one.
3235 * profile.c (compute_branch_probabilities): Add debug info
3236 * tree-inline.c (expand_call_inline): Update debug info.
3237 (optimize_inline_calls): Use update_max_bb_count..
3238 (tree_function_versioning): Use update_max_bb_count..
3239 * value-prof.c (gimple_value_profile_transformations):
3240 Do not use update_max_bb_count.
3242 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3244 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
3245 always use frequencies.
3247 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3249 * bb-reorder.c: Remove frequencies from comments.
3250 (better_edge_p): Use profile counts.
3251 (find_traces): Dump profile counts.
3252 (rotate_loop): Use profile counts.
3253 (find_traces_1_round): Likewise.
3254 (connect_better_edge_p): Use counts instead of probabilities for
3256 (copy_bb_p): Drop early check for non-0 frequency.
3257 (sanitize_hot_paths): Update comments.
3259 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3261 * ipa-split.c (struct split_point): Add count.
3262 (consider_split): Do not compute incoming frequency; compute incoming
3263 count and store it to split_point.
3264 (split_function): Set count of the call to split part correctly.
3266 2017-11-13 Carl Love <cel@us.ibm.com>
3268 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
3270 2017-11-13 Tom Tromey <tom@tromey.com>
3272 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
3274 2017-11-13 Carl Love <cel@us.ibm.com>
3276 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3277 Add support for builtins:
3278 unsigned int vec_first_{,miss}_match_{,or_eos}index,
3279 vector {un,}signed {char,int,short},
3280 vector {un,}signed {char,int,short}) arguments.
3281 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
3282 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
3283 Add BU_P9V_AV_2 expansions for the builtins.
3284 * config/rs6000/altivec.h (vec_first_match_index,
3285 vec_first_mismatch_index, vec_first_match_or_eos_index,
3286 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
3287 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
3288 new extern declaration.
3289 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
3290 * config/rs6000/vsx.md (first_match_index_<mode>,
3291 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
3292 first_mismatch_or_eos_index_<mode>): Add define expand.
3293 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
3294 * doc/extend.texi: Update the built-in documenation file for the new
3297 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
3299 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
3300 into the min/max operations for _Float<N> and _Float<N>X types.
3302 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3305 * dwarf2out.c (do_eh_frame): New static variable.
3306 (dwarf2out_begin_prologue): Set it.
3307 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
3309 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3311 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
3313 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
3315 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3318 * debug.h (dwarf2out_do_eh_frame): Declare.
3319 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
3320 (dwarf2out_do_frame): Use it.
3321 (dwarf2out_do_cfi_asm): Likewise.
3322 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
3323 (dwarf2out_assembly_start): Likewise.
3324 (dwarf2out_begin_prologue): Fix comment.
3325 * toplev.c (compile_file): Always call dwarf2out_frame_finish
3326 if the target needs either debug or unwind DWARF2 info.
3327 * lto-opts.c (lto_write_options): Do not save -fexceptions,
3328 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
3329 -ftrapping-math, -ftrapv and -fwrapv.
3331 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3333 * cgraph.c (cgraph_edge::sreal_frequency): New function.
3334 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
3335 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
3336 (estimate_edge_size_and_time): Likewise.
3337 (ipa_merge_fn_summary_after_inlining): Likewise.
3338 * ipa-inline.c (cgraph_freq_base_rec): Remove.
3339 (compute_uninlined_call_time): Use sreal_frequency.
3340 (compute_inlined_call_time): Likewise.
3341 (ipa_inline): Do not initialize cgraph_freq_base_rec.
3342 * profile-count.c: Include sreal.h.
3343 (profile_count::to_sreal_scale): New.
3344 * profile-count.h: Forward declare sreal.
3345 (profile_count::to_sreal_scale): Declare.
3347 2017-11-13 Nathan Sidwell <nathan@acm.org>
3349 * diagnostic.c (maybe_line_and_column): New.
3350 (diagnostic_get_location_text): Use it.
3351 (diagnostic_report_current_module): Likewise.
3352 (test_diagnostic_get_location_text): Add tests.
3354 2017-11-13 Luis Machado <luis.machado@linaro.org>
3356 * doc/md.texi (Specifying processor pipeline description): Fix
3357 incorrect latency for the div instruction example.
3359 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3361 PR tree-optimization/78821
3362 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
3363 that bit_not_p is the same.
3364 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
3365 (split_group): Count precisely bit_not_p bits in each statement.
3366 (invert_op): New function.
3367 (imm_store_chain_info::output_merged_store): Use invert_op to
3368 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
3369 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
3371 2017-11-13 Martin Liska <mliska@suse.cz>
3373 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
3374 (struct source_info): Likewise.
3375 (add_branch_counts): Likewise.
3376 (add_line_counts): Likewise.
3377 (function_summary): Likewise.
3378 (output_intermediate_line): Likewise.
3379 (generate_results): Likewise.
3381 2017-11-13 Martin Liska <mliska@suse.cz>
3383 * gcov.c (struct block_info): Remove typedef for block_t.
3384 (struct line_info): Likewise.
3385 (line_info::has_block): Likewise.
3386 (EXIT_BLOCK): Likewise.
3387 (unblock): Likewise.
3388 (circuit): Likewise.
3389 (get_cycles_count): Likewise.
3390 (process_file): Likewise.
3391 (read_graph_file): Likewise.
3392 (solve_flow_graph): Likewise.
3393 (find_exception_blocks): Likewise.
3394 (add_line_counts): Likewise.
3395 (accumulate_line_info): Likewise.
3396 (output_line_details): Likewise.
3398 2017-11-13 Martin Liska <mliska@suse.cz>
3400 * gcov.c (struct arc_info): Remove typedef for arc_t.
3401 (struct line_info): Likewise.
3402 (add_branch_counts): Likewise.
3403 (output_branch_count): Likewise.
3404 (function_info::~function_info): Likewise.
3405 (circuit): Likewise.
3406 (output_intermediate_line): Likewise.
3407 (read_graph_file): Likewise.
3408 (solve_flow_graph): Likewise.
3409 (find_exception_blocks): Likewise.
3410 (add_line_counts): Likewise.
3411 (accumulate_line_info): Likewise.
3412 (output_line_details): Likewise.
3413 (output_function_details): Likewise.
3415 2017-11-13 Martin Liska <mliska@suse.cz>
3417 * gcov.c (struct function_info): Remove typedef for function_t.
3418 (struct source_info): Likewise.
3419 (source_info::get_functions_at_location): Likewise.
3420 (solve_flow_graph): Likewise.
3421 (find_exception_blocks): Likewise.
3422 (add_line_counts): Likewise.
3423 (output_intermediate_file): Likewise.
3424 (process_file): Likewise.
3425 (generate_results): Likewise.
3426 (release_structures): Likewise.
3427 (read_graph_file): Likewise.
3428 (read_count_file): Likewise.
3429 (accumulate_line_counts): Likewise.
3430 (output_lines): Likewise.
3432 2017-11-13 Martin Liska <mliska@suse.cz>
3434 * gcov.c (function_info::function_info): Remove num_counts
3435 and add vector<gcov_type>.
3436 (function_info::~function_info): Use the vector.
3437 (process_file): Likewise.
3438 (read_graph_file): Likewise.
3439 (read_count_file): Likewise.
3440 (solve_flow_graph): Likewise.
3442 2017-11-13 Martin Liska <mliska@suse.cz>
3444 * gcov.c (function_info::is_artificial): New function.
3445 (process_file): Erase all artificial early.
3446 (generate_results): Skip as all artificial are already
3449 2017-11-13 Martin Liska <mliska@suse.cz>
3451 * gcov.c (read_graph_file): Store to global vector of functions.
3452 (read_count_file): Iterate the vector.
3453 (process_file): Likewise.
3454 (generate_results): Likewise.
3455 (release_structures): Likewise.
3457 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3459 PR tree-optimization/82954
3460 * gimple-ssa-store-merging.c
3461 (imm_store_chain_info::coalesce_immediate_stores): If
3462 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
3464 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
3466 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
3467 Upddate call to ENDIAN_LANE_N.
3468 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
3469 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
3470 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
3471 and use aarch64_endian_lane_rtx.
3472 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
3474 2017-11-12 Tom de Vries <tom@codesourcery.com>
3476 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
3478 2017-11-12 Tom de Vries <tom@codesourcery.com>
3480 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
3482 2017-11-12 Tom de Vries <tom@codesourcery.com>
3484 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
3486 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3487 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3488 * defaults.h (ASM_OUTPUT_LABELREF): Same.
3490 2017-11-11 Martin Sebor <msebor@redhat.com>
3493 * doc/extend.texi (attribute nonstring): Remove spurious argument.
3496 * prefic.c (translate_name): Replace strncpy with memcpy to
3497 avoid -Wstringop-truncation.
3499 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3501 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
3503 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3505 * predict.c (maybe_hot_frequency_p): Do not use cfun.
3507 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3509 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
3512 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
3514 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
3515 enable generating XXBRH if the value is in a vector register.
3516 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
3517 value is in a vector register.
3518 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
3519 register to register bswap64's instead of doing the GPR sequence
3520 used on previous machines.
3521 (bswapdi2_xxbrd): New insn.
3522 (bswapdi2_reg): Disallow on ISA 3.0.
3523 (register to register bswap64 splitter): Do not split the insn on
3524 ISA 3.0 systems that use XXBRD.
3526 2017-11-10 Martin Sebor <msebor@redhat.com>
3529 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
3531 (find_subframework_file): Same.
3533 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3535 * auto-profile.c (afdo_indirect_call): Drop frequency.
3536 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
3537 (cgraph_node::create_edge): Drop frequency argument.
3538 (cgraph_node::create_indirect_edge): Drop frequency argument.
3539 (cgraph_edge::make_speculative): Drop frequency arguments.
3540 (cgraph_edge::resolve_speculation): Do not update frequencies
3541 (cgraph_edge::dump_edge_flags): Do not dump frequency.
3542 (cgraph_node::dump): Check consistency in IPA mode.
3543 (cgraph_edge::maybe_hot_p): Use IPA counter.
3544 (cgraph_edge::verify_count_and_frequency): Rename to ...
3545 (cgraph_edge::verify_count): ... this one; drop frequency checking.
3546 (cgraph_node::verify_node): Update.
3547 * cgraph.h (struct cgraph_edge): Drop frequency.
3548 (cgraph_edge::frequency): New function.
3549 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
3551 (cgraph_edge::rebuild_edges): Likewise.
3552 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
3553 (duplicate_thunk_for_node): Do not pass frequency.
3554 (cgraph_node::create_clone): Scale only counts.
3555 (cgraph_node::create_virtual_clone): Do not pass frequency.
3556 (cgraph_node::create_edge_including_clones): Do not pass frequency.
3557 (cgraph_node::create_version_clone): Do not pass frequency.
3558 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
3559 (cgraph_node::expand_thunk): Do not pass frequency.
3560 (cgraph_node::create_wrapper): Do not pass frequency.
3561 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
3563 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
3564 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
3565 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
3566 (gather_caller_stats): Use frequency function.
3567 (ipcp_cloning_candidate_p): Use frequency function.
3568 (ipcp_propagate_stage): Use frequency function.
3569 (get_info_about_necessary_edges): Use frequency function.
3570 (update_profiling_info): Update only IPA profile.
3571 (update_specialized_profile): Use frequency functoin.
3572 (perhaps_add_new_callers): Update only IPA profile.
3573 * ipa-devirt.c (ipa_devirt): Use IPA profile.
3574 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
3575 (dump_ipa_call_summary): Use frequency function.
3576 (estimate_edge_size_and_time): Use frequency function.
3577 (ipa_merge_fn_summary_after_inlining): Use frequency function.
3578 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
3579 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
3580 (update_noncloned_counts): ... ths one; scale counts only.
3581 (clone_inlined_nodes): Do not scale frequency.
3582 (inline_call): Do not pass frequency.
3583 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
3584 (compute_inlined_call_time): Use IPA profile.
3585 (want_inline_small_function_p): Use IPA profile.
3586 (want_inline_self_recursive_call_p): Use IPA profile.
3587 (edge_badness): Use IPA profile.
3588 (lookup_recursive_calls): Use IPA profile.
3589 (recursive_inlining): Do not pass frequency.
3590 (resolve_noninline_speculation): Do not update frequency.
3591 (inline_small_functions): Collect max of IPA profile.
3592 (dump_overall_stats): Dump IPA porfile.
3593 (dump_inline_stats): Dump IPA porfile.
3594 (ipa_inline): Collect IPA stats.
3595 * ipa-inline.h (clone_inlined_nodes): Update prototype.
3596 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
3597 (ipa_propagate_frequency): Use frequency function.
3598 (ipa_profile): Cleanup.
3599 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
3600 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
3601 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
3602 (input_node): Do not stream frequency.
3603 (input_edge): Do not stream frequency.
3604 (merge_profile_summaries): Scale only IPA profiles.
3605 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
3606 * predict.c (drop_profile): Do not recompute frequency.
3607 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
3608 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
3609 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
3610 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
3611 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
3612 * tree-inline.c (copy_bb): Do not scale frequency.
3613 (expand_call_inline): Do not scale frequency.
3614 (tree_function_versioning): Do not scale frequency.
3615 * ubsan.c (ubsan_create_edge): Do not pass frequency.
3617 2017-11-10 Julia Koval <julia.koval@intel.com>
3619 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
3620 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
3621 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
3622 (_mm256_maskz_gf2p8affine_epi64_epi8)
3623 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
3624 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
3625 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
3626 (__builtin_ia32_vgf2p8affineqb_v32qi)
3627 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
3628 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
3630 2017-11-10 Tamar Christina <tamar.christina@arm.com>
3634 (arm_option_override): Refactor.
3635 (arm_option_reconfigure_globals): New.
3636 (arm_options_perform_arch_sanity_checks): New.
3637 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
3639 (arm_options_perform_arch_sanity_checks): Likewise
3641 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
3643 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
3644 (power9-qpmul): New.
3645 * rs6000/rs6000.md ("type" attr): Add qmul.
3646 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
3647 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
3648 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
3650 2017-11-10 Martin Sebor <msebor@redhat.com>
3653 * builtins.c (compute_objsize): Handle arrays that
3654 compute_builtin_object_size likes to fail for. Make extern.
3655 * builtins.h (compute_objsize): Declare.
3656 (check_strncpy_sizes): New function.
3657 (expand_builtin_strncpy): Call check_strncpy_sizes.
3658 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
3659 -Wstringop-truncation.
3660 (gimple_fold_builtin_strncat): Same.
3661 * gimple.c (gimple_build_call_from_tree): Set call location.
3662 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
3663 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
3664 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
3665 (handle_builtin_strlen): Use strlen_to_stridx.
3666 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
3668 Use strlen_to_stridx.
3669 (pass_strlen::execute): Release strlen_to_stridx.
3670 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
3671 (-Wstringop-truncation): Document new option.
3673 2017-11-10 Martin Liska <mliska@suse.cz>
3675 PR gcov-profile/82702
3676 * gcov.c (main): Handle intermediate files in a different
3678 (get_gcov_intermediate_filename): New function.
3679 (output_gcov_file): Remove support of intermediate files.
3680 (generate_results): Allocate intermediate file.
3681 (release_structures): Clean-up properly fn_end.
3682 (output_intermediate_file): Start iterating with line 1.
3684 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3686 PR tree-optimization/82929
3687 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3688 ops_swapped_p non-static data member.
3689 (store_immediate_info::store_immediate_info): Clear it.
3690 (imm_store_chain_info::coalesce_immediate_stores): If swapping
3691 ops set ops_swapped_p.
3692 (count_multiple_uses): Handle ops_swapped_p.
3694 2017-11-10 Martin Liska <mliska@suse.cz>
3696 * coverage.c (coverage_init): Stream information about
3697 support of has_unexecuted_blocks.
3698 * doc/gcov.texi: Document that.
3699 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
3700 * gcov.c (read_graph_file): Likewise.
3701 (output_line_beginning): Fix a small issue with
3704 2017-11-10 Bin Cheng <bin.cheng@arm.com>
3706 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
3707 reference of trivial component.
3709 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3712 * gimple-ssa-store-merging.c
3713 (pass_store_merging::terminate_all_aliasing_chains): For
3714 gimple_store_p stmts also call refs_output_dependent_p.
3716 PR rtl-optimization/82913
3717 * compare-elim.c (try_merge_compare): Punt if def_insn is not
3720 2017-11-09 Jeff Law <law@redhat.com>
3722 * vr-values.h: New file with vr_values class.
3723 * tree-vrp.c: Include vr-values.h
3724 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
3725 data objects into the vr_values class.
3726 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
3727 (get_value_range): Make it a member function within vr_values class.
3728 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
3729 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
3730 (op_with_constant_singleton_value_range): Likewise.
3731 (extract_range_for_var_from_comparison_expr): Likewise.
3732 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
3733 (extract_range_from_binary_expr): Likewise.
3734 (extract_range_from_unary_expr): Likewise.
3735 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
3736 (check_for_binary_op_overflow, extract_range_basic): Likewise.
3737 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
3738 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
3739 (compare_name_with_value, compare_names): Likewise.
3740 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
3741 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
3742 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
3743 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
3744 (extract_range_from_phi_node): Likewise.
3745 (simplify_truth_ops_using_ranges): Likewise.
3746 (simplify_div_or_mod_using_ranges): Likewise.
3747 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
3748 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
3749 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
3750 (simplify_float_conversion_using_ranges): Likewise.
3751 (simplify_internal_call_using_ranges): Likewise.
3752 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
3753 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
3754 poitner via x_vr_values for calls into gimple folder.
3755 (vrp_initialize_lattice): Make this the vr_values ctor.
3756 (vrp_free_lattice): Make this the vr_values dtor.
3757 (set_vr_value): New function.
3758 (class vrp_prop): Add vr_values data member. Add various member
3759 functions as well as member functions that delegate to vr_values.
3760 (check_array_ref): Make a member function within vrp_prop class.
3761 (search_for_addr_array, vrp_initialize): Likewise.
3762 (vrp_finalize): Likewise. Revamp to avoid direct access to
3763 vr_value, values_propagated, etc.
3764 (check_array_bounds): Extract vrp_prop class instance pointer from
3765 walk info structure. Use it to call member functions.
3766 (check_all_array_refs): Make a member function within vrp_prop class.
3767 Smuggle class instance pointer via walk info structure.
3768 (x_vr_values): New local static.
3769 (vrp_valueize): Use x_vr_values to get class instance.
3770 (vr_valueize_1): Likewise.
3771 (class vrp_folder): Add vr_values data member. Add various member
3772 functions as well as member functions that delegate to vr_values.
3773 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
3774 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
3775 class instance from vr_values. Use it to call member functions.
3776 (vrp_dom_walker): Add vr_values data member.
3777 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
3778 instance via x_vr_values.
3779 (identify_jump_threads): Accept vr_values as argument. Store
3780 it into the walker structure.
3781 (evrp_dom_walker): Add vr_values class data member. Add various
3783 (evrp_dom_walker::try_find_new_range): Use vr_values data
3784 member to access the memory allocator.
3785 (evrp_dom_walker::before_dom_children): Store vr_values class
3786 instance into the vrp_folder class.
3787 (evrp_dom_walker::push_value_range): Rework to avoid direct
3788 access to num_vr_values and vr_value.
3789 (evrp_dom_walker::pop_value_range): Likewise.
3790 (execute_early_vrp): Remove call to vrp_initialize_lattice.
3791 Use vr_values to get to dump_all_value_ranges member function.
3792 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
3793 and simplify_cond_using_ranges_2 via vrp_prop class instance.
3794 Pass vr_values class instance down to identify_jump_threads.
3795 Remove call to vrp_free_lattice.
3796 (debug_all_value_ranges): Remove.
3798 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
3799 (vrp_folder): Likewise.
3801 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
3802 Get it from the existing bitmap instead.
3803 (vrp_intersect_ranges_1): Likewise.
3805 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3807 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3809 (store_immediate_info::store_immediate_info): Add bitnotp argument,
3810 set bit_not_p to it.
3811 (imm_store_chain_info::coalesce_immediate_stores): Break group
3812 if bit_not_p is different.
3813 (count_multiple_uses, split_group,
3814 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
3815 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
3816 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
3817 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
3818 to store_immediate_info ctor.
3820 2017-11-09 Jim Wilson <jimw@sifive.com>
3822 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
3823 (scan_prog_file): Likewise.
3825 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3827 * bb-reorder.c (max_entry_frequency): Remove.
3828 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
3829 connect_traces, push_to_next_round_p): Remove prototypes.
3830 (find_traces_1_round): Use counts only.
3831 (push_to_next_round_p): Likewise.
3832 (find_traces): Likewise.
3833 (rotate_loop): Likewise.
3834 (find_traces_1_round): Likewise.
3835 (connect_traces): Likewise.
3836 (edge_order): Likewise.
3838 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
3840 * config/arm/arm.c (output_return_instruction): Add comments to
3841 indicate requirement for cmse_nonsecure_entry return to account
3842 for the size of clearing instruction output here.
3843 (thumb_exit): Likewise.
3844 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
3845 return in hardfloat mode.
3847 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
3849 * config/rs6000/rs6000.c (machine_function): Add a bool,
3850 "toc_is_wrapped_separately".
3851 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
3852 if it wasn't explicitly set or unset, we are optimizing for speed, and
3853 doing separate shrink-wrapping.
3854 (rs6000_get_separate_components): Enable the TOC component if
3855 saving the TOC register in the prologue.
3856 (rs6000_components_for_bb): Handle the TOC component.
3857 (rs6000_emit_prologue_components): Store the TOC register where needed.
3858 (rs6000_set_handled_components): Mark TOC as handled, if handled.
3859 (rs6000_emit_prologue): Don't save the TOC if that is already done.
3861 2017-11-09 Martin Jambor <mjambor@suse.cz>
3863 * ipa-param-manipulation.c: New file.
3864 * ipa-param-manipulation.h: Likewise.
3865 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
3866 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
3867 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
3868 (ipa_parm_adjustment): Likewise.
3869 (ipa_parm_adjustment_vec): Likewise.
3870 (ipa_get_vector_of_formal_parms): Moved declaration to
3871 ipa-param-manipulation.h.
3872 (ipa_get_vector_of_formal_parm_types): Likewise.
3873 (ipa_modify_formal_parameters): Likewise.
3874 (ipa_modify_call_arguments): Likewise.
3875 (ipa_combine_adjustments): Likewise.
3876 (ipa_dump_param_adjustments): Likewise.
3877 (ipa_modify_expr): Likewise.
3878 (ipa_get_adjustment_candidate): Likewise.
3879 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
3880 ipa-param-manipulation.c.
3881 (ipa_get_vector_of_formal_parm_types): Likewise.
3882 (ipa_modify_formal_parameters): Likewise.
3883 (ipa_modify_call_arguments): Likewise.
3884 (ipa_modify_expr): Likewise.
3885 (get_ssa_base_param): Likewise.
3886 (ipa_get_adjustment_candidate): Likewise.
3887 (index_in_adjustments_multiple_times_p): Likewise.
3888 (ipa_combine_adjustments): Likewise.
3889 (ipa_dump_param_adjustments): Likewise.
3890 * tree-sra.c: Also include ipa-param-manipulation.h
3891 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
3894 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3895 Alan Hayward <alan.hayward@arm.com>
3896 David Sherwood <david.sherwood@arm.com>
3898 * doc/sourcebuild.texi (vect_masked_store): Document.
3900 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3901 Alan Hayward <alan.hayward@arm.com>
3902 David Sherwood <david.sherwood@arm.com>
3904 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
3906 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3907 Alan Hayward <alan.hayward@arm.com>
3908 David Sherwood <david.sherwood@arm.com>
3910 * doc/sourcebuild.texi (vect_variable_length): Document.
3912 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3913 Alan Hayward <alan.hayward@arm.com>
3914 David Sherwood <david.sherwood@arm.com>
3916 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
3918 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3919 Alan Hayward <alan.hayward@arm.com>
3920 David Sherwood <david.sherwood@arm.com>
3922 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
3924 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3925 Alan Hayward <alan.hayward@arm.com>
3926 David Sherwood <david.sherwood@arm.com>
3928 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
3929 previously undocumented selectors.
3930 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
3932 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3934 * doc/rtl.texi (const_vector): Say that elements can be
3935 const_wide_ints too.
3936 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
3937 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
3938 (gen_vec_duplicate): Use it instead of CONSTANT_P.
3939 * optabs.c (expand_vector_broadcast): Likewise.
3941 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3942 Alan Hayward <alan.hayward@arm.com>
3943 David Sherwood <david.sherwood@arm.com>
3945 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
3946 scaled index even if the unscaled address was invalid.
3947 Don't increase the complexity of using a scale in that case.
3949 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3950 Alan Hayward <alan.hayward@arm.com>
3951 David Sherwood <david.sherwood@arm.com>
3953 * doc/rtl.texi: Rewrite the subreg rules so that they partition
3954 the inner register into REGMODE_NATURAL_SIZE bytes rather than
3955 UNITS_PER_WORD bytes.
3956 * emit-rtl.c (validate_subreg): Divide subregs into blocks
3957 based on REGMODE_NATURAL_SIZE of the inner mode.
3958 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
3959 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
3960 * expmed.c (lowpart_bit_field_p): Divide the value up into
3961 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
3962 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
3963 whether something is likely to occupy more than one register.
3965 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3968 * ipa-inline-transform.c (update_noncloned_frequencies): Use
3969 profile_count::adjust_for_ipa_scaling.
3970 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
3971 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
3973 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
3975 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3977 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
3978 (split_group): Add total_orig and total_new arguments, estimate the
3979 number of statements related to the store group without store merging
3980 and with store merging.
3981 (imm_store_chain_info::output_merged_store): Adjust split_group
3982 callers, punt if estimated number of statements with store merging
3983 is not smaller than estimated number of statements without it.
3985 (handled_load): Remove has_single_use checks.
3986 (pass_store_merging::process_store): Likewise.
3988 2017-11-09 Richard Biener <rguenther@suse.de>
3990 PR tree-optimization/82902
3991 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
3993 2017-11-09 Martin Liska <mliska@suse.cz>
3996 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
3999 2017-11-09 Martin Liska <mliska@suse.cz>
4001 PR tree-optimization/82669
4002 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4004 2017-11-09 Martin Liska <mliska@suse.cz>
4006 PR gcov-profile/48463
4007 * coverage.c (coverage_begin_function): Output also end locus
4008 of a function and information whether the function is
4010 * gcov-dump.c (tag_function): Parse and print the information.
4011 * gcov.c (INCLUDE_MAP): Add include.
4012 (INCLUDE_SET): Likewise.
4013 (struct line_info): Move earlier in the source file because
4014 of vector<line_info> in function_info structure.
4015 (line_info::line_info): Likewise.
4016 (line_info::has_block): Likewise.
4017 (struct source_info): Add new member index.
4018 (source_info::get_functions_at_location): New function.
4019 (function_info::group_line_p): New function.
4020 (output_intermediate_line): New function.
4021 (output_intermediate_file): Use the mentioned function.
4022 (struct function_start): New.
4023 (struct function_start_pair_hash): Likewise.
4024 (process_file): Add code that identifies group functions.
4025 Assign lines either to global or function scope.
4026 (generate_results): Skip artificial functions.
4027 (find_source): Assign index for each source file.
4028 (read_graph_file): Read new flag artificial and end_line.
4029 (add_line_counts): Assign it either to global of function scope.
4030 (accumulate_line_counts): Isolate core of the function to
4031 accumulate_line_info and call it for both function and global
4033 (accumulate_line_info): New function.
4034 (output_line_beginning): Fix GNU coding style.
4035 (print_source_line): New function.
4036 (output_line_details): Likewise.
4037 (output_function_details): Likewise.
4038 (output_lines): Iterate both source (global) scope and function
4040 (struct function_line_start_cmp): New class.
4041 * doc/gcov.texi: Reflect changes in documentation.
4043 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4046 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4047 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4048 and similarly for not instead of neg.
4050 2017-11-08 Andi Kleen <ak@linux.intel.com>
4052 * config/i386/i386.opt: Add -mforce-indirect-call.
4053 * config/i386/predicates.md: Check for flag_force_indirect_call.
4054 * doc/invoke.texi: Document -mforce-indirect-call
4056 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4058 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
4060 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
4061 config/riscv/riscv.c (predict.h): New include.
4062 (riscv_slow_unaligned_access_p): No longer static.
4063 (riscv_block_move_straight): Add require.
4064 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4066 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4069 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
4070 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
4071 nonimmediate_operand predicate for operand 1 instead of
4074 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4076 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
4078 * config/aarch64/constraints.md (Uml): New constraint.
4079 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
4082 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4084 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
4085 of two vec_duplicates into a vec_concat.
4087 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4089 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4090 Simplify vec_merge of vec_duplicate and vec_concat.
4091 * config/aarch64/constraints.md (Utq): New constraint.
4092 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
4095 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4097 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4098 Simplify vec_merge of vec_duplicate and const_vector.
4099 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
4101 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
4102 mode iterator. Update predicate on operand 1 to
4103 handle non-const_vec constants. Delete constraints.
4104 (*aarch64_combinez_be<mode>): Likewise for operand 2.
4106 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4108 PR tree-optimization/78821
4109 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
4111 (store_operand_info::store_operand_info): Initialize it to false.
4112 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
4113 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
4114 store in the group, and if chain_info is non-NULL, to ignore altogether
4116 (compatible_load_p): Fail if bit_not_p does not match.
4117 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
4118 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
4119 (pass_store_merging::process_store): Adjust
4120 terminate_all_aliasing_chains calls to pass NULL in all current spots,
4121 call terminate_all_aliasing_chains newly when adding a store into
4122 a chain with non-NULL chain_info.
4124 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4126 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
4128 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4130 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4132 (aarch64_layout_frame): Initialise emit_frame_chain.
4133 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
4134 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
4136 2017-11-08 Martin Liska <mliska@suse.cz>
4138 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
4139 of gimple_call_internal_p.
4141 2017-11-07 Tom de Vries <tom@codesourcery.com>
4143 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
4145 2017-11-07 Tom de Vries <tom@codesourcery.com>
4147 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4150 2017-11-08 Martin Liska <mliska@suse.cz>
4153 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
4155 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4157 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
4159 2017-11-07 Jakub Jelinek <jakub@redhat.com>
4162 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
4163 (*cmp<mode>_ccz_1): New insn with $k alternative.
4166 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
4167 RTX_COMM_COMPARE as commutative as well.
4168 (ix86_binary_operator_ok): Formatting fix.
4169 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
4170 *<code><mode>3<mask_name><round_saeonly_name>,
4171 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
4172 *<s>mul<mode>3_highpart<mask_name>,
4173 *vec_widen_umult_even_v16si<mask_name>,
4174 *vec_widen_umult_even_v8si<mask_name>,
4175 *vec_widen_umult_even_v4si<mask_name>,
4176 *vec_widen_smult_even_v16si<mask_name>,
4177 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
4178 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
4179 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
4180 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
4181 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
4182 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
4183 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
4184 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
4185 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
4186 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
4187 ix86_binary_operator_ok. Formatting fixes.
4188 (*<plusminus_insn><mode>3<mask_name><round_name>,
4189 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
4192 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
4194 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
4195 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
4196 eq and ne if TARGET_ISEL.
4197 (cmp): New code_iterator.
4198 (UNS, UNSU_, UNSIK): New code_attrs.
4199 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
4200 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
4202 ("eq<mode>3"): ... this.
4203 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
4205 ("ne<mode>3"): ... this.
4207 2017-11-07 Julia Koval <julia.koval@intel.com>
4210 * common/config/i386/i386-common.c
4211 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
4212 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
4213 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
4214 * config/i386/i386.opt: Ditto.
4215 * config/i386/i386.c (ix86_target_string): Ditto.
4216 (ix86_option_override_internal): Ditto.
4217 (ix86_init_mpx_builtins): Move MPX to args2.
4218 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
4219 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4220 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4221 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4222 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4223 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4224 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
4226 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
4229 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
4230 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
4231 (zero-extendsidi peephole2): Remove peephole.
4233 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4236 * stor-layout.c: Include attribs.h.
4237 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
4238 explicit lookup of "aligned" attribute.
4240 2017-11-07 Andrew Waterman <andrew@sifive.com>
4242 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
4243 (riscv_expand_block_move): Likewise.
4244 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
4246 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
4247 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
4248 * config/riscv/riscv.c (riscv_block_move_straight): New function.
4249 (riscv_adjust_block_mem): Likewise.
4250 (riscv_block_move_loop): Likewise.
4251 (riscv_expand_block_move): Likewise.
4252 * config/riscv/riscv.md (movmemsi): New pattern.
4254 2017-11-07 Michael Clark <michaeljclark@mac.com>
4256 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
4257 (MUSL_DYNAMIC_LINKER): Likewise.
4259 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
4261 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
4262 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
4265 2017-11-07 Richard Biener <rguenther@suse.de>
4267 * match.pd: Fix build.
4269 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
4270 Jackson Woodruff <jackson.woodruff@arm.com>
4272 PR tree-optimization/71026
4273 * match.pd: Canonicalize negate in division.
4275 2017-11-07 Sudakshina Das <sudi.das@arm.com>
4278 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
4280 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4282 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
4283 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
4285 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4287 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
4288 non-scalar integral types.
4289 * match.pd (negate_expr_p): Handle MINUS_EXPR.
4290 (-(A-B), -(~A)): New transformations.
4292 2017-11-07 Tom de Vries <tom@codesourcery.com>
4294 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
4295 semicolon after "do {} while (0)".
4296 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4297 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4298 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4299 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4300 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4302 2017-11-07 Tom de Vries <tom@codesourcery.com>
4304 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
4305 after "do {} while (0)".
4306 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4307 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4308 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4309 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4310 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4312 2017-11-07 Tom de Vries <tom@codesourcery.com>
4315 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
4317 (arm_rtx_costs_internal): Add missing semicolon after
4318 HANDLE_NARROW_SHIFT_ARITH call.
4320 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4322 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4323 disable isel if it was not set explicitly.
4325 2017-11-06 James Bowman <james.bowman@ftdichip.com>
4327 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
4328 (add_type_attribute) likewise.
4330 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
4332 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
4333 of struct ix86_frame.
4334 (ix86_initial_elimination_offset): Likewise.
4335 (ix86_expand_split_stack_prologue): Likewise.
4337 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
4339 * tree-vrp.h (enum value_range_type): Update stale comment.
4341 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4342 Alan Hayward <alan.hayward@arm.com>
4343 David Sherwood <david.sherwood@arm.com>
4345 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
4346 (aarch64_expand_vec_perm_const): Take the number of units too.
4347 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
4348 (aarch64_expand_vec_perm_const): Likewise.
4349 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
4350 (vec_perm<mode>): Update accordingly.
4352 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4353 Alan Hayward <alan.hayward@arm.com>
4354 David Sherwood <david.sherwood@arm.com>
4356 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
4357 Take the number of units too.
4358 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
4359 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
4360 but check for a vector mode before rather than after the call.
4361 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
4362 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
4363 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
4364 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
4365 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
4366 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
4367 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
4368 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
4369 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
4370 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
4371 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
4372 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
4373 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
4374 (aarch64_sqdmull2_n<mode>): Update accordingly.
4376 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4377 Alan Hayward <alan.hayward@arm.com>
4378 David Sherwood <david.sherwood@arm.com>
4380 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
4381 the number of units too.
4382 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
4383 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
4384 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
4385 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
4386 (vec_store_lanesxi<mode>): Update accordingly.
4388 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4389 Alan Hayward <alan.hayward@arm.com>
4390 David Sherwood <david.sherwood@arm.com>
4392 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
4393 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
4394 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
4395 of units rather than the mode.
4396 * config/aarch64/iterators.md (nunits): New mode attribute.
4397 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4398 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
4399 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
4400 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
4401 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
4402 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
4403 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
4404 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
4405 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
4406 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
4407 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4408 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
4409 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
4410 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
4411 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
4412 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
4413 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
4414 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
4415 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
4416 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
4417 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
4418 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
4419 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
4420 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4421 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
4422 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
4423 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4424 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
4425 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
4426 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
4427 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
4428 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
4429 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
4430 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
4431 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
4432 (aarch64_simd_vec_setv2di): Likewise.
4434 2017-11-06 Carl Love <cel@us.ibm.com>
4436 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
4438 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
4439 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
4441 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
4443 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
4444 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
4445 Add power 8 macro expansions.
4446 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
4447 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
4448 power 8 instructions. (VSX_XXBR): Add iterator.
4450 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
4452 * config/arm/arm.md (predicable_short_it): Change default to "no",
4453 improve documentation, remove uses that are identical to the default.
4454 (enabled_for_depr_it): Rename to enabled_for_short_it.
4455 * gcc/config/arm/arm-fixed.md (predicable_short_it):
4456 Remove default uses.
4457 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
4458 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
4459 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
4460 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
4462 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
4465 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
4466 float128 helper macros, which are no longer used after deleting
4467 the old 'q' built-in functions, and moving the round to odd
4468 built-in functions to being special built-in functions.
4469 (BU_FLOAT128_2): Likewise.
4470 (BU_FLOAT128_1_HW): Likewise.
4471 (BU_FLOAT128_2_HW): Likewise.
4472 (BU_FLOAT128_3_HW): Likewise.
4473 (FABSQ): Delete old 'q' built-in functions.
4474 (COPYSIGNQ): Likewise.
4475 (SQRTF128_ODD): Move round to odd built-in functions to be
4476 special built-in functions, so that we can handle
4477 -mabi=ieeelongdouble.
4478 (TRUNCF128_ODD): Likewise.
4479 (ADDF128_ODD): Likewise.
4480 (SUBF128_ODD): Likewise.
4481 (MULF128_ODD): Likewise.
4482 (DIVF128_ODD): Likewise.
4483 (FMAF128_ODD): Likewise.
4484 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
4485 built-in names to 'f128'.
4486 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
4487 old 'q' built-in functions, as the machine independent code for
4488 'f128' built-in functions handles this.
4489 (rs6000_expand_builtin): Add expansion for float128 round to odd
4490 functions, keying off on -mabi=ieeelongdouble of whether to use
4491 the KFmode or TFmode variant.
4492 (rs6000_init_builtins): Initialize the _Float128 round to odd
4494 * doc/extend.texi (PowerPC Built-in Functions): Document the old
4495 _Float128 'q' built-in functions are now mapped into the new
4496 'f128' built-in functions.
4498 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
4500 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
4501 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
4503 2017-11-06 David Malcolm <dmalcolm@redhat.com>
4506 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
4507 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
4508 * toplev.c: Include "ipa-fnsummary.h".
4509 (toplev::finalize): Call ipa_fnsummary_c_finalize.
4511 2017-11-06 Jakub Jelinek <jakub@redhat.com>
4513 PR tree-optimization/82838
4514 * gimple-ssa-store-merging.c
4515 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
4516 on a separate gimple_seq which is then appended to seq.
4518 2017-11-06 Jeff Law <law@redhat.com>
4521 * config/i386/i386.c (PROBE_INTERVAL): Remove.
4522 (get_probe_interval): New functions.
4523 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
4524 (ix86_adjust_stack_and_probe): Likewise.
4525 (output_adjust_stack_and_probe): Likewise.
4526 (ix86_emit_probe_stack_range): Likewise.
4527 (ix86_expand_prologue): Likewise.
4529 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4531 PR tree-optimization/82816
4532 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
4533 if the modes of the two types are the same.
4534 (convert_plusminus_to_widen): Likewise.
4536 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4538 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
4540 (usadv16qi): New define_expand.
4541 (usadv8hi): New define_expand.
4543 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
4546 * ipa-inline-transform.c (update_noncloned_frequencies): Always
4548 (inline_transform): Likewise.
4549 * predict.c (counts_to_freqs): Remove useless conditional.
4550 * profile-count.h (profile_count::apply_scale): Move sanity check.
4551 * tree-inline.c (copy_bb): Always scale.
4552 (copy_cfg_body): Likewise.
4554 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
4557 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
4559 (*cmp_ite0): Add enabled_for_depr_it attribute.
4560 (*cmp_ite1): Likewise.
4562 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4564 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
4567 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4569 * tree-vrp.c (vrp_int_const_binop): Return true on success and
4570 return the value by pointer.
4571 (extract_range_from_multiplicative_op_1): Update accordingly.
4572 Return as soon as an operation fails.
4574 2017-11-05 Tom de Vries <tom@codesourcery.com>
4577 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
4578 (DEF_SANITIZER_BUILTIN): ... here.
4579 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
4580 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
4582 2017-11-05 Tom de Vries <tom@codesourcery.com>
4585 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
4587 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
4588 ASM_OUTPUT_BEFORE_CASE_LABEL call.
4589 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
4591 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4592 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4593 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4595 2017-11-05 Tom de Vries <tom@codesourcery.com>
4598 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
4601 2017-11-04 Michael Clark <michaeljclark@mac.com>
4603 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
4604 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
4606 (*addsi3_extended): Likewise.
4607 (*addsi3_extended2): Likewise.
4608 (<optab>si3): Likewise.
4609 (<optab>di3): Likewise.
4610 (<optab><mode>3): Likewise.
4611 (<*optabe>si3_internal): Likewise.
4612 (zero_extendqi<SUPERQI:mode>2): Likewise.
4613 (*add<mode>hi3): Likewise.
4614 (*xor<mode>hi3): Likewise.
4615 (<optab>di3): Likewise.
4616 (*<optab>si3_extend): Likewise.
4617 (*sge<u>_<X:mode><GPR:mode>): Likewise.
4618 (*slt<u>_<X:mode><GPR:mode>): Likewise.
4619 (*sle<u>_<X:mode><GPR:mode>): Likewise.
4621 2017-11-04 Andrew Waterman <andrew@sifive.com>
4623 * config/riscv/riscv.c (riscv_option_override): Conditionally set
4624 TARGET_STRICT_ALIGN based upon -mtune argument.
4626 2017-11-04 Andrew Waterman <andrew@sifive.com>
4628 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
4630 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
4632 * config/i386/i386.c (choose_basereg): Use optional scratch
4633 register and add assertion.
4634 (x86_emit_outlined_ms2sysv_save): Use scratch register when
4635 needed, and don't allocate stack.
4636 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
4637 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
4638 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
4640 2017-11-03 Jeff Law <law@redhat.com>
4642 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
4643 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
4644 to probe at the start of a noreturn function.
4646 2017-11-03 Jakub Jelinek <jakub@redhat.com>
4648 PR tree-optimization/78821
4649 * gimple-ssa-store-merging.c: Update the file comment.
4650 (MAX_STORE_ALIAS_CHECKS): Define.
4651 (struct store_operand_info): New type.
4652 (store_operand_info::store_operand_info): New constructor.
4653 (struct store_immediate_info): Add rhs_code and ops data members.
4654 (store_immediate_info::store_immediate_info): Add rhscode, op0r
4655 and op1r arguments to the ctor, initialize corresponding data members.
4656 (struct merged_store_group): Add load_align_base and load_align
4658 (merged_store_group::merged_store_group): Initialize them.
4659 (merged_store_group::do_merge): Update them.
4660 (merged_store_group::apply_stores): Pick the constant for
4661 encode_tree_to_bitpos from one of the two operands, or skip
4662 encode_tree_to_bitpos if neither operand is a constant.
4663 (class pass_store_merging): Add process_store method decl. Remove
4664 bool argument from terminate_all_aliasing_chains method decl.
4665 (pass_store_merging::terminate_all_aliasing_chains): Remove
4666 var_offset_p argument and corresponding handling.
4667 (stmts_may_clobber_ref_p): New function.
4668 (compatible_load_p): New function.
4669 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
4670 if there is overlap and rhs_code is not INTEGER_CST. For
4671 non-overlapping stores terminate group if rhs is not mergeable.
4672 (get_alias_type_for_stmts): Change first argument from
4673 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
4674 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
4675 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
4677 (get_location_for_stmts): Change first argument from
4678 auto_vec<gimple *> & to vec<gimple *> &.
4679 (struct split_store): Remove orig_stmts data member, add orig_stores.
4680 (split_store::split_store): Create orig_stores rather than orig_stmts.
4681 (find_constituent_stmts): Renamed to ...
4682 (find_constituent_stores): ... this. Change second argument from
4683 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
4684 to info structures rather than the statements.
4685 (split_group): Rename ALLOW_UNALIGNED argument to
4686 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
4687 it. Adjust find_constituent_stores caller.
4688 (imm_store_chain_info::output_merged_store): Handle rhs_code other
4689 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
4690 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
4691 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
4692 (mem_valid_for_store_merging): New function.
4693 (handled_load): New function.
4694 (pass_store_merging::process_store): New method.
4695 (pass_store_merging::execute): Use process_store method. Adjust
4696 terminate_all_aliasing_chains caller.
4698 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4700 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4701 Return true for more constants, symbols and label references.
4702 (aarch64_valid_floating_const): Remove unused function.
4704 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4707 * config/aarch64/aarch64.c (aarch64_layout_frame):
4708 Undo forcing of LR at bottom of frame.
4710 2017-11-03 Jeff Law <law@redhat.com>
4713 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
4714 for int_registers_saved.
4716 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
4717 extracted from tree-ssa-dom.c.
4718 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
4719 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
4720 (record_equivalences_from_incoming_edge): Add additional argument
4721 to single_pred_edge_ignoring_loop_edges call.
4722 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
4723 (uncprop_dom_walker::before_dom_children): Add additional argument
4724 to single_pred_edge_ignoring_loop_edges call.
4725 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
4726 single_pred_edge_ignoring_loop_edges rather than open coding.
4727 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
4729 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
4731 * match.pd (-(-A)): Rewrite.
4733 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4735 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
4736 (rs6000_emit_int_cmove): New declaration.
4737 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
4738 (rs6000_emit_sISEL): Delete.
4739 (rs6000_emit_int_cmove): Make non-static.
4740 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
4741 instead of rs6000_emit_sISEL.
4743 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
4745 * asan.c (create_cond_insert_point): Maintain profile.
4746 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
4748 * basic-block.h (struct basic_block_def): Remove frequency.
4749 (EDGE_FREQUENCY): Use to_frequency
4750 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
4752 (find_traces): Update to use to_frequency.
4753 (find_traces_1_round): Likewise; use only IPA counts.
4754 (bb_to_key): Likewise.
4755 (connect_traces): Use IPA counts only.
4756 (copy_bb_p): Update to use to_frequency.
4757 (fix_up_crossing_landing_pad): Likewise.
4758 (sanitize_hot_paths): Likewise.
4759 * bt-load.c (basic_block_freq): Likewise.
4760 * cfg.c (init_flow): Set count_max to uninitialized.
4761 (check_bb_profile): Remove frequencies; check counts.
4762 (dump_bb_info): Do not dump frequencies.
4763 (update_bb_profile_for_threading): Update counts only.
4764 (scale_bbs_frequencies_int): Likewise.
4765 (MAX_SAFE_MULTIPLIER): Remove.
4766 (scale_bbs_frequencies_gcov_type): Update counts only.
4767 (scale_bbs_frequencies_profile_count): Update counts only.
4768 (scale_bbs_frequencies): Update counts only.
4769 * cfg.h (struct control_flow_graph): Add count-max.
4770 (update_bb_profile_for_threading): Update prototype.
4771 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
4772 (find_many_sub_basic_blocks): Likewise.
4773 * cfgcleanup.c (try_forward_edges): Likewise.
4774 (try_crossjump_to_edge): Likewise.
4775 * cfgexpand.c (expand_gimple_cond): Likewise.
4776 (expand_gimple_tailcall): Likewise.
4777 (construct_init_block): Likewise.
4778 (construct_exit_block): Likewise.
4779 * cfghooks.c (verify_flow_info): Check consistency of counts.
4780 (dump_bb_for_graph): Do not dump frequencies.
4781 (split_block_1): Do not update frequencies.
4782 (split_edge): Do not update frequencies.
4783 (make_forwarder_block): Do not update frequencies.
4784 (duplicate_block): Do not update frequencies.
4785 (account_profile_record): Do not update frequencies.
4786 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
4787 for global heuristics.
4788 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
4789 (expected_loop_iterations_unbounded): Use counts only.
4790 * cfgloopmanip.c (scale_loop_profile): Simplify.
4791 (create_empty_loop_on_edge): Simplify
4793 (duplicate_loop_to_header_edge): Simplify
4794 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
4795 (update_br_prob_note): Take care of removing note when profile
4797 (relink_block_chain): Do not dump frequency.
4798 (rtl_account_profile_record): Use to_frequency.
4799 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
4800 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
4801 (cgraph_update_edges_for_call_stmt_node): Likewise.
4802 (cgraph_edge::verify_count_and_frequency): Update.
4803 (cgraph_node::verify_node): Temporarily disable frequency verification.
4804 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
4805 to_cgraph_frequency.
4806 (cgraph_edge::rebuild_edges): Convert to ipa counts.
4807 * cgraphunit.c (init_lowered_empty_function): Do not initialize
4809 (cgraph_node::expand_thunk): Update profile.
4810 * except.c (dw2_build_landing_pads): Do not update frequency.
4811 * final.c (compute_alignments): Use to_frequency.
4812 (dump_basic_block_info): Do not dump frequency.
4813 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
4814 (dump_gimple_bb_header): Do not dump frequency.
4815 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
4817 * gimple-streamer-in.c (input_bb): Do not stream frequency.
4818 * gimple-streamer-out.c (output_bb): Do not stream frequency.
4819 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
4820 (init_before_recovery): Do not update frequency.
4821 (sched_create_recovery_edges): Do not update frequency.
4822 * hsa-gen.c (convert_switch_statements): Do not update frequency.
4823 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
4824 (ipa_cp_c_finalize): Set max_count to uninitialized.
4825 * ipa-fnsummary.c (get_minimal_bb): Use counts.
4826 (param_change_prob): Use counts.
4827 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
4829 * ipa-split.c (consider_split): Use to_frequency.
4830 (split_function): Use to_frequency.
4831 * ira-build.c (loop_compare_func): Likewise.
4832 (mark_loops_for_removal): Likewise.
4833 (mark_all_loops_for_removal): Likewise.
4834 * loop-doloop.c (doloop_modify): Do not update frequency.
4835 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
4837 * lto-streamer-in.c (input_function): Update count_max.
4838 * omp-expand.c (expand_omp_taskreg): Update count_max.
4839 * omp-simd-clone.c (simd_clone_adjust): Update profile.
4840 * predict.c (maybe_hot_frequency_p): Use to_frequency.
4841 (maybe_hot_count_p): Use ipa counts only.
4842 (maybe_hot_bb_p): Simplify.
4843 (maybe_hot_edge_p): Simplify.
4844 (probably_never_executed): Do not take frequency argument.
4845 (probably_never_executed_bb_p): Do not pass frequency.
4846 (probably_never_executed_edge_p): Likewise.
4847 (combine_predictions_for_bb): Check that profile is nonzero.
4848 (propagate_freq): Do not set frequency.
4849 (drop_profile): Simplify.
4850 (counts_to_freqs): Simplify.
4851 (expensive_function_p): Use to_frequency.
4852 (propagate_unlikely_bbs_forward): Simplify.
4853 (determine_unlikely_bbs): Simplify.
4854 (estimate_bb_frequencies): Add hack to silence graphite issues.
4855 (compute_function_frequency): Use ipa counts.
4856 (pass_profile::execute): Update.
4857 (rebuild_frequencies): Use counts only.
4858 (force_edge_cold): Use counts only.
4859 * profile-count.c (profile_count::dump): Dump new count types.
4860 (profile_count::differs_from_p): Check compatiblity.
4861 (profile_count::to_frequency): New function.
4862 (profile_count::to_cgraph_frequency): New function.
4863 * profile-count.h (struct function): Declare.
4864 (enum profile_quality): Add profile_guessed_local and
4865 profile_guessed_global0.
4866 (class profile_proability): Decrease number of bits to 29;
4867 update from_reg_br_prob_note and to_reg_br_prob_note.
4868 (class profile_count: Update comment; decrease number of bits
4869 to 61. Check compatibility.
4870 (profile_count::compatible_p): New private member function.
4871 (profile_count::ipa_p): New member function.
4872 (profile_count::operator<): Handle global zero correctly.
4873 (profile_count::operator>): Handle global zero correctly.
4874 (profile_count::operator<=): Handle global zero correctly.
4875 (profile_count::operator>=): Handle global zero correctly.
4876 (profile_count::nonzero_p): New member function.
4877 (profile_count::force_nonzero): New member function.
4878 (profile_count::max): New member function.
4879 (profile_count::apply_scale): Handle IPA scalling.
4880 (profile_count::guessed_local): New member function.
4881 (profile_count::global0): New member function.
4882 (profile_count::ipa): New member function.
4883 (profile_count::to_frequency): Declare.
4884 (profile_count::to_cgraph_frequency): Declare.
4885 * profile.c (OVERLAP_BASE): Delete.
4886 (compute_frequency_overlap): Delete.
4887 (compute_branch_probabilities): Do not use compute_frequency_overlap.
4888 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
4889 * sched-ebb.c (rank): Use counts only.
4890 * shrink-wrap.c (handle_simple_exit): Use counts only.
4891 (try_shrink_wrapping): Use counts only.
4892 (place_prologue_for_one_component): Use counts only.
4893 * tracer.c (find_best_predecessor): Use to_frequency.
4894 (find_trace): Use to_frequency.
4895 (tail_duplicate): Use to_frequency.
4896 * trans-mem.c (expand_transaction): Do not update frequency.
4897 * tree-call-cdce.c: Do not update frequency.
4898 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
4899 (gimple_merge_blocks): Likewise.
4900 (gimple_split_edge): Likewise.
4901 (gimple_duplicate_sese_region): Likewise.
4902 (gimple_duplicate_sese_tail): Likewise.
4903 (move_sese_region_to_fn): Likewise.
4904 (gimple_account_profile_record): Likewise.
4905 (insert_cond_bb): Likewise.
4906 * tree-complex.c (expand_complex_div_wide): Likewise.
4907 * tree-eh.c (lower_resx): Update profile.
4908 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
4910 (initialize_cfun): Do not initialize frequencies
4911 (freqs_to_counts): Delete.
4912 (copy_cfg_body): Ignore count parameter.
4913 (copy_body): Update.
4914 (expand_call_inline): Update count_max.
4915 (optimize_inline_calls): Update count_max.
4916 (tree_function_versioning): Update count_max.
4917 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
4918 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
4920 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
4921 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
4922 (try_peel_loop): Likewise.
4923 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
4925 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
4926 (tree_transform_and_unroll_loop): Do not use frequencies
4927 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
4928 Use reliable prediction only.
4929 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
4930 * tree-ssa-sink.c (select_best_block): Use to_frequency.
4931 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
4932 probability scaling.
4933 * tree-ssa-threadupdate.c (create_block_for_threading): Do
4934 not update frequency
4935 (any_remaining_duplicated_blocks): Likewise.
4936 (update_profile): Likewise.
4937 (estimated_freqs_path): Delete.
4938 (freqs_to_counts_path): Delete.
4939 (clear_counts_path): Delete.
4940 (ssa_fix_duplicate_block_edges): Likewise.
4941 (duplicate_thread_path): Likewise.
4942 * tree-switch-conversion.c (gen_inbound_check): Use counts.
4943 * tree-tailcall.c (decrease_profile): Do not update frequency.
4944 (eliminate_tail_call): Likewise.
4945 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
4946 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
4947 (optimize_mask_stores): Likewise.
4948 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4949 * ubsan.c (ubsan_expand_null_ifn): Update profile.
4950 (ubsan_expand_ptr_ifn): Update profile.
4951 * value-prof.c (gimple_ic): Simplify.
4952 * value-prof.h (gimple_ic): Update prototype.
4953 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
4954 * ipa-inline.c (compute_uninlined_call_time): Be sure that
4956 (want_inline_self_recursive_call_p): Likewise.
4957 (resolve_noninline_speculation): Only cummulate defined counts.
4958 (inline_small_functions): Use nonzero_p.
4959 (ipa_inline): Do not access freed node.
4961 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4963 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4964 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
4966 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
4968 * config/riscv/riscv.c (riscv_legitimize_move): Handle
4969 non-legitimate address.
4971 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4973 * config/rs6000/rs6000.md (*lt0_disi): Delete.
4974 (*lt0_<mode>di, *lt0_<mode>si): New.
4976 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4978 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
4979 TARGET_PAIRED_FLOAT.
4981 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
4982 Jim Wilson <jim.wilson@linaro.org>
4984 * config/aarch64/aarch64-cores.def (saphira): New CPU.
4985 * config/aarch64/aarch64-tune.md: Regenerated.
4986 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
4987 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
4989 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
4991 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
4992 (arc_expand_prologue): Restore blink for millicode.
4993 * config/arc/linux.h (LINK_EH_SPEC): Defined.
4995 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
4998 * config/i386/i386.c (ix86_vector_duplicate_value): Use
4999 gen_vec_duplicate after forcing the scalar into a register.
5001 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5003 * combine (try_combine): Print the insns input to try_combine to the
5006 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5009 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5010 Remove second argument from aarch64_process_target_attr call.
5011 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5013 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5015 (aarch64_handle_attr_arch): Remove second argument.
5016 (aarch64_handle_attr_cpu): Ditto.
5017 (aarch64_handle_attr_tune): Ditto.
5018 (aarch64_handle_attr_isa_flags): Ditto.
5019 (aarch64_process_one_target_attr): Ditto.
5020 (aarch64_process_target_attr): Ditto.
5021 (aarch64_option_valid_attribute_p): Remove second argument.
5022 on aarch64_process_target_attr call.
5024 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5026 * diagnostic.c: Include "selftest-diagnostic.h".
5027 (selftest::assert_location_text): New function.
5028 (selftest::test_diagnostic_get_location_text): New function.
5029 (selftest::diagnostic_c_tests): Call it.
5031 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5033 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5034 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5035 (class selftest::test_diagnostic_context): Move to...
5036 * selftest-diagnostic.c: New file.
5037 * selftest-diagnostic.h: New file.
5039 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5041 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5042 offset range for FT32B.
5043 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5044 * config/ft32/ft32.md: Add TARGET_NOPM.
5045 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5046 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5048 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5050 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5052 2017-11-02 Jeff Law <law@redhat.com>
5054 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5055 virtual keyword on FINAL OVERRIDE members.
5057 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5058 virtuals together. Add virtual destructor.
5059 (substitute_and_fold_engine): Similarly.
5061 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
5063 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5065 2017-11-02 Richard Biener <rguenther@suse.de>
5067 PR tree-optimization/82795
5068 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
5070 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5072 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
5074 (gcc_GAS_CHECK_FEATURE): Remove.
5075 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
5076 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
5077 * configure: Regenerate.
5079 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
5081 * config/arc/arc.c (hwloop_optimize): Account for empty
5084 2017-11-02 Richard Biener <rguenther@suse.de>
5087 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
5088 Truncate ARRAY_REF index and element size.
5090 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5092 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
5094 2017-11-01 Jeff Law <law@redhat.com>
5096 * tree-ssa-ccp.c (ccp_folder): New class derived from
5097 substitute_and_fold_engine.
5098 (ccp_folder::get_value): New member function.
5099 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
5100 (ccp_fold_stmt): Remove prototype.
5101 (ccp_finalize): Call substitute_and_fold from the ccp_class.
5102 * tree-ssa-copy.c (copy_folder): New class derived from
5103 substitute_and_fold_engine.
5104 (copy_folder::get_value): Renamed from get_value.
5105 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
5106 * tree-vrp.c (vrp_folder): New class derived from
5107 substitute_and_fold_engine.
5108 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
5109 (vrp_folder::get_value): New member function.
5110 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
5111 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
5112 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
5113 provide a class interface to folder/substitute routines.
5114 (ssa_prop_fold_stmt_fn): Remove typedef.
5115 (ssa_prop_get_value_fn): Likewise.
5116 (subsitute_and_fold): Remove prototype.
5117 (replace_uses_in): Likewise.
5118 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
5119 Renamed from replace_uses_in. Call the virtual member function
5120 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
5121 (substitute_and_fold_dom_walker): Remove initialization of
5122 data member entries for calbacks. Add substitute_and_fold_engine
5123 member and initialize it.
5124 (substitute_and_fold_dom_walker::before_dom_children0: Use the
5125 member functions for get_value, replace_phi_args_in c
5126 replace_uses_in, and fold_stmt calls.
5127 (substitute_and_fold_engine::substitute_and_fold): Renamed from
5128 substitute_and_fold. Remove assert. Update ctor call.
5130 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
5131 (ssa_prop_visit_phi_fn): Likewise.
5132 (class ssa_propagation_engine): New class to provide an interface
5134 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
5136 (ssa_prop_visit_phi): Likewise.
5137 (ssa_propagation_engine::simulate_stmt): Moved into class.
5138 Call visit_phi/visit_stmt from the class rather than via
5139 file scoped static variables.
5140 (ssa_propagation_engine::simulate_block): Moved into class.
5141 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
5142 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
5143 set file scoped statics for the visit_stmt/visit_phi callbacks.
5144 * tree-complex.c (complex_propagate): New class derived from
5145 ssa_propagation_engine.
5146 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
5147 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
5148 (tree_lower_complex): Call ssa_propagate via the complex_propagate
5150 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
5151 ssa_propagation_engine.
5152 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
5153 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
5154 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
5155 * tree-ssa-copy.c (copy_prop): New class derived from
5156 ssa_propagation_engine.
5157 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
5158 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
5159 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
5160 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
5161 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
5162 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
5163 (execute_vrp): Call ssa_propagate from the vrp_prop class.
5165 2017-11-01 Jakub Jelinek <jakub@redhat.com>
5167 PR rtl-optimization/82778
5168 PR rtl-optimization/82597
5169 * compare-elim.c (struct comparison): Add in_a_setter field.
5170 (find_comparison_dom_walker::before_dom_children): Remove killed
5171 bitmap and df_simulate_find_defs call, instead walk the defs.
5172 Compute last_setter and initialize in_a_setter. Merge definitions
5173 with first initialization for a few variables.
5174 (try_validate_parallel): Use insn_invalid_p instead of
5175 recog_memoized. Return insn rather than just the pattern.
5176 (try_merge_compare): Fix up comment. Don't uselessly test if
5177 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
5179 (execute_compare_elim_after_reload): Remove df_chain_add_problem
5182 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5183 Alan Hayward <alan.hayward@arm.com>
5184 David Sherwood <david.sherwood@arm.com>
5186 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
5187 aarch64_hard_regno_nregs to get the number of registers
5190 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5191 Alan Hayward <alan.hayward@arm.com>
5192 David Sherwood <david.sherwood@arm.com>
5194 * config/aarch64/constraints.md (Upl): Rename to...
5196 * config/aarch64/aarch64.md
5197 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
5200 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5201 Alan Hayward <alan.hayward@arm.com>
5202 David Sherwood <david.sherwood@arm.com>
5204 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
5205 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
5208 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5209 Alan Hayward <alan.hayward@arm.com>
5210 David Sherwood <david.sherwood@arm.com>
5212 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
5213 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
5214 (aarch64_evpc_dup): Generate rtl direcly, rather than using
5216 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
5217 of a single element.
5218 * config/aarch64/iterators.md: Add a comment above the permute
5219 unspecs to say that they are generated directly by
5220 aarch64_expand_vec_perm_const.
5221 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
5223 2017-11-01 Nathan Sidwell <nathan@acm.org>
5225 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
5227 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5229 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
5230 and medany code models, and describe what they do.
5232 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5234 Revert accidental duplicate:
5236 * combine.c (can_change_dest_mode): Reject changes in
5237 REGMODE_NATURAL_SIZE.
5239 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
5241 PR rtl-optimization/64682
5242 PR rtl-optimization/69567
5243 PR rtl-optimization/69737
5244 PR rtl-optimization/82683
5245 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
5246 register mentioned in the note, drop the note, unless it came from I3,
5247 in which case it should go to I3 again.
5249 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5251 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
5252 and return false if not.
5253 (clear_bytes_written_by, live_bytes_read): Update accordingly.
5255 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5257 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
5258 range is known to be empty.
5260 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5261 Alan Hayward <alan.hayward@arm.com>
5262 David Sherwood <david.sherwood@arm.com>
5264 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
5265 and CONST_VECTOR_NUNITS instead of computing the number of units from
5266 the byte sizes of the vector and element.
5267 (simplify_binary_operation_1): Likewise.
5268 (simplify_const_binary_operation): Likewise.
5269 (simplify_ternary_operation): Likewise.
5271 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5272 Alan Hayward <alan.hayward@arm.com>
5273 David Sherwood <david.sherwood@arm.com>
5275 * var-tracking.c (INT_MEM_OFFSET): Replace with...
5276 (int_mem_offset): ...this new function.
5277 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
5278 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
5279 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
5282 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5283 Alan Hayward <alan.hayward@arm.com>
5284 David Sherwood <david.sherwood@arm.com>
5286 * lower-subreg.c (interesting_mode_p): New function.
5287 (compute_costs, find_decomposable_subregs, decompose_register)
5288 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
5289 (resolve_clobber, dump_choices): Use it.
5291 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5292 Alan Hayward <alan.hayward@arm.com>
5293 David Sherwood <david.sherwood@arm.com>
5295 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
5297 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5298 Alan Hayward <alan.hayward@arm.com>
5299 David Sherwood <david.sherwood@arm.com>
5301 * alias.c (find_base_value, find_base_term): Only process integer
5302 truncations. Check the precision rather than the size.
5304 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5305 Alan Hayward <alan.hayward@arm.com>
5306 David Sherwood <david.sherwood@arm.com>
5308 * machmode.h (is_narrower_int_mode): New function
5309 * optabs.c (expand_float, expand_fix): Use it.
5310 * dwarf2out.c (rotate_loc_descriptor): Likewise.
5312 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5313 Alan Hayward <alan.hayward@arm.com>
5314 David Sherwood <david.sherwood@arm.com>
5316 * rtl.h (narrower_subreg_mode): New function.
5317 * ira-color.c (update_costs_from_allocno): Use it.
5319 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5320 Alan Hayward <alan.hayward@arm.com>
5321 David Sherwood <david.sherwood@arm.com>
5323 * optabs-query.h (convert_optab_p): New function, split out from...
5324 (convert_optab_handler): ...here.
5325 (widening_optab_handler): Delete.
5326 (find_widening_optab_handler): Remove permit_non_widening parameter.
5327 (find_widening_optab_handler_and_mode): Likewise. Provide an
5328 override that operates on mode class wrappers.
5329 * optabs-query.c (widening_optab_handler): Delete.
5330 (find_widening_optab_handler_and_mode): Remove permit_non_widening
5331 parameter. Assert that the two modes are the same class and that
5332 the "from" mode is narrower than the "to" mode. Use
5333 convert_optab_handler instead of widening_optab_handler.
5334 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
5335 instead of widening_optab_handler.
5336 * expr.c (expand_expr_real_2): Update calls to
5337 find_widening_optab_handler.
5338 * optabs.c (expand_widen_pattern_expr): Likewise.
5339 (expand_binop_directly): Take the insn_code as a parameter.
5340 (expand_binop): Only call find_widening_optab_handler for
5341 conversion optabs; use optab_handler otherwise. Update calls
5342 to find_widening_optab_handler and expand_binop_directly.
5343 Use convert_optab_handler instead of widening_optab_handler.
5344 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
5345 find_widening_optab_handler and use scalar_mode rather than
5347 (convert_plusminus_to_widen): Likewise.
5349 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5350 Alan Hayward <alan.hayward@arm.com>
5351 David Sherwood <david.sherwood@arm.com>
5353 * machmode.h (fixed_size_mode): New class.
5354 * rtl.h (get_pool_mode): Return fixed_size_mode.
5355 * gengtype.c (main): Add fixed_size_mode.
5356 * target.def (get_raw_result_mode): Return a fixed_size_mode.
5357 (get_raw_arg_mode): Likewise.
5358 * doc/tm.texi: Regenerate.
5359 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
5360 * targhooks.c (default_get_reg_raw_mode): Likewise.
5361 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
5362 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
5363 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
5364 (msp430_get_raw_result_mode): Likewise.
5365 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
5366 * dbxout.c (dbxout_parms): Require fixed-size modes.
5367 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
5368 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5369 * omp-low.c (lower_oacc_reductions): Likewise.
5370 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
5371 (simplify_subreg): Update accordingly.
5372 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
5373 (force_const_mem): Update accordingly. Return NULL_RTX for modes
5374 that aren't fixed-size.
5375 (get_pool_mode): Return a fixed_size_mode.
5376 (output_constant_pool_2): Take a fixed_size_mode.
5378 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5379 Alan Hayward <alan.hayward@arm.com>
5380 David Sherwood <david.sherwood@arm.com>
5382 * doc/rtl.texi (vec_series): Document.
5383 (const): Say that the operand can be a vec_series.
5384 * rtl.def (VEC_SERIES): New rtx code.
5385 * rtl.h (const_vec_series_p_1): Declare.
5386 (const_vec_series_p): New function.
5387 * emit-rtl.h (gen_const_vec_series): Declare.
5388 (gen_vec_series): Likewise.
5389 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
5390 (gen_vec_series): Likewise.
5391 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
5392 * simplify-rtx.c (simplify_unary_operation): Handle negations
5394 (simplify_binary_operation_series): New function.
5395 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
5396 (test_vector_ops_series): New function.
5397 (test_vector_ops): Call it.
5398 * config/powerpcspe/altivec.md (altivec_lvsl): Use
5399 gen_const_vec_series.
5400 (altivec_lvsr): Likewise.
5401 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): 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 * doc/rtl.texi (const): Update description of address constants.
5408 Say that vector constants are allowed too.
5409 * common.md (E, F): Use CONSTANT_P instead of checking for
5411 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
5412 checking for CONST_VECTOR.
5413 * expmed.c (make_tree): Use build_vector_from_val for a CONST
5415 * expr.c (expand_expr_real_2): Check for vector modes instead
5416 of checking for CONST_VECTOR.
5417 * rtl.h (const_vec_p): New function.
5418 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
5419 (unwrap_const_vec_duplicate): Handle them here too.
5421 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5422 David Malcolm <dmalcolm@redhat.com>
5423 Alan Hayward <alan.hayward@arm.com>
5424 David Sherwood <david.sherwood@arm.com>
5426 * rtl.h (vec_duplicate_p): New function.
5427 * selftest-rtl.c (assert_rtx_eq_at): New function.
5428 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
5429 (assert_rtx_eq_at): Declare.
5430 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
5431 * selftest-run-tests.c (selftest::run_tests): Call it.
5432 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
5433 (simplify_unary_operation_1): Recursively handle vector duplicates.
5434 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
5436 (simplify_subreg): Handle subregs of vector duplicates.
5437 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
5438 (selftest::simplify_rtx_c_tests): New functions.
5440 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5441 Alan Hayward <alan.hayward@arm.com>
5442 David Sherwood <david.sherwood@arm.com>
5444 * emit-rtl.h (gen_const_vec_duplicate): Declare.
5445 (gen_vec_duplicate): Likewise.
5446 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
5448 (gen_const_vector): ...here.
5449 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
5450 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
5451 whose elements are all equal.
5452 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
5453 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
5454 (simplify_relational_operation): Likewise.
5455 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
5457 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
5458 (aarch64_expand_vector_init): Likewise.
5459 * config/arm/arm.c (neon_vdup_constant): Likewise.
5460 (neon_expand_vector_init): Likewise.
5461 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
5462 (arm_block_set_unaligned_vect): Likewise.
5463 (arm_block_set_aligned_vect): Likewise.
5464 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
5465 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
5466 (expand_vec_perm_even_odd_pack): Likewise.
5467 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
5468 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
5469 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
5470 gen_const_vec_duplicate.
5471 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
5472 * config/mips/mips.c (mips_gen_const_int_vector): Use
5473 gen_const_vec_duplicate.
5474 (mips_expand_vector_init): Use CONST0_RTX.
5475 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
5476 (define_split): Use gen_const_vec_duplicate.
5477 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
5478 (define_split): Use gen_const_vec_duplicate.
5479 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
5480 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
5481 * config/spu/spu.c (spu_const): Likewise.
5483 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5484 Alan Hayward <alan.hayward@arm.com>
5485 David Sherwood <david.sherwood@arm.com>
5487 * combine.c (can_change_dest_mode): Reject changes in
5488 REGMODE_NATURAL_SIZE.
5490 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
5492 * configure.ac (--enable-libssp): New.
5493 (gcc_cv_libc_provides_ssp): Check for explicit setting before
5494 trying to determine target-specific default. Adjust indentation.
5495 * configure: Regenerated.
5496 * doc/install.texi (Configuration): Expand --disable-libssp
5499 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
5501 config/i386/i386.c (ix86_expand_epilogue): Correct stack
5504 2017-10-31 Martin Jambor <mjambor@suse.cz>
5507 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
5509 2017-10-31 David Malcolm <dmalcolm@redhat.com>
5511 * auto-profile.c (autofdo_source_profile::read): Use
5512 UNKNOWN_LOCATION rather than 0.
5513 * diagnostic-core.h (warning_at_rich_loc): Rename to...
5514 (warning_at): ...this overload.
5515 (warning_at_rich_loc_n): Rename to...
5516 (warning_n): ...this overload.
5517 (error_at_rich_loc): Rename to...
5518 (error_at): ...this overload.
5519 (pedwarn_at_rich_loc): Rename to...
5520 (pedwarn): ...this overload.
5521 (permerror_at_rich_loc): Rename to...
5522 (permerror): ...this overload.
5523 (inform_at_rich_loc): Rename to...
5524 (inform): ...this overload.
5525 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
5526 (diagnostic_n_impl_richloc): Rename to...
5527 (diagnostic_n_impl): ...this rich_location *-based decl.
5528 (inform_at_rich_loc): Rename to...
5529 (inform): ...this, and add an assertion.
5530 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
5531 (warning_at_rich_loc): Rename to...
5532 (warning_at): ...this, and add an assertion.
5533 (warning_at_rich_loc_n): Rename to...
5534 (warning_n): ...this, and add an assertion.
5535 (warning_n): Update location_t-based implementation for removal of
5536 location_t-based diagnostic_n_impl.
5537 (pedwarn_at_rich_loc): Rename to...
5538 (pedwarn): ...this, and add an assertion.
5539 (permerror_at_rich_loc): Rename to...
5540 (permerror): ...this, and add an assertion.
5541 (error_n): Update for removal of location_t-based diagnostic_n_impl.
5542 (error_at_rich_loc): Rename to...
5543 (error_at): ...this, and add an assertion.
5544 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
5545 (driver::do_spec_on_infiles): Likewise.
5546 * substring-locations.c (format_warning_va): Update for renaming
5547 of inform_at_rich_loc.
5549 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
5551 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
5552 _Float<N>X built-in functions so that the variant without the
5553 "__builtin_" prefix is only enabled for the GNU C and Objective C
5554 languages when they are in non-strict ANSI/ISO mode.
5555 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
5556 * target.def (floatn_builtin_p): Add a target hook to control
5557 whether _Float<N> and _Float<N>X built-in functions without the
5558 "__builtin_" prefix are enabled, and return true for C and
5559 Objective C in the default hook. Include langhooks.h in
5561 * targhooks.h (default_floatn_builtin_p): Likewise.
5562 * targhooks.c (default_floatn_builtin_p): Likewise.
5563 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
5564 floatn_builtin_p target hook.
5565 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
5567 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
5568 Eric Botcazou <ebotcazou@adacore.com>
5570 PR rtl-optimization/81803
5571 * lra-constraints.c (curr_insn_transform): Also reload the whole
5572 register for a strict subreg no wider than a word if this is for
5573 a WORD_REGISTER_OPERATIONS target.
5575 2017-10-31 Jason Merrill <jason@redhat.com>
5577 * gdbinit.in: Skip over inlines from timevar.h.
5579 2017-10-31 Martin Liska <mliska@suse.cz>
5581 * doc/gcov.texi: Document new option.
5582 * gcov.c (print_usage): Likewise print it.
5583 (process_args): Support the argument.
5584 (format_count): New function.
5585 (format_gcov): Use the function.
5587 2017-10-31 Martin Liska <mliska@suse.cz>
5589 * gcov.c (struct name_map): do not use typedef.
5590 Define operator== and operator<.
5591 (name_search): Remove.
5592 (name_sort): Remove.
5593 (main): Do not allocate names.
5594 (process_file): Add vertical space.
5595 (generate_results): Use std::find.
5596 (release_structures): Do not release memory.
5597 (find_source): Use std::find.
5599 2017-10-31 Martin Liska <mliska@suse.cz>
5601 * gcov.c (struct line_info): Remove it's typedef.
5602 (line_info::line_info): Add proper ctor.
5603 (line_info::has_block): Do not use a typedef.
5604 (struct source_info): Do not use typedef.
5605 (circuit): Likewise.
5606 (get_cycles_count): Likewise.
5607 (output_intermediate_file): Iterate via vector iterator.
5608 (add_line_counts): Use std::vector methods.
5609 (accumulate_line_counts): Likewise.
5610 (output_lines): Likewise.
5612 2017-10-31 Martin Liska <mliska@suse.cz>
5614 * gcov.c (struct source_info): Remove typedef.
5615 (source_info::source_info): Add proper ctor.
5616 (accumulate_line_counts): Use struct, not it's typedef.
5617 (output_gcov_file): Likewise.
5618 (output_lines): Likewise.
5619 (main): Do not allocate an array.
5620 (output_intermediate_file): Use size of vector container.
5621 (process_file): Resize the vector.
5622 (generate_results): Do not preallocate, use newly added vector
5624 (release_structures): Do not release sources.
5625 (find_source): Use vector methods.
5626 (add_line_counts): Do not use typedef.
5628 2017-10-31 Martin Liska <mliska@suse.cz>
5630 * doc/gcov.texi: Document that.
5631 * gcov.c (add_line_counts): Mark lines with a non-executed
5633 (output_line_beginning): Handle such lines.
5634 (output_lines): Pass new argument.
5635 (output_intermediate_file): Print it in intermediate format.
5637 2017-10-31 Martin Liska <mliska@suse.cz>
5639 * color-macros.h: New file.
5640 * diagnostic-color.c: Factor out color related to macros to
5642 * doc/gcov.texi: Document -k option.
5643 * gcov.c (INCLUDE_STRING): Include string.h.
5644 (print_usage): Add -k option.
5645 (process_args): Parse it.
5646 (pad_count_string): New function.
5647 (output_line_beginning): Likewise.
5648 (DEFAULT_LINE_START): New macro.
5649 (output_lines): Support color output.
5651 2017-10-31 Martin Liska <mliska@suse.cz>
5653 PR gcov-profile/82633
5654 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
5655 their interaction with GCOV infrastructure.
5656 * configure.ac: Add -fkeep-{inline,static}-functions to
5658 * configure: Regenerate.
5660 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
5663 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
5665 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
5668 * config/rs6000/rs6000.md (allocate_stack): Force update interval
5669 into a register if it does not fit into an immediate offset field.
5671 2017-10-31 Olivier Hainque <hainque@adacore.com>
5673 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
5675 2017-10-31 Julia Koval <julia.koval@intel.com>
5677 * config.gcc: Add gfniintrin.h.
5678 * config/i386/gfniintrin.h: New.
5679 * config/i386/i386-builtin-types.def
5680 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5681 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5682 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5683 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5684 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5685 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
5686 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
5687 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
5688 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
5689 V64QI_FTYPE_V64QI_V64QI_INT): New types.
5690 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5691 * config/i386/immintrin.h: Include gfniintrin.h.
5692 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
5694 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
5696 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
5698 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5700 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
5701 (arm_ashldi3_1bit): Remove pattern.
5702 (ashrdi3): Remove shift by 1 expansion.
5703 (arm_ashrdi3_1bit): Remove pattern.
5704 (lshrdi3): Remove shift by 1 expansion.
5705 (arm_lshrdi3_1bit): Remove pattern.
5706 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
5707 cost of ashldi3 by 1.
5708 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
5709 (<shift>di3_neon): Likewise.
5711 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5713 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
5714 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
5715 and (*aarch64_simd_mov<VQ:mode>).
5716 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
5717 pattern alternative.
5718 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
5719 attributes to match pattern alternative.
5721 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
5723 * config.gcc (powerpc*-*-*): Add emmintrin.h.
5724 * config/rs6000/emmintrin.h: New file.
5725 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
5727 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5729 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
5730 * (movdi_vfp_cortexa8): Remove pattern.
5732 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5734 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
5736 (Specific, alpha*-dec-osf5.1): Remove.
5737 (Specific, mips-sgi-irix5): Remove.
5738 (Specific, mips-sgi-irix6): Remove.
5740 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5743 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
5744 arguments to clear_bit_region_be.
5746 2017-10-30 Jim Wilson <wilson@tuliptree.org>
5748 * gimplify.c: Include memmodel.h.
5750 2017-10-30 Martin Jambor <mjambor@suse.cz>
5752 * omp-grid.c (grid_attempt_target_gridification): Also insert a
5753 condition whether loop should be executed at all.
5755 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
5757 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5758 gimple folding of vec_madd() intrinsics.
5759 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
5760 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
5761 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
5763 2017-10-30 Richard Biener <rguenther@suse.de>
5765 PR tree-optimization/82762
5767 2017-10-23 Richard Biener <rguenther@suse.de>
5769 PR tree-optimization/82129
5771 2017-08-01 Richard Biener <rguenther@suse.de>
5773 PR tree-optimization/81181
5774 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5775 (compute_antic): ... end of iteration here.
5777 2017-10-30 Joseph Myers <joseph@codesourcery.com>
5779 * doc/invoke.texi (C Dialect Options): Document -std=c17,
5780 -std=iso9899:2017 and -std=gnu17.
5781 * doc/standards.texi (C Language): Document C17 support.
5782 * doc/cpp.texi (Overview): Mention -std=c17.
5783 (Standard Predefined Macros): Document C11 and C17 values of
5784 __STDC_VERSION__. Do not refer to C99 support as incomplete.
5785 * doc/extend.texi (Inline): Do not list individual options for
5786 standards newer than C99.
5787 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
5789 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
5792 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
5794 * asan.c (asan_finish_file): Align asan globals array by shadow
5797 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5800 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
5801 (struct store_immediate_info): Add bitregion_start and bitregion_end
5803 (store_immediate_info::store_immediate_info): Add brs and bre
5804 arguments and initialize bitregion_{start,end} from those.
5805 (struct merged_store_group): Add bitregion_start, bitregion_end,
5806 align_base and mask fields. Drop unnecessary struct keyword from
5807 struct store_immediate_info. Add do_merge method.
5808 (clear_bit_region_be): Use memset instead of loop storing zeros.
5809 (merged_store_group::do_merge): New method.
5810 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
5811 stores as long as the surrounding bitregions have no gaps.
5812 (merged_store_group::merge_overlapping): Use do_merge.
5813 (merged_store_group::apply_stores): Test that bitregion_{start,end}
5814 is byte aligned, rather than requiring that start and width are
5815 byte aligned. Drop unnecessary struct keyword from
5816 struct store_immediate_info. Allocate and populate also mask array.
5817 Make start of the arrays relative to bitregion_start rather than
5818 start and size them according to bitregion_{end,start} difference.
5819 (struct imm_store_chain_info): Drop unnecessary struct keyword from
5820 struct store_immediate_info.
5821 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
5822 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
5823 struct keyword from struct store_immediate_info.
5824 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
5825 between stores as long as the surrounding bitregions have no gaps.
5827 (struct split_store): Add orig non-static data member.
5828 (split_store::split_store): Initialize orig to false.
5829 (find_constituent_stmts): Return store_immediate_info *, non-NULL
5830 if there is exactly a single original stmt. Change stmts argument
5831 to pointer from reference, if NULL, don't push anything to it. Add
5832 first argument, use it to optimize skipping over orig stmts that
5833 are known to be before bitpos already. Simplify.
5834 (split_group): Return unsigned int count how many stores are or
5835 would be needed rather than a bool. Add allow_unaligned argument.
5836 Change split_stores argument from reference to pointer, if NULL,
5837 only do a dry run computing how many stores would be produced.
5838 Rewritten algorithm to use both alignment and misalign if
5839 !allow_unaligned and handle bitfield stores with gaps.
5840 (imm_store_chain_info::output_merged_store): Set start_byte_pos
5841 from bitregion_start instead of start. Compute allow_unaligned
5842 here, if true, do 2 split_group dry runs to compute which one
5843 produces fewer stores and prefer aligned if equal. Punt if
5844 new count is bigger or equal than original before emitting any
5845 statements, rather than during that. Remove no longer needed
5846 new_ssa_names tracking. Replace num_stmts with
5847 split_stores.length (). Use 32-bit stack allocated entries
5848 in split_stores auto_vec. Try to reuse original store lhs/rhs1
5849 if possible. Handle bitfields with gaps.
5850 (pass_store_merging::execute): Ignore bitsize == 0 stores.
5851 Compute bitregion_{start,end} for the stores and construct
5852 store_immediate_info with that. Formatting fixes.
5854 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
5857 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
5858 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
5860 2017-10-29 Jim Wilson <wilson@tuliptree.org>
5862 * gimplify.c: Include tm_p.h.
5864 * common.opt (gcoff): Re-add as ignored option.
5865 (gcoff1, gcoff2, gcoff3): Likewise.
5867 * Makefile.in (OBJS): Delete sdbout.o.
5868 (GTFILES): Delete $(srcdir)/sdbout.c.
5869 * debug.h: Delete sdb_debug_hooks.
5870 * final.c: Delete sdbout.h include.
5871 (final_scan_insn): Delete SDB_DEBUG check.
5872 (rest_of_clean_state): Likewise.
5873 * output.h: Delete sdb_begin_function_line.
5876 * toplev.c: Delete sdbout.h include.
5877 (process_options): Delete SDB_DEBUG check.
5878 * tree-core.h (tree_type_common): Delete pointer field of
5880 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
5881 TYPE_SYMTAB_POINTER.
5882 * tree.h (TYPE_SYMTAB_POINTER): Delete.
5883 (TYPE_SYMTAB_IS_POINTER): Delete.
5884 (TYPE_SYMTAB_IS_DIE): Renumber.
5885 * xcoffout.c: Refer to former sdbout.c file.
5886 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
5888 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
5889 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
5890 (Debugging Options): Delete -gcoff.
5891 (-gstabs): Delete SDB reference.
5893 (-gcoff@var{level}): Delete.
5894 * doc/passes.texi (Debugging information output): Delete SDB and
5895 sdbout.c references.
5896 * doc/tm.texi: Regenerate.
5897 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
5898 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
5900 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
5901 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
5902 (SDB_DEBUGGING_INFO): Delete.
5903 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
5904 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
5905 * target.def (output_source_filename): Delete COFF reference.
5907 * common.opt (gcoff): Delete.
5908 (gxcoff+): Update Negative chain.
5909 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
5911 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
5912 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
5913 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
5915 (expand_function_start): Change sdb reference to past tense.
5916 (expand_function_end): Change sdb reference to past tense.
5917 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
5918 * opts.c (debug_type_names): Delete coff entry.
5919 (common_handle_option): Delete OPT_gcoff case.
5920 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
5922 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
5923 * config/cris/cris.h: Delete SDB reference in comment.
5924 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
5925 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
5926 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
5927 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
5929 (ix86_expand_prologue): Likewise.
5930 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
5931 * config/ia64/ia64.h: Likewise.
5932 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
5933 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
5935 * config/mmix/mmix.h: Likewise.
5936 * config/nds32/nds32.c: Likewise.
5937 * config/stormy/storym16.h: Likewise.
5938 * config/visium/visium.h: Likewise.
5939 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
5941 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
5943 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
5944 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
5945 FRAME_POINTER_REGNUM point at high end of local var area.
5947 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
5949 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
5950 Move comment around. Do not reset best_edge for a copiable
5951 destination if the copy would cause a partition change.
5952 (better_edge_p): Remove redundant check.
5954 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5956 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
5958 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5960 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
5961 for math functions that have _Float<N> and _Float<N>X variants.
5962 (mathfn_built_in_2): Add support for math functions that have
5963 _Float<N> and _Float<N>X variants.
5964 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
5965 (expand_builtin_mathfn_ternary): Add support for fma with
5966 _Float<N> and _Float<N>X variants.
5967 (expand_builtin): Likewise.
5968 (fold_builtin_3): Likewise.
5969 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
5970 create math function _Float<N> and _Float<N>X variants as external
5972 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
5973 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
5974 the __builtin_ prefix and if not strict ansi, without the prefix.
5975 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
5976 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
5977 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
5978 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
5979 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
5980 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
5981 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
5982 function signatures for fma _Float<N> and _Float<N>X variants.
5983 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
5984 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
5985 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
5986 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
5987 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
5988 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
5989 * gencfn-macros.c (print_case_cfn): Add support for math functions
5990 that have _Float<N> and _Float<N>X variants.
5991 (print_define_operator_list): Likewise.
5992 (fltfn_suffixes): Likewise.
5994 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
5995 for math functions that have _Float<N> and _Float<N>X variants.
5996 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
5997 and _Float<N>X variants.
5998 (COPYSIGN): Likewise.
6001 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6002 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6004 (integer_valued_read_call_p): Likewise.
6005 * fold-const-call.c (fold_const_call_ss): Likewise.
6006 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6007 _Float<N> and _Float<N>X variants.
6008 (fold_const_call_ssss): Add support for fma _Float<N> and
6009 _Float<N>X variants.
6010 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6011 support for copysign and fma _Float<N> and _Float<N>X variants.
6012 (backprop::process_builtin_call_use): Likewise.
6013 * tree-call-cdce.c (can_test_argument_range); Add support for
6014 sqrt _Float<N> and _Float<N>X variants.
6015 (edom_only_function): Likewise.
6016 (get_no_error_domain): Likewise.
6017 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6018 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6019 copysign _Float<N> and _Float<N>X variants.
6020 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6021 handled by machine independent code.
6022 (FMAF128): Likewise.
6023 * doc/cpp.texi (Common Predefined Macros): Document defining
6024 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6025 fma _Float<N> and _Float<N>X variants.
6027 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6030 * config/i386/i386-modes.def (CCFPU): Remove definition.
6031 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6032 (ix86_cc_modes_compatible): Ditto.
6033 (ix86_expand_carry_flag_compare): Ditto.
6034 (ix86_expand_int_movcc): Ditto.
6035 (ix86_expand_int_addcc): Ditto.
6036 (ix86_reverse_condition): Ditto.
6037 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6038 Return true/false for unordered/ordered fp comparisons.
6039 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6040 (ix86_prepare_fp_compare_args): Update for rename.
6041 (ix86_expand_fp_compare): Update for rename. Generate unordered
6042 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6043 (ix86_expand_sse_compare_and_jump): Ditto.
6044 * config/i386/predicates.md (fcmov_comparison_operator):
6045 Remove CCFPU mode handling.
6046 (ix86_comparison_operator): Ditto.
6047 (ix86_carry_flag_operator): Ditto.
6048 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6049 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6050 (*cmpu<mode>_cc_i387): Ditto.
6051 (FPCMP): Remove mode iterator.
6052 (unord): Remove mode attribute.
6053 (unord_subst): New define_subst transformation
6054 (unord): New define_subst attribute.
6056 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6057 (*cmpi<unord>xf_i387): Ditto.
6058 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6059 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
6060 using unord_subst transformation.
6061 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
6062 (round_saeonly): Also handle CCFP mode.
6063 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
6064 Remove UNSPEC_SAHF unspec handling.
6066 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6068 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
6070 2017-10-27 Jeff Law <law@redhat.com>
6072 * gimple-ssa-sprintf.c: Include domwalk.h.
6073 (class sprintf_dom_walker): New class, derived from dom_walker.
6074 (sprintf_dom_walker::before_dom_children): New function.
6075 (struct call_info): Moved into sprintf_dom_walker class
6076 (compute_formath_length, handle_gimple_call): Likewise.
6077 (sprintf_length::execute): Call the dominator walker rather
6078 than walking the statements.
6080 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
6081 gimple statement locations.
6082 (check_array_bounds): Corresponding changes. Get the statement's
6083 location directly from wi->stmt.
6085 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
6088 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
6090 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6092 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
6093 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
6095 2017-10-27 Jakub Jelinek <jakub@redhat.com>
6098 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
6099 * config/i386/i386.c (maybe_get_pool_constant): Removed.
6100 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
6101 maybe_get_pool_constant.
6102 * config/i386/predicates.md (zero_extended_scalar_load_operand):
6105 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6107 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
6108 2.26 caveat. Update gas and gld versions.
6109 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
6112 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6114 * cgraph.h (set_malloc_flag): Declare.
6115 * cgraph.c (set_malloc_flag_1): New function.
6116 (set_malloc_flag): Likewise.
6117 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
6118 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
6120 (read_ipa_call_summary): Add support for reading is_return_callee.
6121 (write_ipa_call_summary): Stream is_return_callee.
6122 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
6123 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
6124 ipa-prop.h, ipa-fnsummary.h.
6125 (pure_const_names): Change to static.
6126 (malloc_state_e): Define.
6127 (malloc_state_names): Define.
6128 (funct_state_d): Add field malloc_state.
6129 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
6130 (check_retval_uses): New function.
6131 (malloc_candidate_p): Likewise.
6132 (analyze_function): Add support for malloc attribute.
6133 (pure_const_write_summary): Stream malloc_state.
6134 (pure_const_read_summary): Add support for reading malloc_state.
6135 (dump_malloc_lattice): New function.
6136 (propagate_malloc): New function.
6137 (warn_function_malloc): New function.
6138 (ipa_pure_const::execute): Call propagate_malloc and
6139 ipa_free_fn_summary.
6140 (pass_local_pure_const::execute): Add support for malloc attribute.
6141 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
6142 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
6144 2017-10-27 Martin Liska <mliska@suse.cz>
6146 PR gcov-profile/82457
6147 * doc/invoke.texi: Document that one needs a non-strict ISO mode
6148 for fork-like functions to be properly instrumented.
6150 2017-10-27 Richard Biener <rguenther@suse.de>
6153 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
6154 info when we redirected EH.
6156 2017-10-26 Michael Collison <michael.collison@arm.com>
6158 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
6160 (<optab>_trunchf<GPI:mode>2: New pattern.
6161 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
6162 * config/aarch64/iterators.md (wv): New mode attribute.
6163 (vf, VF): New mode attributes.
6164 (vgp, VGP): New mode attributes.
6165 (s): Update attribute with SImode and DImode prefixes.
6167 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6169 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
6170 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
6171 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
6172 (nios2_option_overide): Initialize it. Don't allow R0-relative
6173 addressing with PIC.
6174 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
6175 (nios2_symbolic_constant_p): Likewise.
6176 (nios2_legitimate_address_p): Likewise.
6177 (nios2_r0rel_section_name_p): New.
6178 (nios2_symbol_ref_in_r0rel_data_p): New.
6179 (nios2_emit_move_sequence): Handle r0rel_constant_p.
6180 (r0rel_constant_p): New.
6181 (nios2_print_operand_address): Handle r0rel_constant_p.
6182 (nios2_cdx_narrow_form_p): Likewise.
6183 * config/nios2/nios2.opt (mr0rel-sec=): New option.
6184 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
6185 (Nios II Options): Document -mr0rel-sec.
6187 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6189 * config/nios2/nios2.c: Include xregex.h.
6190 (nios2_gprel_sec_regex): New.
6191 (nios2_option_overide): Initialize it. Don't allow GP-relative
6192 addressing with PIC.
6193 (nios2_small_section_name_p): Check for regex match.
6194 * config/nios2/nios2.opt (mgprel-sec=): New option.
6195 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
6196 (Nios II Options): Document -mgprel-sec.
6198 2017-10-26 Jim Wilson <wilson@tuliptree.org>
6200 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
6202 2017-10-26 Tom de Vries <tom@codesourcery.com>
6204 PR tree-optimization/82707
6205 * gimple.c (gimple_copy): Fix unsharing of
6206 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
6208 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
6210 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
6211 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
6212 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
6213 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
6214 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
6215 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
6216 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
6217 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
6218 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
6219 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
6220 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
6221 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
6222 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
6223 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
6224 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
6225 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
6226 _mm512_mask_cmpunord_ps_mask): New intrinsics.
6228 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6230 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
6232 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
6233 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
6234 warning to rs6000.c. Remove the Undocumented flag, since it has
6236 (-mabi=ibmlongdouble): Likewise.
6237 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
6238 already set, set the default format for long double.
6239 (rs6000_debug_reg_global): Print whether long double is IBM or
6241 (rs6000_option_override_internal): Rework setting long double
6242 format. Only warn if the user is changing the long double default
6243 and they did not use -Wno-psabi.
6244 * doc/invoke.texi (PowerPC options): Update the documentation for
6245 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
6247 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6248 Alan Hayward <alan.hayward@arm.com>
6249 David Sherwood <david.sherwood@arm.com>
6251 * rtl.h (wider_subreg_mode): New function.
6252 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
6253 rather than an unsigned int *.
6254 * ira-color.c (regno_max_ref_width): Replace with...
6255 (regno_max_ref_mode): ...this new variable.
6256 (coalesced_pseudo_reg_slot_compare): Update accordingly.
6257 Use wider_subreg_mode.
6258 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
6259 rather than an unsigned int *.
6260 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
6261 (process_alt_operands): Likewise.
6262 (invariant_p): Likewise.
6263 * lra-spills.c (assign_mem_slot): Likewise.
6264 (add_pseudo_to_slot): Likewise.
6265 * lra.c (collect_non_operand_hard_regs): Likewise.
6266 (add_regs_to_insn_regno_info): Likewise.
6267 * reload1.c (regno_max_ref_width): Replace with...
6268 (regno_max_ref_mode): ...this new variable.
6269 (reload): Update accordingly. Update call to
6270 ira_sort_regnos_for_alter_reg.
6271 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
6272 (init_eliminable_invariants): Update to use regno_max_ref_mode.
6273 (scan_paradoxical_subregs): Likewise.
6275 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6277 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
6278 (aarch64_frame): Add emit_frame_chain boolean.
6279 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6280 Move eh_return case to aarch64_layout_frame.
6281 (aarch64_layout_frame): Initialize emit_frame_chain.
6282 (aarch64_expand_prologue): Use emit_frame_chain.
6284 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6286 * config/aarch64/aarch64.c (aarch64_layout_frame):
6287 Ensure LR is always stored at the bottom of the callee-saves.
6288 Remove rarely used frame layout which saves callee-saves at top of
6289 frame, so the store of LR can be used as a valid probe in all cases.
6291 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6293 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
6294 Improve unaligned TImode/TFmode base/offset split.
6296 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6297 Alan Hayward <alan.hayward@arm.com>
6298 David Sherwood <david.sherwood@arm.com>
6300 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
6301 * combine.c (find_single_use_1): Likewise.
6302 (expand_field_assignment): Likewise.
6303 (move_deaths): Likewise.
6304 * lra-constraints.c (simplify_operand_subreg): Likewise.
6305 (curr_insn_transform): Likewise.
6306 * lra.c (collect_non_operand_hard_regs): Likewise.
6307 (add_regs_to_insn_regno_info): Likewise.
6308 * rtlanal.c (reg_referenced_p): Likewise.
6309 (covers_regno_no_parallel_p): Likewise.
6311 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6313 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
6314 Don't print any bits outside the precision of the value.
6315 * wide-int.cc (test_printing): Add some new tests.
6317 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6319 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
6320 supports -xbrace_comment option.
6321 * configure: Regenerate.
6322 * config.in: Regenerate.
6323 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
6324 (ASM_CPU_SPEC): Use it.
6326 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6328 * target.def (static_rtx_alignment): New hook.
6329 * targhooks.h (default_static_rtx_alignment): Declare.
6330 * targhooks.c (default_static_rtx_alignment): New function.
6331 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
6332 * doc/tm.texi: Regenerate.
6333 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
6334 instead of targetm.constant_alignment. Remove call to
6336 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6337 (cris_preferred_mininum_alignment): New function, split out from...
6338 (cris_constant_alignment): ...here.
6339 (cris_static_rtx_alignment): New function.
6340 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
6342 (ix86_constant_alignment): ...here.
6343 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6344 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6345 (mmix_static_rtx_alignment): New function.
6346 * config/spu/spu.c (spu_static_rtx_alignment): New function.
6347 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6349 2017-10-26 Tamar Christina <tamar.christina@arm.com>
6352 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
6353 Add flag_trapping_math and flag_fp_int_builtin_inexact.
6355 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
6357 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
6358 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
6359 mark as a sign-extending load.
6360 (local_pic_load_u): Define.
6362 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
6365 * fold-const.c (operand_equal_for_comparison_p): Also return true
6366 if ARG0 is a simple variant of ARG1 with narrower precision.
6367 (fold_ternary_loc): Always pass unstripped operands to the predicate.
6369 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
6371 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
6373 * i386.h (processor_costs): Add gather_static, gather_per_elt,
6374 scatter_static, scatter_per_elt.
6375 * x86-tune-costs.h: Add new cost entries.
6377 2017-10-25 Richard Biener <rguenther@suse.de>
6379 * tree-ssa-sccvn.h (vn_eliminate): Declare.
6380 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
6381 class pass_fre): Move to ...
6382 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
6383 class pass_fre): ... here and adjust for statistics.
6385 2017-10-25 Jakub Jelinek <jakub@redhat.com>
6388 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
6389 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
6390 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
6391 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
6394 2017-10-25 Richard Biener <rguenther@suse.de>
6396 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
6397 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
6398 eliminate_push_avail, eliminate_insert): Move inside...
6399 (class eliminate_dom_walker): ... this class in preparation
6401 (fini_eliminate): Remove by merging with ...
6402 (eliminate): ... this function. Adjust for class changes.
6403 (pass_pre::execute): Remove fini_eliminate call.
6404 (pass_fre::execute): Likewise.
6406 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6409 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
6410 (VPERMI2, VPERMI2I): New mode iterators.
6411 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
6412 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
6414 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
6415 mode iterator. Remove 3 old define_insn patterns.
6416 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
6417 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
6418 VPERMI2 mode iterator, remove the other two expanders.
6419 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
6420 to use VPERMI2 mode iterator, add another alternative for vpermi2*
6421 instructions, remove the other two patterns.
6422 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
6423 mode iterator, remove the other two patterns.
6424 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
6425 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
6426 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
6427 and adjust argument order accordingly.
6428 (ix86_expand_vec_perm): Adjust caller.
6429 (expand_vec_perm_1): Likewise.
6430 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
6431 (expand_vec_perm_vpermt2_vpshub2): ... this.
6432 (ix86_expand_vec_perm_const_1): Adjust caller.
6433 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
6436 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
6437 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
6438 (vec_shl_<mode>): Remove unused expander.
6439 (avx512bw_<shift_insn><mode>3): New define_insn.
6440 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
6441 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
6443 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
6446 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
6449 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6451 PR rtl-optimization/82396
6452 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
6453 (autopref_multipass_init): Simplify initialization.
6454 (autopref_rank_data): Simplify sort order.
6455 * gcc/sched-int.h (autopref_multipass_data_): Remove
6456 multi_mem_insn_p, min_offset and max_offset.
6458 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6461 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6462 Check special value of flag_omit_frame_pointer.
6463 (aarch64_can_eliminate): Likewise.
6464 (aarch64_override_options_after_change_1): Simplify handling of
6465 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
6467 2017-10-24 Richard Biener <rguenther@suse.de>
6469 PR tree-optimization/82697
6470 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
6471 zero for conditional load and unconditional store.
6473 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6475 * doc/install.texi: Document bootstrap-cet.
6477 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6480 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
6481 ENDBR instruction at function entrance if function is only
6484 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6487 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
6488 patterns to better describe from which operation the CF is computed.
6489 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
6490 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
6491 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
6492 is 0, use _0 suffixed expanders instead of emitting a comparison
6495 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
6497 * config/i386/i386.md(*movsf_internal, *movdf_internal):
6498 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
6500 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
6503 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
6504 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
6505 * loop-iv.c (iv_get_reaching_def): Likewise.
6506 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
6507 variable is promoted and the partition contains undefined values.
6509 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6511 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
6513 (nios2_address_cost): Define.
6514 (nios2_legitimize_address): Recognize (exp + constant) directly.
6515 (TARGET_ADDRESS_COST): Define.
6517 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6519 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
6520 (nios2_symbolic_memory_operand_p): Declare.
6521 (nios2_split_large_constant): Declare.
6522 (nios2_split_symbolic_memory_operand): Declare.
6523 * config/nios2/nios2.c: Adjust includes.
6524 (nios2_symbolic_constant_allowed): New.
6525 (nios2_symbolic_constant_p): New.
6526 (nios2_plus_symbolic_constant_p): New.
6527 (nios2_valid_addr_expr_p): Recognize addresses involving
6529 (nios2_legitimate_address_p): Likewise, also LO_SUM.
6530 (nios2_symbolic_memory_operand_p): New.
6531 (nios2_large_constant_p): New.
6532 (nios2_split_large_constant): New.
6533 (nios2_split_plus_large_constant): New.
6534 (nios2_split_symbolic_memory_operand): New.
6535 (nios2_legitimize_address): Code refactoring. Handle addresses
6536 involving symbolic constants.
6537 (nios2_emit_move_sequence): Likewise.
6538 (nios2_print_operand): Improve error output.
6539 (nios2_print_operand_address): Handle LO_SUM.
6540 (nios2_cdx_narrow_form_p): Likewise.
6541 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
6542 operands involving symbolic constants.
6543 (movhi_internal, movsi_internal): Likewise.
6544 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
6545 (extendhisi2, extendqi<mode>2): Likewise.
6547 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6549 * tree-pass.h (PROP_rtl_split_insns): Define.
6550 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
6552 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6554 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
6556 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6559 * target.def (const_not_ok_for_debug_p): Default to
6560 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
6561 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
6562 * targhooks.c (default_const_not_ok_for_debug_p): New function.
6563 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
6564 which targetm.const_not_ok_for_debug_p returned true.
6565 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
6567 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
6569 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
6570 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
6571 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
6573 (ix86_const_not_ok_for_debug_p): New function.
6574 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
6575 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
6577 2017-10-23 David Malcolm <dmalcolm@redhat.com>
6580 * system.h: Conditionally include "unique-ptr.h" if
6581 INCLUDE_UNIQUE_PTR is defined.
6582 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
6583 of defining INCLUDE_UNIQUE_PTR before including "system.h".
6585 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
6587 * config/rl78/rl78.md: New define_expand "subdi3".
6589 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
6592 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
6593 DF_REF_INSN if DF_REF_INSN_INFO is false.
6595 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
6597 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
6598 xmm_move instead of sse_move.
6599 (sse_store_index): New function.
6600 (ix86_register_move_cost): Be more sensible about mismatch stall;
6601 model AVX moves correctly; make difference between sse->integer and
6603 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
6604 moves; make difference between SSE and AVX.
6605 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
6606 and zmm_move. Increase size of sse load and store tables;
6607 add unaligned load and store tables; add ssemmx_to_integer.
6608 * x86-tune-costs.h: Update all entries according to real
6609 move latencies from Agner Fog's manual and chip documentation.
6611 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6614 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
6615 * config/i386/constraints.md (Wf): New constraint.
6616 * config/i386/i386.md (UNSPEC_SBB): New unspec.
6617 (cmp<dwi>_doubleword): Removed.
6618 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
6619 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
6620 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
6621 expand with cmp<dwi>_doubleword. For LTU and GEU use
6622 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
6624 * common.opt (gcolumn-info): Enable by default.
6625 * doc/invoke.texi (gcolumn-info): Document new default.
6627 2017-10-23 Richard Biener <rguenther@suse.de>
6629 PR tree-optimization/82672
6630 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
6631 Fold the stmt if we propagated into it.
6633 2017-10-23 Richard Biener <rguenther@suse.de>
6635 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
6636 (bitmap_remove_expr_from_set): ... this. All callers call this
6637 for non-constant values.
6638 (bitmap_set_subtract): Rename to...
6639 (bitmap_set_subtract_expressions): ... this. Adjust and
6641 (bitmap_set_contains_value): Remove superfluous check.
6642 (bitmap_set_replace_value): Inline into single caller ...
6643 (bitmap_value_replace_in_set): ... here and simplify.
6644 (dependent_clean): Merge into ...
6645 (clean): ... this using an overload. Adjust.
6646 (prune_clobbered_mems): Adjust.
6647 (compute_antic_aux): Likewise.
6648 (compute_partial_antic_aux): Likewise.
6650 2017-10-23 Richard Biener <rguenther@suse.de>
6652 PR tree-optimization/82129
6654 2017-08-01 Richard Biener <rguenther@suse.de>
6656 PR tree-optimization/81181
6657 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6658 (compute_antic): ... end of iteration here.
6660 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6662 * target.def (starting_frame_offset): New hook.
6663 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
6664 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
6665 * doc/tm.texi.in: Regenerate.
6666 * hooks.h (hook_hwi_void_0): Declare.
6667 * hooks.c (hook_hwi_void_0): New function.
6668 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
6669 STARTING_FRAME_OFFSET.
6670 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
6671 * reload1.c (reload): Likewise.
6672 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
6673 instead of STARTING_FRAME_OFFSET.
6674 * function.c (try_fit_stack_local): Likewise.
6675 (assign_stack_local_1): Likewise
6676 (instantiate_virtual_regs): Likewise.
6677 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
6678 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
6679 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
6680 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
6681 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
6682 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
6683 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
6684 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
6685 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
6686 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
6687 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
6688 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
6689 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
6690 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
6691 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
6692 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
6693 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
6694 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
6695 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
6696 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
6697 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
6698 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
6699 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
6700 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
6701 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
6702 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
6703 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
6704 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
6705 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
6706 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
6707 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
6708 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
6709 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
6710 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
6711 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
6712 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
6713 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
6714 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
6715 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
6716 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
6717 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
6718 * config/avr/avr.c (avr_starting_frame_offset): Make static and
6719 return a HOST_WIDE_INT.
6720 (avr_builtin_setjmp_frame_value): Use it instead of
6721 STARTING_FRAME_OFFSET.
6722 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6723 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
6724 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
6726 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6727 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
6728 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
6729 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6730 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
6731 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
6732 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6733 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
6734 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
6735 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6736 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
6737 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
6739 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6740 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
6741 * config/mips/mips.c (mips_compute_frame_info): Refer to
6742 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6743 (mips_starting_frame_offset): New function.
6744 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6745 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
6746 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
6747 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
6748 and return a HOST_WIDE_INT.
6749 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6750 (mmix_initial_elimination_offset): Refer to
6751 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6752 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
6753 * config/pa/pa.c (pa_starting_frame_offset): New function.
6754 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
6755 (pa_expand_prologue): Likewise.
6756 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6757 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
6758 !FRAME_GROWS_DOWNWARD handling to...
6759 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6760 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
6761 !FRAME_GROWS_DOWNWARD handling to...
6762 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6763 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
6764 !FRAME_GROWS_DOWNWARD handling to...
6765 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6766 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
6768 (rs6000_starting_frame_offset): New function.
6769 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
6770 !FRAME_GROWS_DOWNWARD handling to...
6771 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6772 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
6773 !FRAME_GROWS_DOWNWARD handling to...
6774 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6775 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
6776 !FRAME_GROWS_DOWNWARD handling to...
6777 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6778 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
6779 (rs6000_starting_frame_offset): New function.
6780 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
6781 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
6782 * config/vax/vax.c (vax_starting_frame_offset): New function.
6783 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
6784 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6785 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
6786 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
6787 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6788 * system.h (STARTING_FRAME_OFFSET): Poison.
6790 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6792 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
6793 SCALAR_TYPE_MODE instead of TYPE_MODE.
6795 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6796 Alan Hayward <alan.hayward@arm.com>
6797 David Sherwood <david.sherwood@arm.com>
6799 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
6801 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6802 Alan Hayward <alan.hayward@arm.com>
6803 David Sherwood <david.sherwood@arm.com>
6805 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
6807 2017-10-23 Richard Biener <rguenther@suse.de>
6809 PR tree-optimization/82129
6810 * tree-ssa-pre.c (bitmap_set_and): Remove.
6811 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
6812 canonicalizing expressions in the set to those with lowest
6813 ID rather than taking that from the first edge.
6815 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6817 * combine.c (rtx_equal_for_field_assignment_p): Use
6818 byte_lowpart_offset.
6820 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6821 Alan Hayward <alan.hayward@arm.com>
6822 David Sherwood <david.sherwood@arm.com>
6824 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
6825 to a SUBREG_PROMOTED_VAR.
6827 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6828 Alan Hayward <alan.hayward@arm.com>
6829 David Sherwood <david.sherwood@arm.com>
6831 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
6832 (expand_debug_source_expr): Likewise.
6833 * combine.c (combine_simplify_rtx): Likewise.
6834 * cse.c (fold_rtx): Likewise.
6835 * optabs.c (expand_float): Likewise.
6836 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6837 (simplify_binary_operation_1): Likewise.
6839 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6840 Alan Hayward <alan.hayward@arm.com>
6841 David Sherwood <david.sherwood@arm.com>
6843 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
6844 (record_promoted_value): Likewise.
6845 * expr.c (expand_expr_real_2): Likewise.
6846 * ree.c (update_reg_equal_equiv_notes): Likewise.
6847 (combine_set_extension): Likewise.
6848 * rtlanal.c (low_bitmask_len): Likewise.
6849 * simplify-rtx.c (neg_const_int): Likewise.
6850 (simplify_binary_operation_1): Likewise.
6852 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6853 Alan Hayward <alan.hayward@arm.com>
6854 David Sherwood <david.sherwood@arm.com>
6856 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
6857 * regcprop.c (maybe_mode_change): Likewise.
6858 * reload1.c (alter_reg): Likewise.
6860 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6862 * inchash.h (inchash::hash::add_wide_int): New function.
6863 * lto-streamer-out.c (hash_tree): Use it.
6865 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6867 * inchash.h (inchash::hash::add_wide_int): Rename to...
6868 (inchash::hash::add_hwi): ...this.
6869 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
6870 (polymorphic_call_target_hasher::hash): Likewise.
6871 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
6872 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
6873 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
6874 * lto-streamer-out.c (hash_tree): Likewise.
6875 * optc-save-gen.awk: Likewise.
6876 * tree.c (add_expr): Likewise.
6878 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6881 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
6882 for ordered inequality comparisons even with TARGET_IEEE_FP.
6884 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6887 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
6888 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6889 Expand with cmp<dwi>_doubleword.
6891 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6893 * extend.texi: Add x86 specific to 'nocf_check' attribute.
6894 List CET intrinsics.
6895 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
6896 specific to -fcf-protection option.
6898 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6900 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
6901 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
6902 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
6903 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
6904 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
6905 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
6906 (extra_objs): Add cet.o for Linux/x86 targets.
6907 (tmake_file): Add i386/t-cet for Linux/x86 targets.
6908 * config/i386/cet.c: New file.
6909 * config/i386/cetintrin.h: Likewise.
6910 * config/i386/t-cet: Likewise.
6911 * config/i386/cpuid.h (bit_SHSTK): New.
6912 (bit_IBT): Likewise.
6913 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
6914 pass IBT and SHSTK bits.
6915 * config/i386/i386-builtin-types.def
6916 (VOID_FTYPE_UNSIGNED_PVOID): New.
6917 (VOID_FTYPE_UINT64_PVOID): Likewise.
6918 * config/i386/i386-builtin.def: Add CET intrinsics.
6919 * config/i386/i386-c.c (ix86_target_macros_internal): Add
6920 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
6921 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
6922 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
6924 * config/i386/i386.c (rest_of_insert_endbranch): New.
6925 (pass_data_insert_endbranch): Likewise.
6926 (pass_insert_endbranch): Likewise.
6927 (make_pass_insert_endbranch): Likewise.
6928 (ix86_notrack_prefixed_insn_p): Likewise.
6929 (ix86_target_string): Add -mibt, -mshstk flags.
6930 (ix86_option_override_internal): Add flag_cf_protection
6932 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
6933 (ix86_print_operand): Add 'notrack' prefix output.
6934 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
6935 (ix86_expand_builtin): Expand CET intrinsics.
6936 (x86_output_mi_thunk): Add 'endbranch' instruction.
6937 * config/i386/i386.h (TARGET_IBT): New.
6938 (TARGET_IBT_P): Likewise.
6939 (TARGET_SHSTK): Likewise.
6940 (TARGET_SHSTK_P): Likewise.
6941 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
6942 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
6943 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
6944 (builtin_setjmp_setup): New pattern.
6945 (builtin_longjmp): Likewise.
6946 (rdssp<mode>): Likewise.
6947 (incssp<mode>): Likewise.
6948 (saveprevssp): Likewise.
6949 (rstorssp): Likewise.
6950 (wrss<mode>): Likewise.
6951 (wruss<mode>): Likewise.
6952 (setssbsy): Likewise.
6953 (clrssbsy): Likewise.
6954 (nop_endbr): Likewise.
6955 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
6957 * config/i386/immintrin.h: Include <cetintrin.h>.
6958 * config/i386/linux-common.h
6959 (file_end_indicate_exec_stack_and_cet): New prototype.
6960 (TARGET_ASM_FILE_END): New.
6962 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
6964 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
6965 latencies instead of having separate table; make difference between
6966 integer and float costs.
6967 * i386.h (processor_costs): Remove scalar_stmt_cost,
6968 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
6969 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
6971 * x86-tune-costs.h: Remove entries which has been removed in
6972 procesor_costs from all tables; make cond_taken_branch_cost
6973 and cond_not_taken_branch_cost COST_N_INSNS based.
6975 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
6977 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
6979 2017-10-20 Jakub Jelinek <jakub@redhat.com>
6981 * config/i386/i386.md (isa): Remove fma_avx512f.
6982 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
6983 <avx512>_fmadd_<mode>_mask3<round_name>,
6984 <avx512>_fmsub_<mode>_mask<round_name>,
6985 <avx512>_fmsub_<mode>_mask3<round_name>,
6986 <avx512>_fnmadd_<mode>_mask<round_name>,
6987 <avx512>_fnmadd_<mode>_mask3<round_name>,
6988 <avx512>_fnmsub_<mode>_mask<round_name>,
6989 <avx512>_fnmsub_<mode>_mask3<round_name>,
6990 <avx512>_fmaddsub_<mode>_mask<round_name>,
6991 <avx512>_fmaddsub_<mode>_mask3<round_name>,
6992 <avx512>_fmsubadd_<mode>_mask<round_name>,
6993 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
6994 (*vec_widen_umult_even_v16si<mask_name>,
6995 *vec_widen_smult_even_v16si<mask_name>): Likewise.
6996 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
6998 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7000 * extend.texi: Add 'nocf_check' documentation.
7001 * gimple.texi: Add second parameter to
7002 gimple_build_call_from_tree.
7003 * invoke.texi: Add -fcf-protection documentation.
7004 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7006 2017-10-20 Richard Biener <rguenther@suse.de>
7008 PR tree-optimization/82473
7009 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7010 the largest input type.
7012 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7014 * c-attribs.c (handle_nocf_check_attribute): New function.
7015 (c_common_attribute_table): Add 'nocf_check' handling.
7016 * gimple-parser.c: Add second argument NULL to
7017 gimple_build_call_from_tree.
7018 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7019 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7021 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7022 * common.opt: Add fcf-protection flag.
7023 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7024 * flag-types.h: Add enum cf_protection_level.
7025 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7026 Add 'nocf_check' attribute propagation to gimple call.
7027 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7028 (gimple_build_call_from_tree): Update prototype.
7029 (gimple_call_nocf_check_p): New function.
7030 (gimple_call_set_nocf_check): Likewise.
7031 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7032 * ipa-icf.c: Add nocf_check attribute in statement hash.
7033 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7034 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7035 * toplev.c (process_options): Add flag_cf_protection handling.
7037 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7039 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7041 2017-10-20 Richard Biener <rguenther@suse.de>
7043 PR tree-optimization/82603
7044 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7045 remove false predicated stores.
7047 2017-10-20 Richard Biener <rguenther@suse.de>
7049 * graphite-isl-ast-to-gimple.c
7050 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7051 Remove return value and simplify, dump copied stmt after lhs
7053 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7054 Reduce dump verbosity.
7055 (gsi_insert_earliest): Likewise.
7056 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7057 * graphite.c (print_global_statistics): Adjust dumping.
7058 (print_graphite_scop_statistics): Likewise.
7059 (print_graphite_statistics): Do not dump loops here.
7060 (graphite_transform_loops): But here.
7062 2017-10-20 Nicolas Roche <roche@adacore.com>
7064 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
7065 * configure: Regenerate.
7067 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7070 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
7071 functions when optimizing replace GIMPLE_RETURN stmts with
7072 calls to __builtin_unreachable ().
7075 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
7076 for -fsanitize=thread link of executables.
7077 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
7078 link of executables.
7081 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
7083 (<shift_insn><mode>3<mask_name>): Change the last of the 3
7084 define_insns for logical vector shifts to use VI248_AVX512BW
7085 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
7086 condition, useless isa and prefix attributes. Change the first
7087 2 of these define_insns to ...
7088 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
7089 define_insn for avx512vl.
7090 (<shift_insn><mode>3): ... and this, new define_insn without
7091 masking for non-avx512vl.
7094 * config/i386/sse.md (*andnot<mode>3,
7095 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
7096 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
7097 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
7098 not applied use empty suffix even for TARGET_AVX512VL.
7099 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
7100 is applied, supply evex,evex or evex,evex,evex instead of just
7103 2017-10-20 Julia Koval <julia.koval@intel.com>
7105 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
7106 (OPTION_MASK_ISA_GFNI_UNSET): New.
7107 (ix86_handle_option): Handle OPT_mgfni.
7108 * config/i386/cpuid.h (bit_GFNI): New.
7109 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
7110 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
7111 * config/i386/i386.c (ix86_target_string): Add -mgfni.
7112 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
7113 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
7114 * config/i386/i386.opt: Add mgfni.
7116 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
7118 * config/msp430/msp430.c (msp430_option_override): Disable
7119 -fdelete-null-pointer-checks.
7120 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
7122 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7124 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
7125 of x87 and SSE instructions.
7127 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7129 * asan.c (create_cond_insert_point): Do not update edge count.
7130 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
7131 (afdo_propagate_circuit): Likewise.
7132 (afdo_calculate_branch_prob): Likewise.
7133 (afdo_annotate_cfg): Likewise.
7134 * basic-block.h (struct edge_def): Remove count.
7135 (edge_def::count): New accessor.
7136 * bb-reorder.c (rotate_loop): Update.
7137 (find_traces_1_round): Update.
7138 (connect_traces): Update.
7139 (sanitize_hot_paths): Update.
7140 * cfg.c (unchecked_make_edge): Update.
7141 (make_single_succ_edge): Update.
7142 (check_bb_profile): Update.
7143 (dump_edge_info): Update.
7144 (update_bb_profile_for_threading): Update.
7145 (scale_bbs_frequencies_int): Update.
7146 (scale_bbs_frequencies_gcov_type): Update.
7147 (scale_bbs_frequencies_profile_count): Update.
7148 (scale_bbs_frequencies): Update.
7149 * cfganal.c (connect_infinite_loops_to_exit): Update.
7150 * cfgbuild.c (compute_outgoing_frequencies): Update.
7151 (find_many_sub_basic_blocks): Update.
7152 * cfgcleanup.c (try_forward_edges): Update.
7153 (try_crossjump_to_edge): Update
7154 * cfgexpand.c (expand_gimple_cond): Update
7155 (expand_gimple_tailcall): Update
7156 (construct_exit_block): Update
7157 * cfghooks.c (verify_flow_info): Update
7158 (redirect_edge_succ_nodup): Update
7159 (split_edge): Update
7160 (make_forwarder_block): Update
7161 (duplicate_block): Update
7162 (account_profile_record): Update
7163 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
7164 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
7165 * cfgloopmanip.c (scale_loop_profile): Update.
7167 (lv_adjust_loop_entry_edge): Update.
7168 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7169 (force_nonfallthru_and_redirect): Update.
7170 (purge_dead_edges): Update.
7171 (rtl_flow_call_edges_add): Update.
7172 * cgraphunit.c (init_lowered_empty_function): Update.
7173 (cgraph_node::expand_thunk): Update.
7174 * gimple-pretty-print.c (dump_probability): Update.
7175 (dump_edge_probability): Update.
7176 * gimple-ssa-isolate-paths.c (isolate_path): Update.
7177 * haifa-sched.c (sched_create_recovery_edges): Update.
7178 * hsa-gen.c (convert_switch_statements): Update.
7179 * ifcvt.c (dead_or_predicable): Update.
7180 * ipa-inline-transform.c (inline_transform): Update.
7181 * ipa-split.c (split_function): Update.
7182 * ipa-utils.c (ipa_merge_profiles): Update.
7183 * loop-doloop.c (add_test): Update.
7184 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
7185 * lto-streamer-in.c (input_cfg): Update.
7186 (input_function): Update.
7187 * lto-streamer-out.c (output_cfg): Update.
7188 * modulo-sched.c (sms_schedule): Update.
7189 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
7190 * predict.c (maybe_hot_edge_p): Update.
7191 (unlikely_executed_edge_p): Update.
7192 (probably_never_executed_edge_p): Update.
7193 (dump_prediction): Update.
7194 (drop_profile): Update.
7195 (propagate_unlikely_bbs_forward): Update.
7196 (determine_unlikely_bbs): Update.
7197 (force_edge_cold): Update.
7198 * profile.c (compute_branch_probabilities): Update.
7199 * reg-stack.c (better_edge): Update.
7200 * shrink-wrap.c (handle_simple_exit): Update.
7201 * tracer.c (better_p): Update.
7202 * trans-mem.c (expand_transaction): Update.
7203 (split_bb_make_tm_edge): Update.
7204 * tree-call-cdce.c: Update.
7205 * tree-cfg.c (gimple_find_sub_bbs): Update.
7206 (gimple_split_edge): Update.
7207 (gimple_duplicate_sese_region): Update.
7208 (gimple_duplicate_sese_tail): Update.
7209 (gimple_flow_call_edges_add): Update.
7210 (insert_cond_bb): Update.
7211 (execute_fixup_cfg): Update.
7212 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
7213 * tree-complex.c (expand_complex_div_wide): Update.
7214 * tree-eh.c (lower_resx): Update.
7215 (unsplit_eh): Update.
7216 (cleanup_empty_eh_move_lp): Update.
7217 * tree-inline.c (copy_edges_for_bb): Update.
7218 (freqs_to_counts): Update.
7219 (copy_cfg_body): Update.
7220 * tree-ssa-dce.c (remove_dead_stmt): Update.
7221 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7222 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7223 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7224 (unloop_loops): Update.
7225 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7226 * tree-ssa-loop-split.c (connect_loops): Update.
7227 (split_loop): Update.
7228 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
7229 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7230 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7231 * tree-ssa-reassoc.c (branch_fixup): Update.
7232 * tree-ssa-tail-merge.c (replace_block_by): Update.
7233 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7234 (compute_path_counts): Update.
7235 (update_profile): Update.
7236 (recompute_probabilities): Update.
7237 (update_joiner_offpath_counts): Update.
7238 (estimated_freqs_path): Update.
7239 (freqs_to_counts_path): Update.
7240 (clear_counts_path): Update.
7241 (ssa_fix_duplicate_block_edges): Update.
7242 (duplicate_thread_path): Update.
7243 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7244 (case_bit_test_cmp): Update.
7245 (collect_switch_conv_info): Update.
7246 (gen_inbound_check): Update.
7247 (do_jump_if_equal): Update.
7248 (emit_cmp_and_jump_insns): Update.
7249 * tree-tailcall.c (decrease_profile): Update.
7250 (eliminate_tail_call): Update.
7251 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7252 (vect_do_peeling): Update.
7253 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7254 * ubsan.c (ubsan_expand_null_ifn): Update.
7255 (ubsan_expand_ptr_ifn): Update.
7256 * value-prof.c (gimple_divmod_fixed_value): Update.
7257 (gimple_mod_pow2): Update.
7258 (gimple_mod_subtract): Update.
7259 (gimple_ic): Update.
7260 (gimple_stringop_fixed_value): Update.
7262 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7265 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
7267 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
7269 PR rtl-optimization/82395
7270 * ira-color.c (allocno_priority_compare_func): Fix comparison step
7271 based on non_spilled_static_chain_regno_p.
7273 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7275 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
7276 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
7277 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
7279 2017-10-19 Martin Sebor <msebor@redhat.com>
7281 PR tree-optimization/82596
7282 * tree.c (array_at_struct_end_p): Handle STRING_CST.
7284 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7286 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
7287 (get_mem_refs_of_builtin_call): Likewise.
7288 * builtins.c (expand_builtin_apply): Adjust call to
7289 allocate_dynamic_stack_space.
7290 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
7291 the third argument to allocate_dynamic_stack_space, otherwise -1.
7292 (expand_builtin): Deal with all alloca variants.
7293 (is_inexpensive_builtin): Likewise.
7294 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
7295 * calls.c (special_function_p): Deal with all alloca variants.
7296 (initialize_argument_information): Adjust call to
7297 allocate_dynamic_stack_space.
7298 (expand_call): Likewise.
7299 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
7300 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
7301 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
7302 use it for the stack usage computation.
7303 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
7304 * function.c (gimplify_parameters): Call build_alloca_call_expr.
7305 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
7306 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
7307 (in_loop_p): Remove first argument and useless check.
7308 (pass_walloca::execute): Remove useless test and adjust call to above.
7309 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
7310 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
7311 (gimplify_call_expr): Deal with all alloca variants.
7312 * hsa-gen.c (gen_hsa_alloca): Likewise.
7313 (gen_hsa_insns_for_call): Likewise.
7314 * ipa-pure-const.c (special_builtin_state): Likewise.
7315 * tree-chkp.c (chkp_build_returned_bound): Likewise.
7316 * tree-object-size.c (alloc_object_size): Likewise.
7317 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
7318 (call_may_clobber_ref_p_1): Likewise.
7319 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
7320 (ccp_fold_stmt): Likewise.
7321 (optimize_stack_restore): Likewise.
7322 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
7323 (mark_all_reaching_defs_necessary_1): Likewise.
7324 (propagate_necessity): Likewise.
7325 (eliminate_unnecessary_stmts): Likewise.
7326 * tree.c (build_common_builtin_nodes): Build
7327 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
7328 (build_alloca_call_expr): New function.
7329 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
7330 (CASE_BUILT_IN_ALLOCA): Likewise.
7331 (build_alloca_call_expr): Declare.
7332 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
7334 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7337 * dwarf2out.c (new_die_raw): New static inline function.
7338 (new_die): Use it to create the DIE.
7339 (add_AT_external_die_ref): Likewise.
7340 (clone_die): Likewise.
7341 (clone_as_declaration): Likewise.
7342 (dwarf2out_vms_debug_main_pointer): Likewise.
7343 (base_type_die): Likewise. Remove early return for corner cases.
7344 Do not call add_pubtype on the DIE here.
7345 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
7346 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
7347 typedefs for base types with DW_AT_endianity. Make sure a DIE with
7348 native order exists for base types, attach the DIE manually and call
7349 add_pubtype on it. Do not equate a reverse order DIE to the type.
7351 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
7353 * config/arm/arm.c (align_ok_ldrd_strd): New function.
7354 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
7356 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
7358 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7360 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
7361 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
7362 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
7363 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
7364 * opts.c (sanitizer_opts): Add builtin.
7365 * ubsan.c (instrument_builtin): New function.
7366 (pass_ubsan::execute): Call it.
7367 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
7368 * doc/invoke.texi: Document -fsanitize=builtin.
7370 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
7371 builtins, store max (log2 (align), 0) into uchar field instead of
7372 align into uptr field.
7373 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
7374 store uchar 0 field instead of uptr 0 field.
7375 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
7376 instead of passing one address of struct with 2 locations pass
7377 two addresses of structs with 1 location each.
7378 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
7379 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
7380 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
7381 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
7382 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
7383 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
7384 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
7385 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
7387 2017-10-19 Martin Liska <mliska@suse.cz>
7390 * file-find.c (remove_prefix): Remove.
7391 * file-find.h (remove_prefix): Likewise.
7392 * gcc-ar.c: Remove smartness of lookup.
7394 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
7396 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
7397 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
7398 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
7400 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7403 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
7404 (setcc + and to xor + setcc): New peephole2.
7406 2017-10-19 Tom de Vries <tom@codesourcery.com>
7408 * doc/sourcebuild.texi (Test Directives, Variants of
7409 dg-require-support): Add dg-require-stack-size.
7411 2017-10-19 Martin Liska <mliska@suse.cz>
7414 * gimplify.c (gimplify_decl_expr): Do not instrument variables
7415 that have a large alignment.
7416 (gimplify_target_expr): Likewise.
7418 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
7420 PR rtl-optimization/82602
7421 * ira.c (rtx_moveable_p): Return false for volatile asm.
7423 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
7426 * config/i386/i386-modes.def (CCGZ): New CC mode.
7427 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
7428 * config/i386/predicates.md (ix86_comparison_operator):
7430 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7431 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
7432 with double-word subtraction.
7433 (put_condition_code): Handle CCGZmode.
7435 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
7437 * wide-int.cc (debug (const wide_int &)): New.
7438 (debug (const wide_int *)): New.
7439 (debug (const widest_int &)): New.
7440 (debug (const widest_int *)): New.
7442 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
7445 * lra-constraints.c (curr_insn_transform): Use non-input operand
7446 instead of output one for matched reload.
7448 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7450 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
7451 (tree-ssa-loop-ivopts.h): New header file.
7452 (struct builtin_info): New fields.
7453 (classify_builtin_1): Compute and record base and offset parts for
7454 memset builtin partition by calling strip_offset.
7455 (offset_cmp, fuse_memset_builtins): New functions.
7456 (finalize_partitions): Fuse adjacent memset partitions by calling
7458 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
7459 Expose the interface.
7460 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
7462 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7464 PR tree-optimization/82574
7465 * tree-loop-distribution.c (find_single_drs): New parameter. Check
7466 that data reference must be executed exactly once per iteration
7467 against the outermost loop in nest.
7468 (classify_partition): Update call to above function.
7470 2017-10-18 Richard Biener <rguenther@suse.de>
7472 PR tree-optimization/82591
7473 * graphite.c (graphite_transform_loops): Move code gen message
7475 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7476 Here. Handle scop_to_isl_ast failing.
7477 (scop_to_isl_ast): Limit the number of ISL operations.
7479 2017-10-18 Richard Biener <rguenther@suse.de>
7481 * graphite-isl-ast-to-gimple.c
7482 (translate_isl_ast_to_gimple::set_rename): Simplify.
7483 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
7484 (graphite_copy_stmts_from_block): ... here.
7485 (copy_bb_and_scalar_dependences): Simplify.
7486 (add_parameters_to_ivs_params): Canonicalize.
7487 (generate_entry_out_of_ssa_copies): Simplify.
7488 * graphite-sese-to-poly.c (extract_affine_name): Simplify
7489 by passing in ISL dimension.
7490 (parameter_index_in_region_1): Rename to ...
7491 (parameter_index_in_region): ... this.
7492 (extract_affine): Adjust assert, pass down parameter index.
7493 (add_param_constraints): Use range-info when available.
7494 (build_scop_context): Adjust.
7495 * sese.c (new_sese_info): Adjust.
7496 (free_sese_info): Likewise.
7497 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
7498 Remove unused typedefs.
7499 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
7501 2017-10-18 Martin Liska <mliska@suse.cz>
7503 * combine.c (simplify_compare_const): Add gcc_fallthrough.
7505 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7507 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
7508 (s390_sched_init): Do not reset s390_sched_state if we entered the
7509 current basic block via a fallthru edge and all others are unlikely.
7511 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7513 * config/s390/s390.c (NUM_SIDES): New variable.
7514 (LONGRUNNING_THRESHOLD): New variable.
7515 (LATENCY_FACTOR): New variable.
7516 (s390_sched_score): Decrease score for long-running instructions on
7518 (s390_sched_variable_issue): Perform bookkeeping for long-running
7521 2017-10-18 Richard Biener <rguenther@suse.de>
7523 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
7524 Simplify with removal of the parameter rename map.
7525 (set_rename): Likewise.
7526 (should_copy_to_new_region): Likewise.
7527 (graphite_copy_stmts_from_block): Likewise.
7528 (copy_bb_and_scalar_dependences): Remove initialization of
7529 unused copied_bb_map.
7531 (copy_internal_parameters): Likewise.
7532 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
7533 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
7534 Use INTEGRAL_TYPE_P.
7535 (parameter_index_in_region_1): Rename to ...
7536 (assign_parameter_index_in_region): ... this. Assert we have
7537 a parameter we handle.
7538 (scan_tree_for_params): Adjust.
7539 * sese.h (parameter_rename_map_t): Remove.
7540 (struct sese_info_t): Remove unused parameter_rename_map and
7541 copied_bb_map members.
7542 * sese.c (new_sese_info): Adjust.
7543 (free_sese_info): Likewise.
7545 2017-10-18 Martin Liska <mliska@suse.cz>
7548 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
7549 on an abnormal edge.
7551 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
7553 * doc/invoke.texi (ffunction-sections and fdata-sections):
7556 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7558 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
7559 the use statement can throw internally.
7561 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7563 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
7564 any RTX present on the RHS of a SET.
7565 * compare-elim.c (try_eliminate_compare): Restore comment.
7567 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7569 * langhooks.h (struct lang_hooks): Document that tree_size langhook
7570 may be also called on tcc_type nodes.
7571 * langhooks.c (lhd_tree_size): Likewise.
7573 2017-10-17 David Malcolm <dmalcolm@redhat.com>
7575 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
7576 format_warning_at_substring.
7577 (maybe_warn): Convert source_range * param to a location_t. Pass
7578 UNKNOWN_LOCATION rather than NULL to fmtwarn.
7579 (format_directive): Remove code to extract source_ranges and
7580 source_range * in favor of just a location_t.
7581 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
7583 * substring-locations.c (format_warning_va): Convert
7584 source_range * param to a location_t.
7585 (format_warning_at_substring): Likewise.
7586 * substring-locations.h (format_warning_va): Likewise.
7587 (format_warning_at_substring): Likewise.
7589 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
7591 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
7593 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
7594 and scatter/gather ops.
7596 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
7597 vec_gather_load and vec_scatter_store.
7598 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
7599 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
7600 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
7601 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
7602 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
7603 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
7605 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
7607 * reg-stack.c (compare_for_stack_reg): Add bool argument.
7608 Detect FTST instruction and handle its register pops. Only pop
7609 second operand if can_pop_second_op is true.
7610 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
7611 set can_pop_second_op to false in the compare_for_stack_reg call.
7613 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
7614 output_fp_compare for stack register operands.
7615 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
7616 instructions here. Do not emit stack register pops here. Assert
7617 that FCOMPP pops next to top stack register. Rewrite function.
7619 2017-10-17 Nathan Sidwell <nathan@acm.org>
7622 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
7623 use DECL_ASSEMBLER_NAME_RAW.
7626 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
7629 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
7630 Wilco Dijkstra <wilco.dijkstra@arm.com>
7632 * builtins.c (expand_builtin_update_setjmp_buf): Add a
7633 converstion to Pmode from the buf_addr.
7635 2017-10-17 Richard Biener <rguenther@suse.de>
7637 * graphite-dependences.c (scop_get_reads_and_writes): Change
7638 output parameters to references.
7640 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
7642 PR 71026/tree-optimization
7643 * fold-const.c (distribute_real_division): Removed.
7644 (fold_binary_loc): Remove calls to distribute_real_divison.
7646 2017-10-17 Richard Biener <rguenther@suse.de>
7648 * graphite-scop-detection.c
7649 (scop_detection::stmt_has_simple_data_refs_p): Always use
7650 the full nest as region.
7651 (try_generate_gimple_bb): Likewise.
7652 * sese.c (scalar_evolution_in_region): Simplify now that
7653 SCEV can handle instantiation in regions.
7654 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
7655 in the non-loop part of a function if requested.
7657 2017-10-17 Richard Biener <rguenther@suse.de>
7659 PR tree-optimization/82563
7660 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
7662 (graphite_regenerate_ast_isl): Call it.
7663 * graphite-scop-detection.c (build_scops): Remove entry edge split.
7665 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7667 PR tree-optimization/82549
7668 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
7669 Formatting fixes. Instead of calling make_bit_field_ref with negative
7672 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
7674 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
7675 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
7676 _mm_maskz_reduce_ss): New.
7677 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
7678 __builtin_ia32_reducess_mask): Ditto..
7679 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
7680 * config/i386/sse.md (reduces<mode>): Renamed to ...
7681 (reduces<mode><mask_scalar_name>): ... this.
7682 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
7684 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
7685 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
7687 2017-10-16 David Malcolm <dmalcolm@redhat.com>
7689 * Makefile.in (OBJS): Add unique-ptr-tests.o.
7690 * selftest-run-tests.c (selftest::run_tests): Call
7691 selftest::unique_ptr_tests_cc_tests.
7692 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
7693 * unique-ptr-tests.cc: New file.
7695 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
7698 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7700 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7701 (make_hard_regno_born, make_hard_regno_dead): Update
7702 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
7704 2017-10-16 Jeff Law <law@redhat.com>
7706 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7708 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
7710 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
7712 2017-10-16 Olivier Hainque <hainque@adacore.com>
7714 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
7715 with_cpu if we were configured for an e500v2 target cpu name.
7717 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
7719 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
7720 * doc/invoke.texi: Document +nodsp as a valid extension for
7723 2017-10-16 Martin Liska <mliska@suse.cz>
7725 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
7726 (test_set_range): Likewise.
7727 (test_range_functions): Rename to ...
7728 (test_bit_in_range): ... this.
7729 (sbitmap_c_tests): Add new test.
7731 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7733 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
7735 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
7736 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
7738 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7740 * config/aarch64/aarch64-builtins.c
7741 (aarch64_types_quadopu_lane_qualifiers): New.
7742 (TYPES_QUADOPU_LANE): New.
7743 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
7744 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
7745 (aarch64_<sur>dot_laneq<vsi2qi>): New.
7746 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
7747 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
7748 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
7749 (Vdottype, DOTPROD): New.
7750 (sur): Add SDOT and UDOT.
7752 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7754 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
7755 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
7756 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
7758 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
7759 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
7760 Enable TARGET_DOTPROD.
7761 (cortex-a75.cortex-a55): Likewise.
7762 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
7764 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7766 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
7767 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
7769 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
7771 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
7772 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
7773 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
7774 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
7775 * config/arm/types.md (neon_dot, neon_dot_q): New.
7776 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
7778 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7780 * config/arm/arm.h (TARGET_DOTPROD): New.
7781 * config/arm/arm.c (arm_arch_dotprod): New.
7782 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
7783 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
7784 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
7785 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
7786 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
7787 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
7788 * doc/invoke.texi (armv8.2-a): Document dotprod
7790 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7792 * i386.c (ix86_vec_cost): New function.
7793 (ix86_rtx_costs): Handle vector operations better.
7794 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
7795 * x86-tune-costs.h: Add new costs to all tables.
7797 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7799 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
7801 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
7802 divsd, sqrtss and sqrtsd
7803 * x86-tune-costs.h: Add new entries to all costs.
7804 (znver1_cost): Fix to match real instruction latencies.
7806 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7807 Michael Collison <michael.collison@arm.com>
7809 * compare-elim.c: Include emit-rtl.h.
7810 (can_merge_compare_into_arith): New function.
7811 (try_validate_parallel): Likewise.
7812 (try_merge_compare): Likewise.
7813 (try_eliminate_compare): Call the above when no previous clobber
7815 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
7818 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7822 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
7823 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
7827 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
7828 to be any operand_equal_p operands. For & (B - 1) require
7829 B to be power of 2. Recognize
7830 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
7832 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7835 * optabs.c (expand_memory_blockage): Fix call of
7836 targetm.have_memory_blockage.
7838 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7841 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
7842 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
7843 objects to extra_objs instead of overwriting it.
7845 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7847 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
7848 Use any_fp_register_operand as operand[3] predicate. Simplify
7849 equality test for operands[2] and operands[4] memory location.
7850 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
7851 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
7852 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
7853 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
7854 any_fp_register_operand as operand[1] predicate. Simplify
7855 equality test for operands[0] and operands[3] memory location.
7856 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
7857 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
7858 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
7860 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7862 * target-insns.def: Add memory_blockage.
7863 * optabs.c (expand_memory_blockage): New function.
7864 (expand_asm_memory_barrier): Rename ...
7865 (expand_asm_memory_blockage): ... to this.
7866 (expand_mem_thread_fence): Call expand_memory_blockage
7867 instead of expand_asm_memory_barrier.
7868 (expand_mem_singnal_fence): Ditto.
7869 (expand_atomic_load): Ditto.
7870 (expand_atomic_store): Ditto.
7871 * doc/md.texi (Standard Pattern Names For Generation):
7872 Document memory_blockage instruction pattern.
7874 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
7876 * config/rl78/rl78.c (rl78_emit_libcall): New function.
7877 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
7878 * config/rl78/rl78.md: New define_expand "adddi3".
7880 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7882 * cfghooks.c (verify_flow_info): Disable check that all probabilities
7885 2017-10-13 Jeff Law <law@redhat.com>
7887 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
7889 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7892 * internal-fn.c (expand_mul_overflow): If both operands have
7893 the same highpart of -1 or 0 and the topmost bit of lowpart
7894 is different, overflow is if res <= 0 rather than res < 0.
7896 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
7898 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
7899 TARGET_P9_VECTOR code for unaligned_load case.
7901 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7903 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
7905 2017-10-13 Nathan Sidwell <nathan@acm.org>
7907 * tree-core.h (tree_contains_struct): Make bool.
7908 * tree.c (tree_contains_struct): Likewise.
7909 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
7910 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
7911 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
7912 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
7913 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
7915 2017-10-13 Richard Biener <rguenther@suse.de>
7917 * graphite-isl-ast-to-gimple.c
7918 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
7919 parameters and dominance check.
7920 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
7921 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
7922 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
7923 Do not update SSA form here or do intermediate IL verification.
7924 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
7925 (graphite_initialize): Remove check on the number of loops in
7926 the function and inline into graphite_transform_loops.
7927 (graphite_finalize): Inline into graphite_transform_loops.
7928 (graphite_transform_loops): Perform SSA update and IL verification
7930 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
7932 2017-10-13 Richard Biener <rguenther@suse.de>
7934 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
7935 graphite_expression_type_precision): Avoid global constructor
7937 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
7938 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
7939 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
7940 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
7941 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
7942 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
7944 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
7947 * config/i386/i386.h (ix86_red_zone_size): New.
7948 * config/i386/i386.md (push peephole2s): Replace
7949 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
7951 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7952 Alan Hayward <alan.hayward@arm.com>
7953 David Sherwood <david.sherwood@arm.com>
7955 * combine.c (can_change_dest_mode): Reject changes in
7956 REGMODE_NATURAL_SIZE.
7958 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7959 Alan Hayward <alan.hayward@arm.com>
7960 David Sherwood <david.sherwood@arm.com>
7962 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
7963 (expand_debug_source_expr): Likewise.
7964 * combine.c (combine_simplify_rtx): Likewise.
7965 * cse.c (fold_rtx): Likewise.
7966 * fwprop.c (canonicalize_address): Likewise.
7967 * targhooks.c (default_shift_truncation_mask): Likewise.
7969 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7970 Alan Hayward <alan.hayward@arm.com>
7971 David Sherwood <david.sherwood@arm.com>
7973 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
7974 (widened_mode): Likewise.
7975 (expand_unop): Likewise.
7976 * ree.c (transform_ifelse): Likewise.
7977 (merge_def_and_ext): Likewise.
7978 (combine_reaching_defs): Likewise.
7979 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7981 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7982 Alan Hayward <alan.hayward@arm.com>
7983 David Sherwood <david.sherwood@arm.com>
7985 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
7986 * combine.c (gen_lowpart_for_combine): Likewise.
7987 * dwarf2out.c (rtl_for_decl_location): Likewise.
7988 * final.c (alter_subreg): Likewise.
7989 * rtlhooks.c (gen_lowpart_general): Likewise.
7990 (gen_lowpart_if_possible): Likewise.
7992 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7993 Alan Hayward <alan.hayward@arm.com>
7994 David Sherwood <david.sherwood@arm.com>
7996 * calls.c (expand_call): Use subreg_lowpart_offset.
7997 * cse.c (cse_insn): Likewise.
7998 * regcprop.c (copy_value): Likewise.
7999 (copyprop_hardreg_forward_1): Likewise.
8001 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8004 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8005 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8006 =Q constraints to +Q and into insn condition add check
8007 that operands[0] and operands[1] are equal.
8008 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8009 =Q constraints to +Q and into insn condition add check
8010 that operands[0] is equal to either operands[1] or operands[2].
8013 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8014 instead of handling MINUS_EXPR twice (once for each argument),
8015 canonicalize operand order and handle just once, use rtype where
8016 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8019 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8020 any values of __C while still being pattern recognizable as a simple
8023 2017-10-13 Richard Biener <rguenther@suse.de>
8025 PR tree-optimization/82451
8027 2017-10-02 Richard Biener <rguenther@suse.de>
8029 PR tree-optimization/82355
8030 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8031 a mapping for the enclosing loop but avoid generating one for
8033 (copy_bb_and_scalar_dependences): Remove premature codegen
8034 error on PHIs in blocks duplicated into multiple places.
8035 * graphite-scop-detection.c
8036 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8037 in the region use it as loop and nest to analyze the DR in.
8038 (try_generate_gimple_bb): Likewise.
8039 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8040 (add_loop_constraints): For blocks in a loop not in the region
8041 create a dimension with a single iteration.
8042 * sese.h (gbb_loop_at_index): Remove assert.
8044 * cfgloop.c (loop_preheader_edge): For the loop tree root
8045 return the single successor of the entry block.
8046 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8047 Reset the SCEV hashtable and niters.
8048 * graphite-scop-detection.c
8049 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8050 assert that we only have POLYNOMIAL_CHREC that vary in loops
8051 contained in the region.
8052 (scop_detection::graphite_can_represent_expr): Adjust.
8053 (scop_detection::stmt_has_simple_data_refs_p): For loops
8054 not in the region set loop to NULL. The nest is now the
8055 entry edge to the region.
8056 (try_generate_gimple_bb): Likewise.
8057 * sese.c (scalar_evolution_in_region): Adjust for
8058 instantiate_scev change.
8059 * tree-data-ref.h (graphite_find_data_references_in_stmt):
8060 Make nest parameter the edge into the region.
8061 (create_data_ref): Likewise.
8062 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
8063 entry edge into a region and adjust instantiate_scev calls.
8064 (create_data_ref): Likewise.
8065 (graphite_find_data_references_in_stmt): Likewise.
8066 (find_data_references_in_stmt): Pass the loop preheader edge
8067 from the nest argument.
8068 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
8069 parameter the edge into the region.
8070 (instantiate_parameters): Use the loop preheader edge as entry.
8071 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
8073 (get_instantiated_value_entry): Make instantiate_below parameter
8074 the edge into the region.
8075 (instantiate_scev_name): Likewise. Adjust dominance checks,
8076 when we cannot use loop-based instantiation instantiate by
8077 walking use-def chains.
8078 (instantiate_scev_poly): Adjust.
8079 (instantiate_scev_binary): Likewise.
8080 (instantiate_scev_convert): Likewise.
8081 (instantiate_scev_not): Likewise.
8082 (instantiate_array_ref): Remove.
8083 (instantiate_scev_3): Likewise.
8084 (instantiate_scev_2): Likewise.
8085 (instantiate_scev_1): Likewise.
8086 (instantiate_scev_r): Do not blindly handle N-operand trees.
8087 Do not instantiate array-refs. Handle all constants and invariants.
8088 (instantiate_scev): Make instantiate_below parameter
8089 the edge into the region.
8090 (resolve_mixers): Use the loop preheader edge for the region
8091 parameter to instantiate_scev_r.
8092 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
8094 2017-10-13 Richard Biener <rguenther@suse.de>
8096 PR tree-optimization/82525
8097 * graphite-isl-ast-to-gimple.c
8098 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
8100 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
8101 Fail code generation when we cannot represent the isl integer.
8102 (binary_op_to_tree): Elide modulo operations that are no-ops
8103 in the type we code generate. Remove now superfluous code
8106 2017-10-13 Richard Biener <rguenther@suse.de>
8108 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
8109 (scop_detection::harmful_loop_in_region): Remove premature
8110 IV type restriction.
8111 (scop_detection::graphite_can_represent_scev): We can handle
8112 pointer IVs just fine.
8114 2017-10-13 Alan Modra <amodra@gmail.com>
8116 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
8117 "Clobbers and Scratch Registers". Add paragraph on
8118 alternative to clobbers for scratch registers and OpenBLAS
8121 2017-10-13 Alan Modra <amodra@gmail.com>
8123 * doc/extend.texi (Clobbers): Correct vax example. Delete old
8124 example of a memory input for a string of known length. Move
8125 commentary out of table. Add a number of new examples
8126 covering array memory inputs.
8128 2017-10-12 Martin Liska <mliska@suse.cz>
8130 PR tree-optimization/82493
8131 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
8132 (test_range_functions): New function.
8133 (sbitmap_c_tests): Likewise.
8134 * selftest-run-tests.c (selftest::run_tests): Run new tests.
8135 * selftest.h (sbitmap_c_tests): New function.
8137 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8139 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8141 * config/rs6000/amo.h: Fix spacing issue.
8143 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8146 * config/i386/i386.md (*ashl<mode>3_mask_1,
8147 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
8148 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
8151 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8153 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
8154 (profile_probability): Set max_probability
8155 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
8157 * profile-count.c (profile_probability::differs_from_p): Do not
8158 rely on max_probaiblity == 10000
8160 2017-10-12 Jeff Law <law@redhat.com>
8162 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
8165 2017-10-12 Martin Sebor <msebor@redhat.com>
8169 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
8170 (handle_alias_pairs): Call it.
8171 * common.opt (-Wattribute-alias): New option.
8172 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
8173 * doc/invoke.texi (-Wattribute-alias): Document.
8175 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
8178 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8180 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8182 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8183 (make_hard_regno_born, make_hard_regno_dead): Update
8184 bb_killed_pseudos and bb_gen_pseudos.
8186 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8188 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
8190 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
8192 * config/alpha/alpha.c (alpha_split_conditional_move):
8193 Use std::swap instead of manually swapping.
8194 (alpha_stdarg_optimize_hook): Ditto.
8195 (alpha_canonicalize_comparison): Ditto.
8197 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8199 * tree-loop-distribution.c (struct builtin_info): New struct.
8200 (struct partition): Refactor fields into struct builtin_info.
8201 (partition_free): Free struct builtin_info.
8202 (build_size_arg_loc, build_addr_arg_loc): Delete.
8203 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
8204 information from struct builtin_info.
8205 (find_single_drs): New function refactored from classify_partition.
8206 Also moved builtin validity checks to this function.
8207 (compute_access_range, alloc_builtin): New functions.
8208 (classify_builtin_st, classify_builtin_ldst): New functions.
8209 (classify_partition): Refactor code into functions find_single_drs,
8210 classify_builtin_st and classify_builtin_ldst.
8211 (distribute_loop): Don't do runtime alias check when distributing
8213 (find_seed_stmts_for_distribution): New function.
8214 (pass_loop_distribution::execute): Refactor code finding seed
8215 stmts into above function. Support distribution for the innermost
8216 two-level loop nest. Adjust dump information.
8218 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8220 * tree-loop-distribution.c: Adjust the general comment.
8221 (NUM_PARTITION_THRESHOLD): New macro.
8222 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
8223 (classify_partition): Skip builtin pattern of loop nest's inner loop.
8224 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
8225 in call to build_partition_graph.
8226 (finalize_partitions): New parameter. Make loop distribution more
8227 conservative by fusing more partitions.
8228 (distribute_loop): Don't do runtime alias check in case of loop nest
8230 (find_seed_stmts_for_distribution): New function.
8231 (prepare_perfect_loop_nest): New function.
8232 (pass_loop_distribution::execute): Refactor code finding seed stmts
8233 and loop nest into above functions. Support loop nest distribution.
8234 Adjust dump information accordingly.
8236 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8238 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
8239 and set PTYPE_SEQUENTIAL for merged partition.
8241 2017-10-12 Richard Biener <rguenther@suse.de>
8243 PR tree-optimization/69728
8245 2017-09-19 Richard Biener <rguenther@suse.de>
8247 PR tree-optimization/69728
8248 * graphite-sese-to-poly.c (schedule_error): New global.
8249 (add_loop_schedule): Handle empty domain by failing the
8251 (build_original_schedule): Handle schedule_error.
8253 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
8254 domain by returning an unchanged schedule.
8256 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8258 * genrecog.c (validate_pattern): For VEC_SELECT verify that
8259 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
8261 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
8263 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
8266 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8269 * expr.c (store_field): Don't optimize away bitsize == 0 store
8270 from CALL_EXPR with addressable return type.
8272 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8274 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
8275 * config/rs6000/rs6000.md (sel): Delete mode attribute.
8276 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
8277 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
8278 TARGET_ISEL instead of TARGET_ISEL<sel>.
8280 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
8282 * config/rs6000/rs6000.c
8283 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
8285 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8287 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
8289 (reg_or_cint_operand): Fix comment.
8290 (reg_or_zero_operand): New predicate.
8291 * config/rs6000/rs6000-protos.h (output_isel): Delete.
8292 * config/rs6000/rs6000.c (output_isel): Delete.
8293 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
8294 instead of reg_or_cint_operand. Output instruction directly (not via
8296 (isel_unsigned_<mode>): Ditto.
8297 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
8298 gpc_reg_operand. Add an instruction alternative for this. Output
8299 instruction directly.
8300 (*isel_reversed_unsigned_<mode>): Ditto.
8302 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8304 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
8305 (TARGET_CANONICALIZE_COMPARISON): Define.
8307 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
8310 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8311 Check whether the dest is REG before adding REG_EQUIV note.
8313 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8316 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8318 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8319 (make_hard_regno_born, make_hard_regno_dead): Update
8320 bb_killed_pseudos and bb_gen_pseudos.
8322 2017-10-11 Nathan Sidwell <nathan@acm.org>
8324 * incpath.h (enum incpath_kind): Name enum, prefix values.
8325 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
8326 * incpath.c (heads, tails): Use INC_MAX.
8327 (add_env_var_paths, add_standard_paths): Use incpath_kind.
8328 (merge_include_chains, split_quote_chain,
8329 register_include_chains): Update incpath_kind names.
8330 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
8331 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
8333 (add_framework_path, darwin_register_objc_includes): Likewise.
8334 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
8336 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8338 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
8339 Do not use float_operator operator predicate.
8340 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
8341 * config/i386/predicates.md (float_operator): Remove predicate.
8343 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8345 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
8346 (*jccxf_i387): Ditto.
8347 (*jcc<mode>_i387): Ditto.
8348 (*jccu<mode>_i387): Ditto.
8349 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
8350 (*jcc_*_i387 splitters): Remove.
8351 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
8352 * config/i386/i386.c (ix86_split_fp_branch): Remove.
8353 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
8356 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8358 * profile-count.h (slow_safe_scale_64bit): New function.
8359 (safe_scale_64bit): New inline.
8360 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
8361 * profile-count.c: Include wide-int.h
8362 (slow_safe_scale_64bit): New.
8364 2017-10-11 Nathan Sidwell <nathan@acm.org>
8366 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
8367 HAS_DECL_ASSEMBLER_NAME_P.
8368 * gimple-expr.c (gimple_decl_printable_name: Check
8369 HAS_DECL_ASSEMBLER_NAME_P too.
8370 * ipa-utils.h (type_in_anonymous_namespace_p): Check
8371 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
8372 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
8373 * passes.c (rest_of_decl_compilation): Check
8374 HAS_DECL_ASSEMBLER_NAME_P too.
8375 * recog.c (verify_changes): Likewise.
8376 * tree-pretty-print.c (dump_decl_name): Likewise.
8377 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
8379 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
8380 (SET_DECL_ASSEMBLER_NAME): Use it.
8381 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
8382 (COPY_DECL_ASSEMBLER_NAME): Likewise.
8383 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
8385 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8387 * config.gcc (i386, x86_64): Add extra objects.
8388 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
8389 (ix86_min_insn_size): Declare.
8390 (ix86_issue_rate): Declare.
8391 (ix86_adjust_cost): Declare.
8392 (ia32_multipass_dfa_lookahead): Declare.
8393 (ix86_macro_fusion_p): Declare.
8394 (ix86_macro_fusion_pair_p): Declare.
8395 (ix86_bd_has_dispatch): Declare.
8396 (ix86_bd_do_dispatch): Declare.
8397 (ix86_core2i7_init_hooks): Declare.
8398 (ix86_atom_sched_reorder): Declare.
8399 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
8400 (COSTS_N_BYTES): Move to x86-tune-costs.h.
8401 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
8402 (rip_relative_addr_p): Rename to ...
8403 (ix86_rip_relative_addr_p): ... this one; export.
8404 (memory_address_length): Update.
8405 (ix86_issue_rate): Move to x86-tune-sched.c.
8406 (ix86_flags_dependent): Move to x86-tune-sched.c.
8407 (ix86_agi_dependent): Move to x86-tune-sched.c.
8408 (exact_dependency_1): Move to x86-tune-sched.c.
8409 (exact_store_load_dependency): Move to x86-tune-sched.c.
8410 (ix86_adjust_cost): Move to x86-tune-sched.c.
8411 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
8412 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
8413 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
8414 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
8415 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
8416 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
8417 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
8418 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
8419 (min_insn_size): Rename to ...
8420 (ix86_min_insn_size): ... this one; export.
8421 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
8422 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
8423 (core2i7_first_cycle_multipass_backtrack): Move to
8424 x86-tune-sched-core.c.
8425 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
8426 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
8427 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
8428 (ix86_avoid_jump_mispredicts): Update.
8429 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
8430 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
8431 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
8432 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
8433 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
8434 (MAX_INSN): Move to ix86-tune-sched-bd.c.
8435 (MAX_IMM): Move to ix86-tune-sched-bd.c.
8436 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
8437 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
8438 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
8439 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
8440 (MAX_STORE): Move to ix86-tune-sched-bd.c.
8441 (BIG): Move to ix86-tune-sched-bd.c.
8442 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
8443 (enum insn_path): Move to ix86-tune-sched-bd.c.
8444 (get_mem_group): Move to ix86-tune-sched-bd.c.
8445 (is_cmp): Move to ix86-tune-sched-bd.c.
8446 (dispatch_violation): Move to ix86-tune-sched-bd.c.
8447 (is_branch): Move to ix86-tune-sched-bd.c.
8448 (is_prefetch): Move to ix86-tune-sched-bd.c.
8449 (init_window): Move to ix86-tune-sched-bd.c.
8450 (allocate_window): Move to ix86-tune-sched-bd.c.
8451 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
8452 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
8453 (process_end_window): Move to ix86-tune-sched-bd.c.
8454 (allocate_next_window): Move to ix86-tune-sched-bd.c.
8455 (find_constant): Move to ix86-tune-sched-bd.c.
8456 (get_num_immediates): Move to ix86-tune-sched-bd.c.
8457 (has_immediate): Move to ix86-tune-sched-bd.c.
8458 (get_insn_path): Move to ix86-tune-sched-bd.c.
8459 (get_insn_group): Move to ix86-tune-sched-bd.c.
8460 (count_num_restricted): Move to ix86-tune-sched-bd.c.
8461 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
8462 (add_insn_window): Move to ix86-tune-sched-bd.c.
8463 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
8464 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
8465 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
8466 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
8467 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
8468 (do_dispatch): Move to ix86-tune-sched-bd.c.
8469 (has_dispatch): Move to ix86-tune-sched-bd.c.
8470 * config/i386/t-i386: Add new object files.
8471 * config/i386/x86-tune-costs.h: New file.
8472 * config/i386/x86-tune-sched-atom.c: New file.
8473 * config/i386/x86-tune-sched-bd.c: New file.
8474 * config/i386/x86-tune-sched-core.c: New file.
8475 * config/i386/x86-tune-sched.c: New file.
8477 2017-10-11 Liu Hao <lh_mouse@126.com>
8479 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
8480 the generic version below instead.
8481 (should_colorize): Recognize Windows consoles as terminals
8483 * pretty-print.c [__MINGW32__] (write_all): New function.
8484 [__MINGW32__] (find_esc_head): Likewise.
8485 [__MINGW32__] (find_esc_terminator): Likewise.
8486 [__MINGW32__] (eat_esc_sequence): Likewise.
8487 [__MINGW32__] (mingw_ansi_fputs): New function that handles
8489 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
8492 2017-10-11 Richard Biener <rguenther@suse.de>
8494 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
8495 Properly call analyze_scalar_evolution with the loop of the stmt.
8497 2017-10-11 Richard Biener <rguenther@suse.de>
8499 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
8500 * tree-core.h (tree_base): Add chrec_var union member.
8501 * tree.h (CHREC_VAR): Remove.
8502 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
8503 * tree-chrec.h (build_polynomial_chrec): Adjust.
8504 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
8505 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
8507 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
8509 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
8510 * match.pd: ... here.
8511 ((T) X == (T) Y): Relax condition.
8513 2017-10-11 Bin Cheng <bin.cheng@arm.com>
8515 PR tree-optimization/82472
8516 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
8518 (break_alias_scc_partitions): Update postorder number.
8520 2017-10-11 Martin Liska <mliska@suse.cz>
8523 * opts.c (parse_no_sanitize_attribute): Do not use error_value
8525 * opts.h (parse_no_sanitize_attribute): Remove last argument.
8527 2017-10-11 Martin Liska <mliska@suse.cz>
8529 * print-rtl.c (print_insn): Move declaration of idbuf
8530 to same scope as name.
8532 2017-10-11 Martin Liska <mliska@suse.cz>
8537 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8540 2017-10-11 Martin Liska <mliska@suse.cz>
8543 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8546 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
8548 * config.gcc (default_gnu_indirect_function): Default to yes for
8549 arm*-*-linux* with glibc.
8551 2017-10-11 Richard Biener <rguenther@suse.de>
8553 * tree-scalar-evolution.c (get_scalar_evolution): Handle
8554 default-defs and types we do not want to analyze.
8555 (interpret_loop_phi): Replace unreachable code with an assert.
8556 (compute_scalar_evolution_in_loop): Remove and inline ...
8557 (analyze_scalar_evolution_1): ... here, replacing condition with
8558 what makes the intent clearer. Remove handling of cases
8559 get_scalar_evolution now handles.
8561 2017-10-10 Jim Wilson <wilson@tuliptree.org>
8563 PR rtl-optimization/81434
8564 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
8565 comment for main loop. In sched_group_found if, also add checks for
8566 pass and min_cost_group.
8568 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
8570 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
8571 (rs6000_insn_cost): New function.
8572 * config/rs6000/rs6000.md (cost): New attribute.
8574 2017-10-10 Jakub Jelinek <jakub@redhat.com>
8575 H.J. Lu <hongjiu.lu@intel.com>
8579 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
8580 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
8581 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
8582 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
8583 to OPTION_MASK_ISA_AVX512VL - builtins that have both
8584 OPTION_MASK_ISA_MMX and some other bit set require both
8585 mmx and the ISAs without the mmx bit.
8586 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
8587 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
8588 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
8589 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
8590 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
8591 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
8592 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
8593 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
8594 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
8595 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
8596 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
8597 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
8598 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
8599 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
8600 Add OPTION_MASK_ISA_MMX.
8602 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
8604 * config.gcc (armv7*-*-freebsd*): New target.
8605 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
8607 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
8609 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
8610 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
8613 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
8615 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
8616 a template parameter.
8617 (WIDE_INT_REF_FOR): Update accordingly.
8618 * tree.h (wi::int_traits <const_tree>): Delete.
8619 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
8620 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
8621 (wi::tree_to_wide_ref): New typedef.
8622 (wi::to_wide): New function.
8623 * calls.c (get_size_range): Use wi::to_wide when operating on
8625 * cgraph.c (cgraph_node::create_thunk): Likewise.
8626 * config/i386/i386.c (ix86_data_alignment): Likewise.
8627 (ix86_local_alignment): Likewise.
8628 * dbxout.c (stabstr_O): Likewise.
8629 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
8630 * expr.c (const_vector_from_tree): Likewise.
8631 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
8632 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
8633 (fold_negate_expr_1, int_const_binop_1, const_binop)
8634 (fold_convert_const_int_from_real, optimize_bit_field_compare)
8635 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
8636 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
8637 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
8638 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
8639 (fold_not_const, round_up_loc): Likewise.
8640 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
8641 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
8642 (alloca_call_type): Likewise.
8643 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
8644 * godump.c (go_output_typedef): Likewise.
8645 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
8646 * internal-fn.c (get_min_precision): Likewise.
8647 * ipa-cp.c (ipcp_store_vr_results): Likewise.
8648 * ipa-polymorphic-call.c
8649 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
8650 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
8651 (ipa_modify_call_arguments): Likewise.
8652 * match.pd: Likewise.
8653 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
8654 * print-tree.c (print_node_brief, print_node): Likewise.
8655 * stmt.c (expand_case): Likewise.
8656 * stor-layout.c (layout_type): Likewise.
8657 * tree-affine.c (tree_to_aff_combination): Likewise.
8658 * tree-cfg.c (group_case_labels_stmt): Likewise.
8659 * tree-data-ref.c (dr_analyze_indices): Likewise.
8660 (prune_runtime_alias_test_list): Likewise.
8661 * tree-dump.c (dequeue_and_dump): Likewise.
8662 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
8663 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
8664 * tree-pretty-print.c (dump_generic_node): Likewise.
8665 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
8666 (simple_iv_with_niters): Likewise.
8667 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
8668 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
8669 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
8670 * tree-ssa-loop-niter.c (split_to_var_and_offset)
8671 (refine_value_range_using_guard, number_of_iterations_ne_max)
8672 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
8673 (get_cst_init_from_scev, record_nonwrapping_iv)
8674 (scev_var_range_cant_overflow): Likewise.
8675 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
8676 * tree-ssa-pre.c (compute_avail): Likewise.
8677 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
8678 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
8679 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
8680 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
8681 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
8682 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
8683 (dump_case_nodes, try_switch_expansion): Likewise.
8684 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
8685 (vect_do_peeling): Likewise.
8686 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
8687 * tree-vect-stmts.c (vectorizable_load): Likewise.
8688 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
8689 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
8690 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
8691 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
8692 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
8693 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
8694 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
8695 (evrp_dom_walker::before_dom_children): Likewise.
8696 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
8697 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
8698 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
8699 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
8700 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
8701 (get_range_pos_neg): Likewise.
8702 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
8703 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
8704 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
8705 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
8706 * config/avr/avr.c (avr_fold_builtin): Likewise.
8707 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
8708 * config/msp430/msp430.c (msp430_attr): Likewise.
8709 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
8710 * config/powerpcspe/powerpcspe-c.c
8711 (altivec_resolve_overloaded_builtin): Likewise.
8712 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
8713 (rs6000_expand_ternop_builtin): Likewise.
8714 * config/rs6000/rs6000-c.c
8715 (altivec_resolve_overloaded_builtin): Likewise.
8716 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
8717 (rs6000_expand_ternop_builtin): Likewise.
8718 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
8720 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8722 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
8723 when copying loop nest with only one inner loop.
8725 2017-10-10 Richard Biener <rguenther@suse.de>
8727 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
8728 blocks if SCEV is active.
8729 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
8731 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
8732 (scev_initialize): Assert we are not yet initialized.
8734 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8736 * tree-loop-distribution.c (generate_loops_for_partition): Remove
8737 inner loop's exit stmt by making it always exit the loop, otherwise
8738 we would generate an infinite empty loop.
8740 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8742 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
8743 renaming variables in new preheader if it's deleted.
8745 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8747 * tree-loop-distribution.c (struct partition): Remove unused field
8748 loops of the structure.
8749 (partition_alloc, partition_free): Ditto.
8750 (build_rdg_partition_for_vertex): Ditto.
8752 2017-10-09 Jeff Law <law@redhat.com>
8754 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
8755 return type to match prototype and documentation.
8757 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8759 * config/rs6000/rs6000.c (processor_costs): Move to ...
8760 * config/rs6000/rs6000.h: ... here.
8761 (rs6000_cost): Declare.
8763 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
8765 * except.c (setjmp_fn): New global variable.
8766 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
8767 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
8768 if DONT_USE_BUILTIN_SETJMP is defined.
8770 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8772 * target.def (insn_cost): New hook.
8773 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
8774 * doc/tm.texi: Regenerate.
8775 * rtlanal.c (insn_cost): Use the new hook.
8777 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8779 * combine.c (combine_validate_cost): Compute the new insn_cost,
8780 not just pattern_cost.
8781 (try_combine): Adjust comment.
8783 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8785 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
8787 * combine.c (uid_insn_cost): Adjust comment.
8788 (combine_validate_cost): Adjust comment. Use pattern_cost instead
8790 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
8791 * dse.c (find_shift_sequence): Ditto.
8792 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
8793 (bb_valid_for_noce_process_p): Use pattern_cost.
8794 * rtl.h (insn_rtx_cost): Delete.
8795 (pattern_cost): New prototype.
8796 (insn_cost): New prototype.
8797 * rtlanal.c (insn_rtx_cost): Rename to...
8798 (pattern_cost): ... this.
8801 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
8803 * config/i386/i386.md (*jcc_2): Remove insn pattern.
8804 (*jcc<mode>_0_r_i387): Ditto.
8805 (*jccxf_r_i387): Ditto.
8806 (*jcc<mode>_r_i387): Ditto.
8807 (*jccu<mode>_r_i387): Ditto.
8808 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
8809 (*jcc): Rename from *jcc_1.
8811 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8813 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
8814 deferred rescans after the lvx/stvx recombination pre-pass.
8816 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
8818 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
8819 memory operation instruction support.
8820 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
8821 (rs6000-ibm-aix[789]*): Likewise.
8822 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
8823 Document new functions.
8825 2017-10-09 Richard Biener <rguenther@suse.de>
8827 PR tree-optimization/82397
8828 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
8829 equality only for semantically equal trees.
8831 2017-10-09 Richard Biener <rguenther@suse.de>
8833 PR tree-optimization/82449
8834 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
8835 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
8836 allow constant addresses.
8837 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
8840 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8842 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
8845 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8848 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
8851 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8854 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
8856 2017-10-09 Jakub Jelinek <jakub@redhat.com>
8859 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
8860 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
8862 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
8864 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
8865 (WI_BINARY_PREDICATE_RESULT): Likewise.
8866 (wi::binary_traits::operator_result): New type.
8867 (wi::binary_traits::predicate_result): Likewise.
8868 (generic_wide_int::operator~, unary generic_wide_int::operator-)
8869 (generic_wide_int::operator==, generic_wide_int::operator!=)
8870 (generic_wide_int::operator&, generic_wide_int::and_not)
8871 (generic_wide_int::operator|, generic_wide_int::or_not)
8872 (generic_wide_int::operator^, generic_wide_int::operator+
8873 (binary generic_wide_int::operator-, generic_wide_int::operator*):
8875 (operator~, unary operator-, operator==, operator!=, operator&)
8876 (operator|, operator^, operator+, binary operator-, operator*): New
8878 * expr.c (get_inner_reference): Use wi::bit_and_not.
8879 * fold-const.c (fold_binary_loc): Likewise.
8880 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
8881 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
8882 (bit_value_binop): Likewise.
8883 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
8884 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
8885 (extract_range_from_binary_expr_1): Likewise.
8886 (masked_increment): Likewise.
8887 (simplify_bit_ops_using_ranges): Likewise.
8889 2017-10-09 Martin Jambor <mjambor@suse.cz>
8892 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
8893 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
8894 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
8896 (hsa_fixup_mov_insn_type): New function.
8897 (hsa_op_with_type::get_in_type): Use it.
8898 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
8899 immediates in an assert.
8900 (hsa_op_with_type::extend_int_to_32bit): New method.
8901 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
8902 types. Convert to dest type if necessary.
8903 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
8904 (reg_for_gimple_ssa): Pass false as min32int to
8905 hsa_type_for_scalar_tree_type.
8906 (gen_hsa_addr): Fixup type when creating addresable temporary.
8907 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
8908 (gen_hsa_unary_operation): Extend operands and convert to dest type if
8909 necessary. Call hsa_fixup_mov_insn_type.
8910 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
8911 extend operands and convert to dest type if necessary.
8912 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
8913 to dest type if necessary.
8914 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
8915 if conversion nt necessary and size matches.
8916 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
8917 to dest type if necessary.
8918 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
8919 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
8921 (gen_hsa_clrsb): Likewise.
8922 (gen_hsa_ffs): Likewise.
8923 (gen_hsa_divmod): Extend operands and convert to dest type if
8925 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
8927 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8929 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
8930 Remove empty default arguments. Use a brace block as output
8932 (conditional return): Ditto.
8934 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
8935 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
8937 (group_ending_nop): Ditto.
8938 (doloop_end): Ditto.
8939 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
8940 (splitters for those): Ditto.
8942 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8944 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
8945 a conditional jump (and the compare for it) so that pc_rtx is the
8947 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
8948 for the deleted and renamed ctr<mode>_internal[234] patterns.
8949 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
8950 Delete second conditional return pattern.
8951 (ctr<mode>_internal2): Delete this second bdnz pattern.
8952 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
8953 (ctr<mode>_internal4): Delete this second bdz pattern.
8955 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
8957 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
8958 (always_initialized_rtx_for_ssa_name_p): New predicate.
8959 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
8960 (finish_out_of_ssa): Free new field of SA.
8961 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
8962 * tree-ssa-coalesce.c: Include tree-ssa.h.
8963 (get_parm_default_def_partitions): Remove extern keyword.
8964 (get_undefined_value_partitions): New function.
8965 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
8966 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
8968 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
8970 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
8972 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
8974 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8976 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
8977 for targets that preffer 128bit.
8979 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8981 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
8983 2017-10-08 Olivier Hainque <hainque@adacore.com>
8985 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
8986 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
8988 (thumb_set_return_address): Likewise.
8990 2017-10-08 Olivier Hainque <hainque@adacore.com>
8992 * common/config/arm/arm-common.c (arm_except_unwind_info):
8993 Handle DWARF2_UNWIND_INFO.
8995 2017-10-07 Michael Collison <michael.collison@arm.com>
8997 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9000 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9002 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9003 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9004 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9005 defined, force the creation of a new block for a dispatch label.
9007 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9009 * invoke.texi (Wsuggest-attribute=cold): Document.
9010 * common.opt (Wsuggest-attribute=cold): New
9011 * ipa-pure-const.c (warn_function_cold): New function.
9012 * predict.c (compute_function_frequency): Use it.
9013 * predict.h (warn_function_cold): Declare.
9015 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9017 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9020 2017-10-06 Martin Liska <mliska@suse.cz>
9022 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9023 keyword for member functions.
9024 (struct sanopt_tree_couple): New struct.
9025 (struct sanopt_tree_couple_hash): New function.
9026 (struct sanopt_ctx): Add new hash_map.
9027 (has_dominating_ubsan_ptr_check): New function.
9028 (record_ubsan_ptr_check_stmt): Likewise.
9029 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9030 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9031 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9033 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9036 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9037 aarch64_simd_valid_immediate on CONST_VECTORs.
9038 (aarch64_reg_or_bic_imm): Likewise.
9040 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9042 PR rtl-optimization/82396
9043 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9045 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9047 * graphite-dependences.c (scop_get_reads): Move code to...
9048 (scop_get_must_writes): Move code to...
9049 (scop_get_may_writes): Move code to...
9050 (scop_get_reads_and_writes): ... here.
9051 (scop_get_dependences): Call scop_get_reads_and_writes.
9053 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9055 PR tree-optimization/82434
9056 * fold-const.h (can_native_encode_type_p,
9057 can_native_encode_string_p): Remove.
9058 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9059 don't encode anything, just return what would be otherwise returned.
9060 (native_encode_fixed, native_encode_complex, native_encode_vector):
9062 (native_encode_string): Likewise. Inline by hand
9063 can_native_encode_string_p.
9064 (can_native_encode_type_p): Remove.
9065 (can_native_encode_string_p): Remove.
9066 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
9067 STRING_CSTs using can_native_encode_string_p, test all
9068 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
9069 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
9070 argument from native_encode_expr.
9071 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
9072 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
9075 2017-10-06 Richard Biener <rguenther@suse.de>
9077 PR tree-optimization/82397
9078 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
9079 operand_equal_p but rely on data_ref_compare_tree for detecting
9081 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
9082 to match up with dr_group_sort_cmp.
9084 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9087 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
9089 * config/s390/s390-builtin-types.def: Regenerate.
9091 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9094 * config/s390/s390-builtin-types.def: Regenerate.
9095 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
9096 Change flag from B_VXE to B_VX.
9097 (s390_vec_min_dbl): Remove B_VXE flag.
9099 2017-10-06 Richard Biener <rguenther@suse.de>
9101 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
9102 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
9103 translate_isl_ast_to_gimple::is_valid_rename,
9104 translate_isl_ast_to_gimple::get_rename,
9105 translate_isl_ast_to_gimple::get_def_bb_for_const,
9106 translate_isl_ast_to_gimple::get_new_name,
9107 translate_isl_ast_to_gimple::collect_all_ssa_names,
9108 translate_isl_ast_to_gimple::copy_loop_phi_args,
9109 translate_isl_ast_to_gimple::collect_all_ssa_names,
9110 translate_isl_ast_to_gimple::copy_loop_phi_args,
9111 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
9112 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
9113 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
9114 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
9115 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
9116 translate_isl_ast_to_gimple::copy_cond_phi_args,
9117 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
9118 translate_isl_ast_to_gimple::edge_for_new_close_phis,
9119 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
9120 translate_isl_ast_to_gimple::rename_uses,
9121 translate_isl_ast_to_gimple::rename_all_uses): Remove.
9122 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
9123 (set_rename_for_each_def): Likewise.
9124 (graphite_copy_stmts_from_block): Handle debug stmt resetting
9125 here. Handle rewriting SCEV analyzable uses here.
9126 (copy_bb_and_scalar_dependences): Generate code for PHI
9128 (graphite_regenerate_ast_isl): Adjust.
9129 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
9130 (add_write, add_read): New functions.
9131 (build_cross_bb_scalars_def): Use it and simplify.
9132 (build_cross_bb_scalars_use): Likewise.
9133 (graphite_find_cross_bb_scalar_vars): Inline into...
9134 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
9135 simulating out-of-SSA. Compute liveout and add dependencies.
9136 (build_scops): Force an empty entry block.
9137 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
9139 (sese_build_liveouts): Declare.
9140 (sese_trivially_empty_bb_p): Likewise.
9141 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
9142 compute liveout and debug_liveout.
9143 (sese_bad_liveouts_use): Remove.
9144 (sese_reset_debug_liveouts_bb): Likewise.
9145 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
9146 (sese_build_liveouts): Build liveout and debug_liveout and store
9148 (new_sese_info): Adjust.
9149 (free_sese_info): Likewise.
9150 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
9151 do not build liveout here.
9152 (move_sese_in_condition): Adjust region entry.
9153 (scev_analyzable_p): Match up with chrec_apply requirements.
9154 (sese_trivially_empty_bb_p): New.
9155 * tree-into-ssa.c (get_reaching_def): Properly support generating
9156 default-defs for incremental rewrite of anonymous names.
9158 2017-10-06 Richard Biener <rguenther@suse.de>
9160 * graphite-sese-to-poly.c (extract_affine): For casts increasing
9161 precision do not perform modulo reduction.
9163 2017-10-06 Richard Biener <rguenther@suse.de>
9165 PR tree-optimization/82436
9166 * tree-vect-slp.c (vect_supported_load_permutation_p): More
9167 conservatively choose the vectorization factor when checking
9168 whether we can perform the required load permutation.
9169 (vect_transform_slp_perm_load): Assert when we may not fail.
9171 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
9173 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
9174 message for incompatible -msdata=* and -mcall-* options.
9176 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9178 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
9179 rate for post-reload scheduling.
9181 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9183 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
9185 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9187 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
9188 to improve monte carlo in scimark.
9190 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9192 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
9193 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
9194 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
9195 pentium4_cost, nocona_cost): Set reassociation width to 1.
9196 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
9197 width to 2 for fp operations and 1 otherwise.
9198 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
9200 (atom_cost): Set reassociation width to 2.
9201 (slm_cost, generic_cost): Set fp reassociation width
9202 to 2 and 1 otherwise.
9203 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
9204 (core_cost): Set fp reassociation width to 4 and vector to 2.
9205 (ix86_reassociation_width): Rewrite using cost table; special case
9206 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
9207 and TARGET_AVX128_OPTIMAL.
9208 * config/i386/i386.h (processor_costs): Add
9209 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
9210 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
9211 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
9212 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
9213 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
9214 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
9216 2017-10-05 Nathan Sidwell <nathan@acm.org>
9218 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
9220 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9222 * config/arm/arm.c (arm_test_fpu_data): New.
9223 (arm_run_selftests): Call arm_test_fpu_data.
9225 2017-10-04 Nathan Sidwell <nathan@acm.org>
9227 * toplev.c (toplev::main): Remove excess parens on pretty_printer
9229 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
9231 2017-10-04 Sudakshina Das <sudi.das@arm.com>
9233 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
9234 check type for aarch64_simd_valid_immediate.
9235 (aarch64_output_simd_mov_immediate): Update prototype.
9236 (aarch64_simd_valid_immediate): Update prototype.
9237 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
9238 support for ORR-immediate.
9239 (and<mode>3): modified pattern to add support for BIC-immediate.
9240 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
9241 now checks for valid immediate for BIC and ORR based on new enum
9243 (aarch64_output_simd_mov_immediate): Function now used to output
9244 BIC/ORR imm as well based on new enum argument.
9245 * config/aarch64/constraints.md (Do): New vector immediate constraint.
9247 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
9248 (aarch64_reg_or_bic_imm): Likewise.
9250 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9252 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
9253 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
9255 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9259 PR rtl-optimization/82396
9260 * haifa-sched.c (autopref_multipass_init): Simplify
9262 (autopref_rank_data): Simplify sort order.
9263 * sched-int.h (autopref_multipass_data_): Remove
9264 multi_mem_insn_p, min_offset and max_offset.
9266 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9268 * doc/sourcebuild.texi: Document vect_peeling_profitable.
9270 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9272 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
9275 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9277 * doc/sourcebuild.texi: Document vect_long_mult.
9279 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
9281 PR tree-optimization/82413
9282 * fold-const.c (build_range_check): Use widest_int when comparing
9283 the maximum ETYPE value with HIGH.
9285 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9287 PR rtl-optimization/82396
9288 * haifa-sched.c (autopref_multipass_init): Simplify
9290 (autopref_rank_data): Simplify sort order.
9291 * sched-int.h (autopref_multipass_data_): Remove
9292 multi_mem_insn_p, min_offset and max_offset.
9294 2017-10-04 Jakub Jelinek <jakub@redhat.com>
9296 PR tree-optimization/82381
9297 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
9298 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
9301 PR tree-optimization/82374
9302 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
9303 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
9304 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
9305 current_function_decl to the new decl.
9307 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
9309 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
9310 helper macro for IEEE float128 hardware built-in functions.
9311 (SQRTF128_ODD): Add built-in functions with the round-to-odd
9313 (TRUNCF128_ODD): Likewise.
9314 (ADDF128_ODD): Likewise.
9315 (SUBF128_ODD): Likewise.
9316 (MULF128_ODD): Likewise.
9317 (DIVF128_ODD): Likewise.
9318 (FMAF128_ODD): Likewise.
9319 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
9320 UNSPEC_TRUNC_ROUND_TO_ODD.
9321 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
9322 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
9323 floating point round to odd instructions.
9324 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
9325 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
9326 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
9327 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
9328 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
9329 (trunc<mode>sf2_hw): Change the truncate with round to odd
9330 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
9331 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
9332 to odd hardware instructions.
9333 (sub<mode>3_odd): Likewise.
9334 (mul<mode>3_odd): Likewise.
9335 (div<mode>3_odd): Likewise.
9336 (sqrt<mode>2_odd): Likewise.
9337 (fma<mode>4_odd): Likewise.
9338 (fms<mode>4_odd): Likewise.
9339 (nfma<mode>4_odd): Likewise.
9340 (nfms<mode>4_odd): Likewise.
9341 (trunc<mode>df2_odd): Change the truncate with round to odd
9342 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
9344 * doc/extend.texi (PowerPC built-in functions): Update documentation
9345 for existing IEEE float128-bit built-in functions. Add built-in
9346 functions that generate the IEEE 128-bit floating point round to
9349 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
9351 PR rtl-optimization/77729
9352 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
9353 to (X&(C1&~C2))|C2 transformations.
9355 2017-10-03 Martin Jambor <mjambor@suse.cz>
9357 PR tree-optimization/82363
9358 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
9359 mismatch, mark lacc written regardless of racc.
9361 2017-10-03 Jakub Jelinek <jakub@redhat.com>
9363 PR tree-optimization/82381
9364 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
9365 stmt_to_insert nor wheather SSA_NAMEs are default defs.
9366 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
9367 fallthrough into reassoc_stmt_dominates_stmt_p.
9370 * combine.c (combine_instructions): Don't combine in unreachable
9373 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9376 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
9377 function to not use the have_cpu variable. Do not set cpu_index,
9378 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
9380 (rs6000_valid_attribute_p): Remove duplicate initializations of
9381 old_optimize and func_optimize.
9382 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
9383 (rs6000_activate_target_options): Make global.
9384 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
9387 2017-10-02 Jakub Jelinek <jakub@redhat.com>
9389 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
9390 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
9391 Set *poffset to 0 and *psize and *pmax_size to -1 if
9392 *poffset + *psize overflows in HOST_WIDE_INT.
9394 PR tree-optimization/82387
9395 PR tree-optimization/82388
9396 PR tree-optimization/82389
9397 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
9398 instead of live_bytes non-NULL.
9400 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
9403 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
9406 2017-10-02 Richard Biener <rguenther@suse.de>
9408 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
9409 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
9410 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
9412 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9414 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
9415 requested precision matches the type's.
9416 * calls.c (alloc_max_size): Calculate the new candidate size as
9417 a widest_int and use wi::to_widest when comparing it with the
9418 current candidate size.
9419 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
9420 zero rather than integer_zero_node.
9421 * match.pd: Check for a no-op conversion before using wi::add
9422 rather than after. Use tree_to_uhwi when summing small shift
9423 counts into an unsigned int.
9425 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9426 Alan Hayward <alan.hayward@arm.com>
9427 David Sherwood <david.sherwood@arm.com>
9430 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
9431 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
9432 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
9433 POINTER_AND_FP_REGS.
9435 2017-10-02 Richard Biener <rguenther@suse.de>
9437 PR tree-optimization/82355
9438 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9439 a mapping for the enclosing loop but avoid generating one for
9441 (copy_bb_and_scalar_dependences): Remove premature codegen
9442 error on PHIs in blocks duplicated into multiple places.
9443 * graphite-scop-detection.c
9444 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9445 in the region use it as loop and nest to analyze the DR in.
9446 (try_generate_gimple_bb): Likewise.
9447 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9448 (add_loop_constraints): For blocks in a loop not in the region
9449 create a dimension with a single iteration.
9450 * sese.h (gbb_loop_at_index): Remove assert.
9452 2017-10-01 Kevin Buettner <kevinb@redhat.com>
9454 * omp-expand.c (adjust_context_scope): New function.
9455 (expand_parallel_call): Call adjust_context_scope.
9457 2017-10-01 Jeff Law <law@redhat.com>
9459 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
9460 dom_opt_dom_walker class with direct access to private members.
9461 Add comments. Call test_for_singularity.
9462 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
9463 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
9464 m_dummy_cond anymore.
9465 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
9467 (pass_dominator:execute): Build the dummy_cond here and pass it
9468 to the dom_opt_dom_walker ctor.
9469 (test_for_singularity): New function.
9471 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
9472 Maya Rashish <coypu@sdf.org>
9474 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
9475 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
9476 (alpha*-*-netbsd*) Use nbsd_tm_file.
9477 (arm*-*-netbsdelf*) Likewise.
9478 (i[34567]86-*-netbsdelf*) Likewise.
9479 (x86_64-*-netbsd*) Likewise.
9480 (mips*-*-netbsd*) Likewise.
9481 (powerpc-*-netbsd*) Likewise.
9482 (sh*-*-netbsd*) Likewise.
9483 (sparc-*-netbsdelf*) Likewise.
9484 (sparc64-*-netbsd*) Likewise.
9485 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
9487 (vax-*-netbsdelf*) Likewise.
9488 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
9489 (UINT_FAST8_TYPE) Likewise.
9490 (INT_FAST16_TYPE) Check CHAR_FAST16.
9491 (UINT_FAST16_TYPE) Likewise.
9493 2017-09-30 Jakub Jelinek <jakub@redhat.com>
9496 * config/i386/i386.md
9497 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
9498 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
9499 *divmodsi4_zext_2): New define_insn_and_split.
9500 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
9501 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
9502 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
9503 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
9504 New define_insn_and_split.
9505 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
9506 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
9507 operands[1] having DImode when mode is SImode.
9509 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
9510 always SImode for DIV and MOD in REG_EQUAL notes.
9512 2017-09-29 Yury Gribov <tetra2005@gmail.com>
9515 * match.pd: Fix handling of NaNs in pattern.
9517 2017-09-29 Jeff Law <law@redhat.com>
9519 * sbitmap.c (bitmap_bit_in_range_p): New function.
9520 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
9521 * tree-ssa-dse.c (live_bytes_read): New function.
9522 (dse_classify_store): Ignore reads of dead bytes.
9524 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
9525 typos and whitespace errors.
9526 * config/i386/predicates.md (address_no_seg_operand): Likewise.
9527 * config/s390/s390.c (s390_emit_prologue): Likewise.
9529 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9532 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
9533 with a symbol for LRA.
9535 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9537 PR rtl-optimization/82338
9538 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
9540 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
9542 * genmodes.c (calc_wider_mode): Suppress qsort macro.
9543 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
9544 (qsort_chk): Declare.
9545 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
9546 (qsort_chk): New function.
9548 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9550 PR tree-optimization/82337
9551 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9552 phi definition if the PHI result appears in an abnormal PHI.
9553 (find_basis_for_base_expr): Don't record a basis if the LHS of the
9554 basis appears in an abnormal PHI.
9556 2017-09-29 Richard Biener <rguenther@suse.de>
9558 * graphite-isl-ast-to-gimple.c
9559 (translate_isl_ast_to_gimple::set_codegen_error): New function.
9560 (binary_op_to_tree): Use it.
9561 (get_rename_from_scev): Likewise.
9562 (copy_loop_phi_nodes): Likewise.
9563 (copy_bb_and_scalar_dependences): Likewise.
9564 (translate_pending_phi_nodes): Likewise.
9566 2017-09-29 Jakub Jelinek <jakub@redhat.com>
9569 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
9570 for movabsq $(i32 << shift), r64.
9572 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
9574 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
9575 index when encoding %esp as %rsp to avoid 0x67 prefix.
9577 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
9579 * config/i386/i386.md (*movsf_internal, *movdf_internal):
9580 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
9582 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
9584 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
9585 Extensions with more than 16 double VFP registers.
9586 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
9587 to_clear_mask and all code related to it. Replace the remaining
9588 entry by a sbitmap and adapt code accordingly.
9590 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
9592 * brig-builtins.def: Change pure attributes to const.
9594 2017-09-28 Joseph Myers <joseph@codesourcery.com>
9596 * config.gcc (default_gnu_indirect_function): Default to yes for
9597 sparc*-*-linux* with glibc.
9599 2017-09-28 Joseph Myers <joseph@codesourcery.com>
9601 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
9602 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
9603 when creating .init_array and .fini_array sections with priority
9606 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
9609 * config/aarch64/aarch64.c
9610 (aarch64_builtin_support_vector_misalignment): Always return false
9611 when misalignment is unknown.
9613 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
9615 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
9616 this function to return false if the definition used by the swap
9617 instruction is artificial, or if the memory address from which the
9618 constant value is loaded is not represented by a base address held
9619 in a register or if the base address register is a frame or stack
9620 pointer. Additionally, return false if the base address of the
9621 loaded constant is a SYMBOL_REF but is not considered to be a
9623 (replace_swapped_load_constant): New function.
9624 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
9625 loaded constant vector with a load of a swapped constant vector.
9627 2017-09-27 Carl Love <cel@us.ibm.com>
9629 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
9630 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
9631 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
9634 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
9636 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
9637 first, always call autopref_rank_data otherwise.
9639 2017-09-27 Richard Biener <rguenther@suse.de>
9641 * graphite-scop-detection.c (find_scop_parameters): Move
9642 loop bound handling ...
9643 (gather_bbs::before_dom_children): ... here, avoiding the need
9644 to build scop_info->loop_nest.
9645 (record_loop_in_sese): Remove.
9646 * sese.h (sese_info_t::loop_nest): Remove.
9647 * sese.c (new_sese_info): Do not allocate loop_nest.
9648 (free_sese_info): Do not free loop_nest.
9650 2017-09-27 Jakub Jelinek <jakub@redhat.com>
9653 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
9654 lhs from calls if the lhs has addressable type.
9656 2017-09-27 Richard Biener <rguenther@suse.de>
9658 * graphite.h (scop::max_alias_set): New member.
9659 * graphite-scop-detection.c: Remove references to non-existing
9660 --param in comments.
9661 (build_alias_sets): Record the maximum alias set used for drs.
9662 (build_scops): Support zero as unlimited for
9663 --param graphite-max-arrays-per-scop.
9664 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
9666 (build_poly_sr_1): ... here. Compute alias set based on the
9667 maximum alias set used for drs rather than
9668 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
9670 2017-09-27 Richard Biener <rguenther@suse.de>
9672 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
9673 --param loop-block-tile-size=0 to disable tiling.
9675 2017-09-27 Richard Biener <rguenther@suse.de>
9677 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
9678 (graphite-max-nb-scop-params): Document special value zero.
9679 * domwalk.h (dom_walker::STOP): New symbolical constant.
9680 (dom_walker::dom_walker): Add optional parameter for bb to
9682 (dom_walker::~dom_walker): Declare.
9683 (dom_walker::before_dom_children): Document STOP return value.
9684 (dom_walker::m_user_bb_to_rpo): New member.
9685 (dom_walker::m_bb_to_rpo): Likewise.
9686 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
9687 mapping here if not provided by the user.
9688 (dom_walker::~dom_walker): Free bb to RPO mapping if not
9689 provided by the user.
9690 (dom_walker::STOP): Define.
9691 (dom_walker::walk): Do not compute bb to RPO mapping here.
9692 Support STOP return value from before_dom_children to stop
9694 * graphite-optimize-isl.c (optimize_isl): If the schedule
9695 is the same still generate code if -fgraphite-identity
9696 or -floop-parallelize-all are given.
9697 * graphite-scop-detection.c: Include cfganal.h.
9698 (gather_bbs::gather_bbs): Get and pass through bb to RPO
9700 (gather_bbs::before_dom_children): Return STOP for BBs
9702 (build_scops): Compute bb to RPO mapping and pass it to
9703 the domwalk. Treat --param graphite-max-nb-scop-params=0
9704 as not limiting the number of params.
9705 * graphite.c (graphite_initialize): Remove limit on the
9706 number of basic-blocks in a function.
9707 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
9708 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
9709 default value of 10.
9711 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9713 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
9714 Adjust code to eliminate needing to do the shift right 32-bits
9715 operation after XSCVDPSPN.
9717 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9719 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
9720 ((X / Y) != 0 -> X >= Y): Likewise.
9722 2017-09-26 Carl Love <cel@us.ibm.com>
9724 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
9725 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
9726 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
9727 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
9728 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
9729 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
9730 definitions and overloading.
9731 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
9732 statement for P9V_BUILTIN_XST_LEN_R.
9733 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
9734 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
9735 define_expand and define_insn for the instructions and builtins.
9736 * doc/extend.texi: Update the built-in documentation file for the new
9738 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
9739 define_insn for the instructions
9741 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
9744 * gcc/config/netbsd-protos.h: New file.
9745 * gcc/config/netbsd.c: New file.
9746 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
9747 * gcc/config/t-netbsd: New file.
9748 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
9749 (tmake_file) Add t-netbsd.
9750 (extra_objs) Add netbsd.o.
9752 2017-09-26 Janus Weil <janus@gcc.gnu.org>
9756 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
9758 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9760 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
9761 sign extension from a vector register to a GPR by doing a 32-bit
9762 direct move and then an EXTSW.
9763 (extendsi<mode>2 splitter): Likewise.
9764 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
9765 right or vector extract after doing XSCVDPSPN. Use
9766 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
9768 (movdi_from_sf_zero_ext): Likewise.
9769 (reload_gpr_from_vsxsf): Likewise.
9770 (p8_mfvsrd_4_disf): Delete, no longer used.
9771 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
9772 then needing to move the SFmode to a GPR to use the XSCVDPSP
9773 instruction instead of FRSP and XSCVDPSPN.
9774 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
9775 it is adjacent to the other XSCVSPDP insns.
9776 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
9777 SFmode to be in traditional Altivec registers.
9778 (vsx_xscvdpspn): Eliminate useless alternative constraint.
9779 (vsx_xscvspdpn): Likewise.
9780 (vsx_xscvspdpn_scalar): Likewise.
9782 2017-09-26 Martin Jambor <mjambor@suse.cz>
9784 * tree-sra.c (compare_access_positions): Put integral types first,
9785 stabilize sorting of integral types, remove conditions putting
9786 non-full-precision integers last.
9787 (sort_and_splice_var_accesses): Disable scalarization if a
9788 non-integert would be represented by a non-full-precision integer.
9790 2017-09-26 Joseph Myers <joseph@codesourcery.com>
9792 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
9793 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
9794 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
9795 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
9796 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
9797 conditionals inside the function instead of around it. Call
9798 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
9799 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
9801 2017-09-26 Richard Biener <rguenther@suse.de>
9803 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
9805 (scop_detection::build_scop_breadth): ... this. Removed.
9806 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
9807 (scop_detection::harmful_stmt_in_bb): Likewise.
9808 (scop_detection::graphite_can_represent_stmt): Likewise.
9809 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
9810 (scop_detection::can_represent_loop): Remove recursion, fold in ...
9811 (scop_detection::can_represent_loop_1): ... this. Removed.
9812 (scop_detection::harmful_loop_in_region): Simplify after inlining
9813 the above and remove more quadraticness.
9814 (build_scops): Adjust.
9815 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
9818 2017-09-26 Jakub Jelinek <jakub@redhat.com>
9821 * config/i386/i386.c (ix86_print_operand_address_as): Only test
9822 REGNO (base) == SP_REG if base is a REG.
9825 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
9826 if it is different SSA_NAME.
9827 (optimize_range_tests_cmp_bitwise): New function.
9828 (optimize_range_tests): Call it.
9830 2017-09-26 Richard Biener <rguenther@suse.de>
9832 PR tree-optimization/82321
9833 * graphite.c (canonicalize_loop_closed_ssa): Properly check
9834 for the def being inside the loop.
9836 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9838 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
9840 * config/s390/s390-builtins.def: Fix constraint on op4.
9842 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9844 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
9845 independent expanders.
9846 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
9847 ("vec_ordered", "vec_unordered"): New expanders.
9849 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9851 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
9854 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9856 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
9857 vec_unpacks_lo_v16qi.
9858 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
9860 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9862 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
9863 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
9864 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
9866 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9868 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
9870 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
9872 ("*vec_slb<mode>"): New insn pattern.
9873 ("vec_shr_<mode>"): New expander.
9874 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
9875 and force the shift count operand to V16QImode.
9876 ("vec_srb<mode>"): Set shift count mode to V16QI.
9878 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9880 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
9881 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
9882 ("vec_widen_smult_hi_<mode>"): New expander definitions.
9884 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
9887 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
9889 2017-09-26 Richard Biener <rguenther@suse.de>
9891 PR tree-optimization/82320
9892 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
9895 2017-09-25 Jeff Law <law@redhat.com>
9897 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
9898 prototype for new argument.
9899 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
9900 mostly extracted from rs6000_emit_allocate_stack.
9901 (rs6000_emit_probe_stack_range_stack_clash): New function.
9902 (rs6000_emit_allocate_stack): Call
9903 rs6000_emit_probe_stack_range_stack_clash as needed.
9904 (rs6000_emit_probe_stack_range): Add additional argument
9905 to call to gen_probe_stack_range{si,di}.
9906 (output_probe_stack_range): New.
9907 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
9908 (output_probe_stack_range_stack_clash): New.
9909 (rs6000_emit_prologue): Emit notes into dump file as requested.
9910 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
9911 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
9912 Add additional operand and pass it to output_probe_stack_range.
9914 2017-09-25 Bin Cheng <bin.cheng@arm.com>
9916 PR tree-optimization/82163
9917 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
9918 (checking_verify_loop_closed_ssa): New parameter.
9919 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
9920 (check_loop_closed_ssa_stmt): Delete.
9921 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
9922 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
9923 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
9926 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
9928 * brig-builtins.def: Treat HSAIL barrier builtins as
9929 setjmp/longjump style functions.
9931 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9933 * target.def (constant_alignment): New hook.
9934 * defaults.h (CONSTANT_ALIGNMENT): Delete.
9935 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
9936 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
9937 * doc/tm.texi: Regenerate.
9938 * targhooks.h (default_constant_alignment): Declare.
9939 (constant_alignment_word_strings): Likewise.
9940 * targhooks.c (default_constant_alignment): New function.
9941 (constant_alignment_word_strings): Likewise.
9942 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
9943 instead of CONSTANT_ALIGNMENT.
9944 * varasm.c (align_variable, get_variable_align, build_constant_desc)
9945 (force_const_mem): Likewise.
9946 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
9947 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
9948 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
9949 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9950 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
9952 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
9953 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9954 constant_alignment_word_strings.
9955 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
9956 (CONSTANT_ALIGNMENT): Likewise.
9957 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9958 (arm_constant_alignment): New function.
9959 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
9960 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9961 constant_alignment_word_strings.
9962 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
9963 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9964 constant_alignment_word_strings.
9965 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
9966 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9967 (cris_constant_alignment): New function.
9968 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
9969 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9970 (epiphany_constant_alignment): New function.
9971 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
9972 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9973 constant_alignment_word_strings.
9974 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
9975 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9976 constant_alignment_word_strings.
9977 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
9978 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9979 constant_alignment_word_strings.
9980 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
9981 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
9982 * config/i386/i386.c (ix86_constant_alignment): Make static.
9983 Use the same interface as the target hook.
9984 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9985 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
9986 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9987 constant_alignment_word_strings.
9988 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
9989 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
9990 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9991 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
9992 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9993 constant_alignment_word_strings.
9994 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
9995 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9996 constant_alignment_word_strings.
9997 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
9998 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9999 constant_alignment_word_strings.
10000 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10001 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10003 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10004 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10005 * config/mips/mips.c (mips_constant_alignment): New function.
10006 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10007 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10008 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10009 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10010 (mmix_constant_alignment): Make static. Use the same interface
10011 as the target hook.
10012 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10013 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10014 constant_alignment_word_strings.
10015 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10016 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10017 constant_alignment_word_strings.
10018 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10019 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10020 constant_alignment_word_strings.
10021 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10022 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10023 (rs6000_constant_alignment): New function.
10024 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10025 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10026 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10027 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10028 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10029 (rs6000_constant_alignment): New function.
10030 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10031 * config/s390/s390.c (s390_constant_alignment): New function.
10032 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10033 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10034 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10035 constant_alignment_word_strings.
10036 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10037 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10038 (sparc_constant_alignment): New function.
10039 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10040 * config/spu/spu.c (spu_constant_alignment): New function.
10041 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10042 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10043 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10044 constant_alignment_word_strings.
10045 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10046 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10047 constant_alignment_word_strings.
10048 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10049 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10050 constant_alignment_word_strings.
10051 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10052 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10053 (visium_constant_alignment): New function.
10054 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10055 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10056 (xtensa_constant_alignment): New function.
10057 * system.h (CONSTANT_ALIGNMENT): Poison.
10059 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
10061 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10062 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
10063 (rs6000_builtin_valid_without_lhs): New helper function.
10064 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10065 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
10067 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10069 * target.h (vec_perm_indices): Use unsigned short rather than
10071 (auto_vec_perm_indices): Likewise.
10072 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
10073 Use unsigned int rather than unsigned char.
10074 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
10076 2017-09-25 Richard Biener <rguenther@suse.de>
10078 * cfgloop.h (sort_sibling_loops): Declare.
10079 * cfgloop.c (sort_sibling_loops_cmp): New helper.
10080 (sort_sibling_loops): New function sorting the sibling loop list
10082 * graphite.c (graphite_transform_loops): Sort sibling loops.
10084 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
10086 * target.def (vec_perm_const_ok): Change sel parameter to
10088 * optabs-query.c (can_vec_perm_p): Update accordingly.
10089 * doc/tm.texi: Regenerate.
10090 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
10091 auto_vec_perm_indices and remove separate nelt field.
10092 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
10093 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
10094 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
10095 (aarch64_expand_vec_perm_const): Update accordingly.
10096 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
10097 to vec_perm_indices.
10098 * config/arm/arm.c (expand_vec_perm_d): Change perm to
10099 auto_vec_perm_indices and remove separate nelt field.
10100 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
10101 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
10102 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
10104 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
10105 to vec_perm_indices.
10106 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
10107 sel to vec_perm_indices.
10108 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
10109 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
10110 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
10112 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
10115 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
10118 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
10119 on the FUNCTION_DECL function context if it has a DIE that is a
10122 2017-09-25 Richard Biener <rguenther@suse.de>
10124 PR tree-optimization/82285
10125 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
10128 2017-09-25 Tom de Vries <tom@codesourcery.com>
10132 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
10135 2017-09-25 Richard Biener <rguenther@suse.de>
10137 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
10138 ISL errors other than isl_error_quota happen. Dump if the
10139 schedule is the same.
10140 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
10141 errors instead of aborting inside ISL.
10143 2017-09-25 Iain Sandoe <iain@codesourcery.com>
10146 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
10147 of libgcc_eh for m64.
10148 * config/i386/darwin64.h: Likewise.
10150 2017-09-25 Richard Biener <rguenther@suse.de>
10152 PR middle-end/82144
10153 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
10154 attribute for incomplete types nor twice for complete ones.
10156 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
10159 * config/i386/i386.c (ix86_print_operand_address_as): Encode
10160 %esp as %rsp to avoid 0x67 prefix if there is no index or base
10163 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
10166 * config/i386/i386.opt (mprefer-avx256): Use
10167 ix86_target_flags variable.
10168 * config/i386/i386.c (ix86_target_string): Move
10169 -mprefer-avx256 to flag2_opts.
10171 2017-09-22 Jakub Jelinek <jakub@redhat.com>
10173 PR middle-end/35691
10174 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
10175 and x != -1 | y != -1 into (x & y) != -1.
10177 2017-09-22 Steve Ellcey <sellcey@cavium.com>
10179 * config.gcc: Add new case statement to set
10180 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
10181 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
10182 s390x-*-linux* case statements. Added aarch64 to the list of
10183 supported architectures.
10185 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10187 PR tree-optimization/82289
10188 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
10189 STMT_VINFO_RELEVANT_P.
10191 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10192 Alan Hayward <alan.hayward@arm.com>
10193 David Sherwood <david.sherwood@arm.com>
10195 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
10196 for VR_RANGE only; don't allow VR_ANTI_RANGE.
10197 (extract_range_from_binary_expr_1): Don't call
10198 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
10200 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10201 Alan Hayward <alan.hayward@arm.com>
10202 David Sherwood <david.sherwood@arm.com>
10204 * target.def (preferred_vector_alignment): New hook.
10205 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
10207 * doc/tm.texi: Regenerate.
10208 * targhooks.h (default_preferred_vector_alignment): Declare.
10209 * targhooks.c (default_preferred_vector_alignment): New function.
10210 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
10212 (DR_TARGET_ALIGNMENT): New macro.
10213 (aligned_access_p): Update commentary.
10214 (vect_known_alignment_in_bytes): New function.
10215 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
10217 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
10218 Calculate the misalignment based on the target alignment rather than
10220 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
10221 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
10222 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
10223 the target alignment, rather than masking the element misalignment
10224 with the number of elements in a vector. Also use the target
10225 alignment when calculating the maximum number of peels.
10226 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
10227 instead of TYPE_ALIGN_UNIT.
10228 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
10229 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
10230 (vect_create_addr_base_for_vector_ref): Update call accordingly.
10231 (vect_create_data_ref_ptr): Likewise.
10232 (vect_setup_realignment): Realign by ANDing with
10233 -DR_TARGET_MISALIGNMENT.
10234 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
10235 the number of peels based on DR_TARGET_ALIGNMENT.
10236 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
10237 with the guaranteed alignment boundary when deciding whether
10239 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
10240 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
10241 (ensure_base_align): Remove stmt_info parameter. Get the
10242 target base alignment from DR_TARGET_ALIGNMENT.
10243 (vectorizable_store): Update call accordingly. Interpret
10244 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
10246 (vectorizable_load): Likewise.
10248 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10249 Alan Hayward <alan.hayward@arm.com>
10250 David Sherwood <david.sherwood@arm.com>
10252 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
10253 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
10254 (vect_enhance_data_refs_alignment): Likewise.
10256 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10258 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10259 error. Only quit immediately if parsing is complete.
10260 (BEGIN): Initialize fatal_err and parse_done.
10261 (begin fpu, end fpu): Check number of arguments.
10262 (begin arch, end arch): Likewise.
10263 (begin cpu, end cpu): Likewise.
10264 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10265 (optalias): Likewise.
10267 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10269 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10270 * config/arm/arm-isa.h: Delete. Move definitions to ...
10271 * arm-cpus.in: ... here. Use new feature and fgroup values.
10272 * config/arm/arm.c (arm_option_override): Use lower case for feature
10274 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10275 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10276 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10278 (print_isa_bits_for): New function.
10279 (gen_isa): New function.
10280 (gen_comm_data): Use print_isa_bits_for.
10281 (define feature): New keyword.
10282 (define fgroup): New keyword.
10283 * config/arm/t-arm (TM_H): Remove.
10284 (GTM_H): Add arm-isa.h.
10285 (arm-isa.h): Add rule to generate file.
10286 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10287 case for feature bit names.
10289 2017-09-22 Richard Biener <rguenther@suse.de>
10291 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
10293 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
10294 print of no dependency loops ...
10295 * graphite.c (graphite_transform_loops): ... here.
10296 (canonicalize_loop_closed_ssa_form): Work from inner to outer
10298 (same_close_phi_node, remove_duplicate_close_phi,
10299 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
10300 (canonicalize_loop_closed_ssa): ... here and simplify.
10301 * graphite-optimize-isl.c: Include tree-vectorizer.h.
10302 (optimize_isl): Use dump_printf_loc to tell when we stopped
10303 optimizing because of an ISL timeout.
10305 2017-09-22 Richard Biener <rguenther@suse.de>
10307 PR tree-optimization/82291
10308 * tree-if-conv.c (predicate_mem_writes): Make sure to
10309 remove writes in blocks predicated with false.
10311 2017-09-22 Richard Biener <rguenther@suse.de>
10313 * sese.c: Include cfganal.h.
10314 (if_region_set_false_region): Remove.
10315 (create_if_region_on_edge): Likewise.
10316 (move_sese_in_condition): Re-implement without destroying
10319 2017-09-22 Richard Biener <rguenther@suse.de>
10321 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
10322 Verify both BBs contain loop PHI nodes before dispatching to
10323 copy_loop_phi_args.
10324 (graphite_regenerate_ast_isl): Do not recompute dominators,
10325 do not verify three times. Restructure for clarity.
10326 * graphite-scop-detection.c (same_close_phi_node,
10327 remove_duplicate_close_phi, make_close_phi_nodes_unique,
10328 defined_in_loop_p, canonicalize_loop_closed_ssa,
10329 canonicalize_loop_closed_ssa_form): Simplify, remove excess
10330 checking and SSA rewrite, move to ...
10331 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
10332 (graphite_initialize): Do not pass in ctx, do not reset the
10333 SCEV cache, compute only dominators.
10334 (graphite_transform_loops): Allocate ISL ctx after
10335 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
10336 Maintain post-dominators only around build_scops.
10337 * sese.c (if_region_set_false_region): Make static. Free
10338 and recompute dominators.
10339 (move_sese_in_condition): Assert we don't get called with
10340 post-dominators computed.
10341 * sese.h (if_region_set_false_region): Remove.
10343 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
10345 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
10346 mode attribute for TARGET_AVX512VL.
10348 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
10350 * config/i386/i386.opt (mprefer-avx256): New option.
10351 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
10353 (ix86_preferred_simd_mode): Return 256-bit AVX modes
10354 for TARGET_PREFER_AVX256.
10355 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
10357 2017-09-21 Jeff Law <law@redhat.com>
10359 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
10360 Fix dump output if the only stack space is for pushed registers.
10362 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10364 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
10367 2017-09-21 Martin Sebor <msebor@redhat.com>
10370 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
10371 code (in C++) or code that triggers warnings.
10373 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
10375 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
10377 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
10379 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
10380 * sched-rgn.c: Ditto.
10381 * sel-sched-ir.c: Ditto.
10383 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
10385 * toplev.h (set_random_seed): Adjust return type.
10386 * toplev.c (init_local_tick): Move eager initialization of random_seed
10387 to get_random_seed. Adjust comment.
10388 (init_random_seed): Inline to get_random_seed, delete.
10389 (get_random_seed): Initialize random_seed lazily.
10390 (set_random_seed): Do not return previous value.
10391 (print_switch_value): Do not call get_random_seed.
10393 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
10395 * cgraph.c (delete_function_version): New, broken out from...
10396 (cgraph_node::delete_function_version): ...here. Rename to
10397 cgraph_node::delete_function_version_by_decl. Update all uses.
10398 (cgraph_node::remove): Call delete_function_version.
10400 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10403 * tree-inline.c (expand_call_inline): Emit clobber stmts for
10404 VAR_DECLs to which addressable non-volatile parameters are mapped
10405 and for id->retvar after the return value assignment. Clear
10406 id->retval and id->retbnd after inlining.
10408 2017-09-21 Richard Biener <rguenther@suse.de>
10410 PR tree-optimization/82276
10411 PR tree-optimization/82244
10412 * tree-vrp.c (build_assert_expr_for): Set
10413 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
10415 (remove_range_assertions): Revert earlier change.
10417 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
10420 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
10422 2017-09-21 Richard Biener <rguenther@suse.de>
10424 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10425 Restore valid IL after code generation errors.
10426 * graphite.c (graphite_transform_loops): Diagnose code
10427 generation issues as MSG_MISSED_OPTIMIZATION and continue
10428 with processing SCOPs.
10430 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10431 Alan Hayward <alan.hayward@arm.com>
10432 David Sherwood <david.sherwood@arm.com>
10434 * calls.c (compute_argument_addresses): Use simplify_gen_binary
10435 rather than choosing between plus_constant and gen_rtx_<CODE>.
10436 * expr.c (emit_push_insn): Likewise.
10437 (expand_expr_real_2): Likewise.
10439 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10440 Alan Hayward <alan.hayward@arm.com>
10441 David Sherwood <david.sherwood@arm.com>
10443 * loop-unroll.c (split_iv): Call copy_rtx on the step.
10445 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10446 Alan Hayward <alan.hayward@arm.com>
10447 David Sherwood <david.sherwood@arm.com>
10449 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
10450 calling tree_to_uhwi.
10452 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10453 Alan Hayward <alan.hayward@arm.com>
10454 David Sherwood <david.sherwood@arm.com>
10456 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
10457 INTEGER_CST rather than a negative test for ADDR_EXPR.
10459 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10460 Alan Hayward <alan.hayward@arm.com>
10461 David Sherwood <david.sherwood@arm.com>
10463 * tree-vrp.c (extract_range_from_binary_expr_1): Check
10464 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
10466 2017-09-21 Richard Biener <rguenther@suse.de>
10468 PR tree-optimization/71351
10469 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
10470 graphite_create_new_loop_guard): Remove, fold remaining parts
10472 (translate_isl_ast_node_for): ... here and simplify.
10474 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10477 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
10478 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
10479 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
10480 alternative always use QI mode, for -Os imov (=R,R) alternative
10481 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
10484 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10485 Jeff Law <law@redhat.com>
10487 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
10488 (allocate_stack_space): New function, partially extracted from
10489 s390_emit_prologue.
10490 (s390_emit_prologue): Track offset to most recent stack probe.
10491 Code to allocate space moved into allocate_stack_space.
10492 Dump actions when no stack is allocated.
10493 (s390_prologue_plus_offset): New function.
10494 (s390_emit_stack_probe): Likewise.
10496 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
10498 * common.opt (Wa, Wl, Wp, g, gz=): Add
10500 (gno-column-info): Remove.
10501 (gcolumn-info): Drop RejectNegative.
10502 (gno-): New prefix.
10503 (gno-record-gcc-switches): Remove.
10504 (grecord-gcc-switches): Drop RejectNegative.
10505 (gno-split-dwarf): Remove.
10506 (gsplit-dwarf): Drop RejectNegative.
10507 (gno-strict-dwarf): Remove.
10508 (gstrict-dwarf): Drop RejectNegative.
10509 * config/darwin.opt (gfull, gused): Add RejectNegative.
10510 * dwarf2out.c (gen_producer_string): Drop
10511 gno-record-gcc-switches handler.
10512 * optc-gen.awk: Add g to prefixes with negative forms.
10513 * opts-common.c (remapping_prefix_p): New.
10514 (find_opt): Check it.
10515 (generate_canonical_option): Test g prefix.
10516 (option_map): Add -gno- mapping.
10517 (add_misspelling_candidates): Check remapping_prefix_p.
10519 2017-09-20 Jeff Law <law@redhat.com>
10521 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
10522 thinko in stack clash protection support.
10524 * explow.c (compute_stack_clash_protection_loop_data): Use
10525 CONST_INT_P instead of explicit test. Verify object is a
10526 CONST_INT_P before looking at INTVAL.
10527 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
10528 instead of explicit test.
10530 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
10533 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
10534 address instead of to r1 and r11.
10536 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
10538 * config.gcc: Support "knm".
10539 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
10540 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
10542 * config/i386/i386.c (m_KNM): Define.
10543 (processor_target_table): Add "knm".
10545 (ix86_option_override_internal): Add "knm".
10546 (ix86_issue_rate): Add PROCESSOR_KNM.
10547 (ix86_adjust_cost): Ditto.
10548 (ia32_multipass_dfa_lookahead): Ditto.
10549 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
10550 (fold_builtin_cpu): Add M_INTEL_KNM.
10551 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
10552 (processor_type): Add PROCESSOR_KNM.
10553 * config/i386/x86-tune.def: Add m_KNM.
10554 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
10556 2017-09-20 Richard Biener <rguenther@suse.de>
10558 PR tree-optimization/80213
10559 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
10560 are allowed in empty BBs as well.
10561 (canonicalize_loop_closed_ssa): Also look for other complex
10563 (scop_detection::get_sese): Include the loop-closed PHI block
10565 (scop_detection::merge_sese): Remove code adding extra blocks.
10566 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
10567 (build_scops): Assert the final returned scop is invalid.
10569 2017-09-20 Richard Biener <rguenther@suse.de>
10571 PR tree-optimization/82264
10572 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
10574 (vn_phi_lookup): Likewise.
10575 (vn_phi_insert): Likewise.
10577 2017-09-20 Jakub Jelinek <jakub@redhat.com>
10579 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
10580 that fits into uhwi or shwi, add DW_AT_const_value regardless
10581 of early_dwarf without going through RTL, using add_AT_unsigned
10584 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
10585 (DEBUG_LTO_ABBREV_SECTION): Likewise.
10586 (DEBUG_LTO_MACINFO_SECTION): Likewise.
10587 (DEBUG_MACRO_SECTION): Likewise.
10588 (DEBUG_LTO_MACRO_SECTION): Likewise.
10589 (DEBUG_STR_DWO_SECTION): Likewise.
10590 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
10591 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
10592 (DEBUG_LTO_DWO_LINE_SECTION): Define.
10593 (DEBUG_LTO_LINE_STR_SECTION): Define.
10594 (init_sections_and_labels): Initialize debug_line_str_section
10595 variable. Initialize debug_loc_section for -gdwarf-5 to
10596 DEBUG_LOCLISTS_SECTION. Formatting fixes.
10598 2017-09-20 Richard Biener <rguenther@suse.de>
10600 * graphite-sese-to-poly.c (extract_affine): Properly handle
10601 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
10603 2017-09-20 Richard Biener <rguenther@suse.de>
10605 PR tree-optimization/81373
10606 * graphite-scop-detection.c (build_cross_bb_scalars_def):
10607 Force SESE live-out defs to be handled even if they are
10610 2017-09-19 Jeff Law <law@redhat.com>
10612 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
10613 nothing for stack adjustments with REG_STACK_CHECK.
10614 * sched-deps.c (parse_add_or_inc): Reject insns with
10615 REG_STACK_CHECK from dependency breaking.
10616 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
10617 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
10618 * reg-notes.def (STACK_CHECK): New note.
10620 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
10621 (ix86_expand_prologue): Dump stack clash info as needed.
10622 Call ix86_adjust_stack_and_probe_stack_clash as needed.
10624 * function.c (dump_stack_clash_frame_info): New function.
10625 * function.h (dump_stack_clash_frame_info): Prototype.
10626 (enum stack_clash_probes): New enum.
10628 * config/alpha/alpha.c (alpha_expand_prologue): Also check
10629 flag_stack_clash_protection.
10630 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
10631 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
10632 (arm_frame_pointer_required): Likewise.
10633 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
10634 (ia64_expand_prologue): Likewise.
10635 * config/mips/mips.c (mips_expand_prologue): Likewise.
10636 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
10637 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10638 (sparc_flat_expand_prologue): Likewise.
10639 * config/spu/spu.c (spu_expand_prologue): Likewise.
10641 * explow.c: Include "params.h".
10642 (anti_adjust_stack_and_probe_stack_clash): New function.
10643 (get_stack_check_protect): Likewise.
10644 (compute_stack_clash_protection_loop_data): Likewise.
10645 (emit_stack_clash_protection_loop_start): Likewise.
10646 (emit_stack_clash_protection_loop_end): Likewise.
10647 (allocate_dynamic_stack_space): Use get_stack_check_protect.
10648 Use anti_adjust_stack_and_probe_stack_clash.
10649 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
10650 (emit_stack_clash_protection_loop_start): Likewise.
10651 (emit_stack_clash_protection_loop_end): Likewise.
10652 * rtl.h (get_stack_check_protect): Prototype.
10653 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
10654 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
10655 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
10657 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
10659 * doc/tm.texi: Rebuilt.
10660 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
10661 get_stack_check_protect.
10662 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
10663 * config/arm/arm.c (arm_expand_prologue): Likewise.
10664 (arm_frame_pointer_required): Likewise.
10665 * config/i386/i386.c (ix86_expand_prologue): Likewise.
10666 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
10667 * config/mips/mips.c (mips_expand_prologue): Likewise.
10668 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
10669 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
10670 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10671 (sparc_flat_expand_prologue): Likewise.
10673 * common.opt (-fstack-clash-protection): New option.
10674 * flag-types.h (enum stack_check_type): Note difference between
10675 -fstack-check= and -fstack-clash-protection.
10676 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
10677 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
10678 * toplev.c (process_options): Issue warnings/errors for cases
10679 not handled with -fstack-clash-protection.
10680 * doc/invoke.texi (-fstack-clash-protection): Document new option.
10681 (-fstack-check): Note additional problem with -fstack-check=generic.
10682 Note that -fstack-check is primarily for Ada and refer users
10683 to -fstack-clash-protection for stack-clash-protection.
10684 Document new params for stack clash protection.
10686 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
10688 * config/i386/i386.c (ix86_split_long_move): Do not handle
10689 address used for LEA in a special way.
10691 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
10693 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
10695 2017-09-19 Martin Sebor <msebor@redhat.com>
10698 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
10699 of incompatible types.
10701 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
10703 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10704 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
10705 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10706 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
10708 2017-09-19 Richard Biener <rguenther@suse.de>
10710 PR tree-optimization/82244
10711 * tree-vrp.c (remove_range_assertions): Do not propagate
10712 a constant to abnormals but replace the assert with a copy.
10714 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
10716 PR rtl-optimization/57878
10717 PR rtl-optimization/68988
10718 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
10719 avoidance test involving non_reload_pseudos. Move frequency test
10720 below the general fragmentation avoidance test.
10722 2017-09-19 Richard Biener <rguenther@suse.de>
10724 PR tree-optimization/69728
10725 * graphite-sese-to-poly.c (schedule_error): New global.
10726 (add_loop_schedule): Handle empty domain by failing the
10728 (build_original_schedule): Handle schedule_error.
10730 2017-09-19 Richard Biener <rguenther@suse.de>
10732 * graphite-scop-detection.c (scop_detection::can_represent_loop):
10733 Do not iterate to sibling loops but only to siblings of inner
10736 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
10739 * config/m68k/m68k.md (moveq feeding equality comparison): Check
10740 that the registers are different.
10742 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
10744 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
10745 to processor_model and "amdfam17h" to arch_names_table.
10746 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
10748 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10751 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
10753 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10754 Alan Hayward <alan.hayward@arm.com>
10755 David Sherwood <david.sherwood@arm.com>
10757 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
10759 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
10761 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
10762 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
10763 vect_schedule_slp_instance.
10764 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
10765 Update call to vect_slp_analyze_node_operations. Simplify return
10767 (vect_slp_analyze_bb_1): Update call accordingly.
10768 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
10769 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
10770 (vect_schedule_slp): Update call accordingly.
10772 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10773 Alan Hayward <alan.hayward@arm.com>
10774 David Sherwood <david.sherwood@arm.com>
10776 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
10777 with types that aren't in fact scalar.
10779 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10781 * tree-vect-slp.c (vect_record_max_nunits): New function,
10783 (vect_build_slp_tree_1): ...here.
10784 (vect_build_slp_tree_2): Call it for phis too.
10786 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10788 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
10789 to vect_get_vec_def_for_operand when getting the mask operand.
10791 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10792 Alan Hayward <alan.hayward@arm.com>
10793 David Sherwood <david.sherwood@arm.com>
10795 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
10798 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10799 Alan Hayward <alan.hayward@arm.com>
10800 David Sherwood <david.sherwood@arm.com>
10802 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
10803 calculation for vector booleans.
10805 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10806 Alan Hayward <alan.hayward@arm.com>
10807 David Sherwood <david.sherwood@arm.com>
10809 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
10811 (vect_transform_stmt): ...here.
10812 (vect_analyze_stmt): Use it instead of calling
10813 vectorizable_live_operation directly.
10815 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
10817 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
10818 non-SIMT targets in acc vector loops.
10820 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
10822 * configure.ac: Add arc and check if assembler supports gdwarf2.
10823 * configure: Regenerate.
10825 2017-09-18 Richard Biener <rguenther@suse.de>
10827 PR tree-optimization/82220
10828 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
10829 epilogue niters from the min_profitable_iters compute.
10831 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10834 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
10835 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
10836 (ix86_init_pic_reg): Revert 2017-09-01 changes.
10838 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
10841 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
10842 switching to a new text section.
10844 2017-09-18 Richard Biener <rguenther@suse.de>
10846 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
10848 (build_alias_set): Reject aliases with no access function.
10850 2017-09-18 Richard Biener <rguenther@suse.de>
10852 PR tree-optimization/79622
10853 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
10855 (build_cross_bb_scalars_use): Likewise.
10857 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
10859 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
10861 2017-09-18 Alan Modra <amodra@gmail.com>
10864 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
10865 stack_pointer_rtx for count 0. Update comments. Break up
10866 large rtl expression.
10868 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
10870 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
10871 Increase to 20 bytes.
10872 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
10873 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
10874 or avx version of the stub.
10876 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
10879 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
10880 compute the minimum stack alignment. Also update preferred stack
10881 boundary for leaf functions.
10883 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
10885 PR tree-optimization/82228
10886 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
10889 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10891 * common/config/nds32/nds32-common.c
10892 (nds32_option_optimization_table): Refine formatting.
10893 (nds32_option_optimization_table): Use -fsched-pressure and
10894 -fomit-frame-pointer for specific optimization level.
10896 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10898 * config/nds32/nds32.c: Refine formatting and comments.
10899 * config/nds32/nds32.h: Likewise.
10900 * config/nds32/nds32.md: Likewise.
10901 * config/nds32/nds32-cost.c: Likewise.
10902 * config/nds32/nds32-isr.c: Likewise.
10903 * config/nds32/nds32-md-auxiliary.c: Likewise.
10904 * config/nds32/nds32-multiple.md: Likewise.
10905 * config/nds32/nds32-predicates.c: Likewise.
10907 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
10908 Jakub Jelinek <jakub@redhat.com>
10910 Add support for -std=c++2a.
10911 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
10913 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
10915 2017-09-15 Steve Ellcey <sellcey@cavium.com>
10918 * doc/extend.texi (Common Function Attributes): Add
10919 references to ARM, AArch64, and S/390 specific attributes.
10920 (Function Specific Option Pragmas): Add AArch64 and S/390
10921 to list of back ends that support the target pragma.
10923 2017-09-15 Nathan Sidwell <nathan@acm.org>
10925 * doc/standards.texi: Fix C++17 description. Update URLs for
10928 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
10930 * common.opt (Wcast-align=strict): New warning option.
10931 * doc/invoke.texi: Document -Wcast-align=strict.
10933 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
10935 * cgraph.h (cgraph_thunk_info): Add comments.
10936 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
10937 assert for VIRTUAL_* arguments stricter.
10939 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
10941 PR tree-optimization/71026
10942 * match.pd: Move RDIV patterns from fold-const.c
10943 * fold-const.c (distribute_real_division): Removed.
10944 (fold_binary_loc): Remove calls to distribute_real_divison.
10946 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10948 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
10949 c++1z and gnu++1z as deprecated. Change other references to
10950 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
10951 Change -Wc++1z-compat to -Wc++17-compat.
10952 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
10953 * dwarf2out.c (highest_c_language): Handle C++17.
10954 (gen_compile_unit_die): Likewise.
10956 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10958 PR rtl-optimization/82192
10959 * combine.c (make_extraction): Don't look through non-paradoxical
10960 SUBREGs or TRUNCATE if pos + len is or might be bigger than
10963 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10964 Alan Hayward <alan.hayward@arm.com>
10965 David Sherwood <david.sherwood@arm.com>
10967 * target.def (function_arg_offset): New hook.
10968 * targhooks.h (default_function_arg_offset): Declare.
10969 * targhooks.c (default_function_arg_offset): New function.
10970 * function.c (locate_and_pad_parm): Use
10971 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
10972 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
10973 (TARGET_FUNCTION_ARG_OFFSET): ...this.
10974 * doc/tm.texi: Regenerate.
10975 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
10976 * config/spu/spu.c (spu_function_arg_offset): New function.
10977 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
10978 * system.h (FUNCTION_ARG_OFFSET): Poison.
10980 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10981 Alan Hayard <alan.hayward@arm.com>
10982 David Sherwood <david.sherwood@arm.com>
10984 * target.def (truly_noop_truncation): New hook.
10985 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
10986 than TRULY_NOOP_TRUNCATION.
10987 * hooks.h (hook_bool_uint_uint_true): Declare.
10988 * hooks.c (hook_bool_uint_uint_true): New function.
10989 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
10990 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
10991 * doc/tm.texi: Regenerate.
10992 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
10993 rather than TRULY_NOOP_TRUNCATION in comments.
10994 (simplify_comparison): Likewise.
10995 (record_truncated_value): Likewise.
10996 * expmed.c (extract_bit_field_1): Likewise.
10997 (extract_split_bit_field): Likewise.
10998 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
10999 instead of TRULY_NOOP_TRUNCATION.
11000 * function.c (assign_parm_setup_block): Likewise.
11001 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11002 * rtlhooks.c: Include target.h.
11003 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11004 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11005 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11006 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11007 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11008 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11009 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11010 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11011 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11012 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11013 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11014 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11015 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11016 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11017 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11018 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11019 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11020 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11021 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11022 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11023 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11024 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11025 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11026 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11027 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11028 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11029 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11030 rather than TRULY_NOOP_TRUNCATION in comments.
11031 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11032 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11033 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11034 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11035 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11036 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11037 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11038 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11039 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11040 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11041 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11042 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11043 rather than TRULY_NOOP_TRUNCATION in comments.
11044 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11045 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11046 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11047 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11048 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11049 TRULY_NOOP_TRUNCATION condition.
11050 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11051 (TRULY_NOOP_TRUNCATION): Delete.
11052 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11053 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11054 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11055 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11056 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11057 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11058 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11059 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11060 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11061 rather than TRULY_NOOP_TRUNCATION in comments.
11062 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
11063 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
11064 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
11065 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
11066 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
11067 * system.h (TRULY_NOOP_TRUNCATION): Poison.
11069 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
11072 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
11073 (*cmp_ior): Likewise.
11074 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
11075 (*ior_scc_scc_cmp): Likewise.
11076 (*and_scc_scc): Likewise.
11077 (*and_scc_scc_cmp): Likewise.
11079 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11080 Alan Hayard <alan.hayward@arm.com>
11081 David Sherwood <david.sherwood@arm.com>
11083 * target.def (can_change_mode_class): New hook.
11084 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
11085 (hard_regno_nregs): Likewise.
11086 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
11087 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
11088 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
11089 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
11090 (LOAD_EXTEND_OP): Update accordingly.
11091 * doc/tm.texi: Regenerate.
11092 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
11093 CANNOT_CHANGE_MODE_CLASS.
11094 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
11095 (REG_CAN_CHANGE_MODE_P): ...this new macro.
11096 * combine.c (simplify_set): Update accordingly.
11097 * emit-rtl.c (validate_subreg): Likewise.
11098 * recog.c (general_operand): Likewise.
11099 * regcprop.c (mode_change_ok): Likewise.
11100 * reload1.c (choose_reload_regs): Likewise.
11101 (inherit_piecemeal_p): Likewise.
11102 * rtlanal.c (simplify_subreg_regno): Likewise.
11103 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
11104 instead of CANNOT_CHANGE_MODE_CLASS.
11105 (reload_cse_simplify_operands): Likewise.
11106 * reload.c (push_reload): Use targetm.can_change_mode_class
11107 instead of CANNOT_CHANGE_MODE_CLASS.
11108 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
11109 REG_CANNOT_CHANGE_MODE_P.
11110 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11111 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
11112 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11113 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11114 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11115 (arm_can_change_mode_class): New function.
11116 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11117 than CANNOT_CHANGE_MODE_CLASS in comments.
11118 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11119 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
11120 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
11121 (ix86_can_change_mode_class): ...this new function, inverting the
11122 sense of the return value.
11123 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11124 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11125 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11126 (ia64_can_change_mode_class): New function.
11127 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11128 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
11129 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
11130 (m32c_can_change_mode_class): ...this new function, inverting the
11131 sense of the return value.
11132 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11133 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11134 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
11135 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
11136 (mips_can_change_mode_class): ...this new function, inverting the
11137 sense of the return value.
11138 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11139 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11140 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11141 (msp430_can_change_mode_class): New function.
11142 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11143 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
11144 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11145 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11146 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11147 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
11148 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11149 (pa_cannot_change_mode_class): Replace with...
11150 (pa_can_change_mode_class): ...this new function, inverting the
11151 sense of the return value.
11152 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11153 than CANNOT_CHANGE_MODE_CLASS in comments.
11154 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11155 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
11156 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11157 (pdp11_cannot_change_mode_class): Replace with...
11158 (pdp11_can_change_mode_class): ...this new function, inverting the
11159 sense of the return value.
11160 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11161 * config/powerpcspe/powerpcspe-protos.h
11162 (rs6000_cannot_change_mode_class_ptr): Delete.
11163 * config/powerpcspe/powerpcspe.c
11164 (rs6000_cannot_change_mode_class_ptr): Delete.
11165 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11166 (rs6000_option_override_internal): Assign to
11167 targetm.can_change_mode_class instead of
11168 rs6000_cannot_change_mode_class_ptr.
11169 (rs6000_cannot_change_mode_class): Replace with...
11170 (rs6000_can_change_mode_class): ...this new function, inverting the
11171 sense of the return value.
11172 (rs6000_debug_cannot_change_mode_class): Replace with...
11173 (rs6000_debug_can_change_mode_class): ...this new function.
11174 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11175 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
11176 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11177 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11178 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
11180 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
11181 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11182 (rs6000_option_override_internal): Assign to
11183 targetm.can_change_mode_class instead of
11184 rs6000_cannot_change_mode_class_ptr.
11185 (rs6000_cannot_change_mode_class): Replace with...
11186 (rs6000_can_change_mode_class): ...this new function, inverting the
11187 sense of the return value.
11188 (rs6000_debug_cannot_change_mode_class): Replace with...
11189 (rs6000_debug_can_change_mode_class): ...this new function.
11190 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11191 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
11192 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
11193 (s390_can_change_mode_class): ...this new function, inverting the
11194 sense of the return value.
11195 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11196 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11197 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
11198 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11199 (sh_cannot_change_mode_class): Replace with...
11200 (sh_can_change_mode_class): ...this new function, inverting the
11201 sense of the return value.
11202 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11203 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11204 (sparc_can_change_mode_class): New function.
11205 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11206 * config/spu/spu.c (spu_can_change_mode_class): New function.
11207 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11208 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11209 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11210 (visium_can_change_mode_class): New function.
11211 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
11213 2017-09-15 Richard Biener <rguenther@suse.de>
11215 PR tree-optimization/82217
11216 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
11217 but not undefined case.
11219 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11222 * postreload.c (reload_cse_simplify_operands): Skip
11223 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
11225 2017-09-15 Richard Biener <rguenther@suse.de>
11227 PR tree-optimization/68823
11228 * graphite-scop-detection.c (build_alias_set): If we have a
11229 possible dependence check whether we can handle them by just
11230 looking at the DRs DR_ACCESS_FNs.
11231 (build_scops): If build_alias_set fails, fail the SCOP.
11233 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11235 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
11236 to support float128 built-in functions that require the ISA 3.0
11238 (BU_FLOAT128_3_HW): Likewise.
11239 (SQRTF128): Add support for the IEEE 128-bit square root and fma
11240 built-in functions.
11241 (FMAF128): Likewise.
11243 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
11244 support for built-in functions that need the ISA 3.0 IEEE 128-bit
11245 floating point instructions.
11246 (rs6000_invalid_builtin): Likewise.
11247 (rs6000_builtin_mask_names): Likewise.
11248 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
11249 (RS6000_BTM_FLOAT128_HW): Likewise.
11250 (RS6000_BTM_COMMON): Likewise.
11251 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
11253 * doc/extend.texi (RS/6000 built-in functions): Document the
11254 IEEE 128-bit floating point square root and fused multiply-add
11255 built-in functions.
11257 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
11259 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
11260 reg (r2) isn't in the set of registers defined in the prologue.
11262 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11263 Alan Hayward <alan.hayward@arm.com>
11264 David Sherwood <david.sherwood@arm.com>
11266 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
11267 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
11268 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
11269 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
11270 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
11272 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
11273 max_vectorization_factor.
11274 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
11276 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11277 Alan Hayward <alan.hayward@arm.com>
11278 David Sherwood <david.sherwood@arm.com>
11280 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
11281 (vect_worthwhile_without_simd_p): Declare.
11282 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
11283 (vectorizable_reduction): Use it.
11284 * tree-vect-stmts.c (vectorizable_shift): Likewise.
11285 (vectorizable_operation): Likewise.
11287 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11288 Alan Hayward <alan.hayward@arm.com>
11289 David Sherwood <david.sherwood@arm.com>
11291 * tree-vectorizer.h (vect_get_num_copies): New function.
11292 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
11293 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11294 (vectorizable_induction): Likewise.
11295 (vectorizable_live_operation): Likewise.
11296 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11297 (vectorizable_bswap): Likewise.
11298 (vectorizable_call): Likewise.
11299 (vectorizable_conversion): Likewise.
11300 (vectorizable_assignment): Likewise.
11301 (vectorizable_shift): Likewise.
11302 (vectorizable_operation): Likewise.
11303 (vectorizable_store): Likewise.
11304 (vectorizable_load): Likewise.
11305 (vectorizable_condition): Likewise.
11306 (vectorizable_comparison): Likewise.
11307 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
11309 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11310 Alan Hayward <alan.hayward@arm.com>
11311 David Sherwood <david.sherwood@arm.com>
11313 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
11314 of vect_init_vector.
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 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
11321 an inline wrapper that provides a location.
11322 (gimple_build_vector): Likewise.
11323 * gimple-fold.c (gimple_build_vector_from_val): New function.
11324 (gimple_build_vector): Likewise.
11325 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
11326 functions to build the initial value. Always return a gimple value.
11327 (get_initial_defs_for_reduction): Likewise. Only compute
11329 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
11330 vect_init_vector on the results from get_initial_def(s)_for_reduction.
11331 (vectorizable_induction): Use gimple_build_vector rather than
11334 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11335 Alan Hayward <alan.hayward@arm.com>
11336 David Sherwood <david.sherwood@arm.com>
11338 * target.h (vec_perm_indices): New typedef.
11339 (auto_vec_perm_indices): Likewise.
11340 * optabs-query.h: Include target.h
11341 (can_vec_perm_p): Take a vec_perm_indices *.
11342 * optabs-query.c (can_vec_perm_p): Likewise.
11343 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
11344 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11345 * tree-vect-generic.c (lower_vec_perm): Likewise.
11346 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
11347 (vect_grouped_load_supported): Likewise.
11348 (vect_shift_permute_load_chain): Likewise.
11349 (vect_permute_store_chain): Use auto_vec_perm_indices.
11350 (vect_permute_load_chain): Likewise.
11351 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
11352 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
11353 Update uses of can_vec_perm_p.
11354 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
11355 mode with a number of elements. Take a vec_perm_indices *.
11356 (vect_create_epilog_for_reduction): Update accordingly.
11357 Use auto_vec_perm_indices.
11358 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
11359 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
11360 (vect_transform_slp_perm_load): Likewise.
11361 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
11362 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
11363 (vect_gen_perm_mask_checked): Likewise.
11364 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
11365 (vect_gen_perm_mask_checked): Likewise.
11366 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
11367 (vectorizable_store): Likewise.
11368 (vectorizable_load): Likewise.
11369 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
11370 (vectorizable_bswap): Likewise.
11372 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11373 Alan Hayward <alan.hayward@arm.com>
11374 David Sherwood <david.sherwood@arm.com>
11376 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
11377 * tree.c (build_vector): Likewise.
11378 (build_vector_from_ctor): Update accordingly.
11379 (build_vector_from_val): Likewise.
11380 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
11381 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11382 * tree-vect-generic.c (add_rshift): Likewise.
11383 (expand_vector_divmod): Likewise.
11384 (optimize_vector_constructor): Likewise.
11385 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
11386 (vect_transform_slp_perm_load): Likewise.
11387 (vect_schedule_slp_instance): Likewise.
11388 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
11389 (vectorizable_call): Likewise.
11390 (vect_gen_perm_mask_any): Likewise. Add elements in order.
11391 * expmed.c (make_tree): Likewise.
11392 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
11393 a vector passed to build_vector.
11394 (fold_convert_const): Likewise.
11395 (exact_inverse): Likewise.
11396 (fold_ternary_loc): Likewise.
11397 (fold_relational_const): Likewise.
11398 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
11399 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
11400 (const_unop): Likewise. Store the reduction accumulator in a
11401 variable rather than an array.
11402 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
11403 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
11404 the new vector, rather than constructing it after the input arrays.
11405 (native_interpret_vector): Use auto_vec<tree> when building
11406 a vector passed to build_vector. Add elements in order.
11407 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
11408 auto_vec<tree> when building a vector passed to build_vector.
11409 (vect_create_epilog_for_reduction): Likewise.
11410 (vectorizable_induction): Likewise.
11411 (get_initial_def_for_reduction): Likewise. Fix indentation of
11413 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
11415 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
11416 passed to build_vector.
11418 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11419 Alan Hayward <alan.hayward@arm.com>
11420 David Sherwood <david.sherwood@arm.com>
11422 * tree-core.h (tree_base::u): Add an "nelts" field.
11423 (tree_vector): Use VECTOR_CST_NELTS as the length.
11424 * tree.c (tree_size): Likewise.
11425 (make_vector): Initialize VECTOR_CST_NELTS.
11426 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
11427 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
11428 TYPE_VECTOR_SUBPARTS.
11429 * expr.c (const_vector_mask_from_tree): Consistently use "units"
11430 as the number of units, setting it from VECTOR_CST_NELTS.
11431 (const_vector_from_tree): Likewise.
11432 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
11433 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
11434 (fold_negate_expr_1): Likewise.
11435 (fold_convert_const): Likewise.
11436 (const_binop): Likewise. Differentiate the number of output and
11438 (const_unop): Likewise.
11439 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
11440 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
11441 in cases that did the opposite.
11443 2017-09-14 Richard Biener <rguenther@suse.de>
11445 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
11448 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
11450 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
11452 2017-09-14 Jakub Jelinek <jakub@redhat.com>
11455 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
11456 if and where to split a bb, except for splitting before debug insn
11457 sequences followed by non-label real insn. Delete debug insns
11458 in between basic blocks.
11460 * combine.c (make_compound_operation_int): Formatting fixes.
11462 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
11463 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11464 * config/netbsd.h (LINK_EH_SPEC): Likewise.
11465 * config/sol2.h (LINK_EH_SPEC): Likewise.
11466 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11467 * config/s390/linux.h (LINK_SPEC): Likewise.
11468 * config/freebsd.h (LINK_EH_SPEC): Likewise.
11469 * config/openbsd.h (LINK_EH_SPEC): Likewise.
11470 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11471 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
11472 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
11473 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11474 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
11475 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
11477 2017-09-13 Jakub Jelinek <jakub@redhat.com>
11479 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
11481 (ENDFILE_LINUX_SPEC): Likewise.
11482 (LINK_EH_SPEC): Likewise.
11483 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
11484 (LINK_OS_LINUX_SPEC32): Likewise.
11485 (LINK_OS_LINUX_SPEC64): Likewise.
11486 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
11487 (LINK_OS_LINUX_SPEC): Likewise.
11489 2017-09-13 Martin Liska <mliska@suse.cz>
11491 PR middle-end/82154
11492 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
11493 CASE_HIGH is NULL_TREE.
11495 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11496 Alan Hayward <alan.hayward@arm.com>
11497 David Sherwood <david.sherwood@arm.com>
11499 * target.def (secondary_memory_needed): New hook.
11500 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
11501 instead of SECONDARY_MEMORY_NEEDED.
11502 (secondary_memory_needed_mode): Likewise.
11503 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
11504 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
11505 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
11506 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
11507 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
11508 * doc/tm.texi: Regenerate.
11509 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
11510 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
11511 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11512 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
11513 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
11514 * config/i386/i386.c (inline_secondary_memory_needed): Put the
11515 mode argument first and change the reg_class arguments to reg_class_t.
11516 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
11517 Make static. Update the call to inline_secondary_memory_needed.
11518 (ix86_register_move_cost): Update the call to
11519 inline_secondary_memory_needed.
11520 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11521 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
11523 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
11524 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11526 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
11527 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
11528 * config/mips/mips.c (mips_secondary_memory_needed): Make static
11529 and match hook interface. Add comment from mips.h.
11530 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11531 * config/mmix/mmix.md (truncdfsf2): Refer to
11532 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11534 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
11535 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
11536 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11537 (pa_secondary_memory_needed): New function.
11538 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
11539 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
11540 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11541 (pdp11_secondary_memory_needed): Make static and match hook interface.
11542 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
11543 * config/powerpcspe/powerpcspe-protos.h
11544 (rs6000_secondary_memory_needed_ptr): Delete.
11545 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
11547 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11548 (rs6000_option_override_internal): Assign to
11549 targetm.secondary_memory_needed rather than
11550 rs6000_secondary_memory_needed_ptr.
11551 (rs6000_secondary_memory_needed): Match hook interface.
11552 (rs6000_debug_secondary_memory_needed): Likewise.
11553 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
11554 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
11555 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
11556 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11557 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
11558 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
11560 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
11561 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11562 (rs6000_option_override_internal): Assign to
11563 targetm.secondary_memory_needed rather than
11564 rs6000_secondary_memory_needed_ptr.
11565 (rs6000_secondary_memory_needed): Match hook interface.
11566 (rs6000_debug_secondary_memory_needed): Likewise.
11567 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
11568 * config/s390/s390.c (s390_secondary_memory_needed): New function.
11569 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11570 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
11571 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11572 (sparc_secondary_memory_needed): New function.
11573 * lra-constraints.c (check_and_process_move): Refer to
11574 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11576 (curr_insn_transform): Likewise.
11577 (process_alt_operands): Use targetm.secondary_memory_needed
11578 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11579 (check_secondary_memory_needed_p): Likewise.
11580 (choose_split_class): Likewise.
11581 * reload.c: Unconditionally include code that was previously
11582 conditional on SECONDARY_MEMORY_NEEDED.
11583 (push_secondary_reload): Use targetm.secondary_memory_needed
11584 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11585 (push_reload): Likewise.
11586 * reload1.c: Unconditionally include code that was previously
11587 conditional on SECONDARY_MEMORY_NEEDED.
11588 (choose_reload_regs): Use targetm.secondary_memory_needed
11589 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11590 (gen_reload): Likewise.
11591 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
11593 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11594 Alan Hayward <alan.hayward@arm.com>
11595 David Sherwood <david.sherwood@arm.com>
11597 * target.def (secondary_memory_needed_mode): New hook:
11598 * targhooks.c (default_secondary_memory_needed_mode): Declare.
11599 * targhooks.h (default_secondary_memory_needed_mode): New function.
11600 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
11601 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
11602 * doc/tm.texi: Regenerate.
11603 * lra-constraints.c (check_and_process_move): Use
11604 targetm.secondary_memory_needed_mode instead of
11605 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
11606 (curr_insn_transform): Likewise.
11607 * reload.c (get_secondary_mem): Likewise.
11608 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11609 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
11611 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11612 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11613 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
11614 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11615 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
11617 * config/powerpcspe/powerpcspe-protos.h
11618 (rs6000_secondary_memory_needed_mode): Delete.
11619 * config/powerpcspe/powerpcspe.c
11620 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11621 (rs6000_secondary_memory_needed_mode): Make static.
11622 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11623 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
11625 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11627 (rs6000_secondary_memory_needed_mode): Make static.
11628 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11629 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
11630 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11631 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11632 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11634 (sparc_secondary_memory_needed_mode): New function.
11635 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
11637 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
11639 * config/aarch64/constraints.md (Umq): New constraint.
11640 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
11642 (mov<mode>): Update condition.
11644 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11646 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
11647 when bitposition is the same.
11649 2017-09-13 Richard Biener <rguenther@suse.de>
11651 * dwarf2out.c (output_die_symbol): Remove.
11652 (output_die): Do not output a DIEs symbol.
11654 2017-09-13 Richard Biener <rguenther@suse.de>
11656 PR middle-end/82128
11657 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
11658 default-def to avoid breaking iterator update with the weird
11659 interaction with cgraph_update_edges_for_call_stmt_node.
11661 2017-09-13 Richard Biener <rguenther@suse.de>
11663 * tree-cfg.c (verify_gimple_assign_binary): Add verification
11664 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
11665 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
11666 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
11668 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
11670 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
11671 Disable pc relative literal load irrespective of
11672 TARGET_FIX_ERR_A53_84341 for default.
11674 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
11676 * config/sparc/sparc.c (output_return): Output the source location of
11677 the insn in the delay slot, if any.
11678 (output_sibcall): Likewise.
11680 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
11683 * common.opt (-static-pie): New alias.
11684 (shared): Negate static-pie.
11685 (-no-pie): Update help text.
11687 (static-pie): New option.
11688 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
11689 -static-pie support.
11690 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
11691 (LINK_EH_SPEC): Likewise.
11692 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11693 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11694 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11695 * gcc.c (LINK_COMMAND_SPEC): Likewise.
11696 (init_gcc_specs): Likewise.
11697 (init_spec): Likewise.
11698 (display_help): Update help message for -pie.
11699 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
11702 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
11704 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
11705 (movdi_aarch64): Likewise.
11706 (movti_aarch64): Likewise.
11708 2017-09-12 Simon Wright <simon@pushface.org>
11711 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
11712 calculation of the minor version, always output as 0.
11714 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11717 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11718 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
11719 on it early, rather than manual conversion late. For
11720 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
11721 instead of performing manual conversion.
11723 2017-09-12 Carl Love <cel@us.ibm.com>
11725 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
11726 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
11728 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11729 VMULOSW): Add definitions.
11730 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11731 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11732 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11733 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11734 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11736 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11738 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
11740 (movti_aarch64): Likewise.
11741 (movdf_aarch64): Likewise.
11742 (movtf_aarch64): Likewise.
11743 (load_pairdi): Likewise.
11744 (store_pairdi): Likewise.
11745 (load_pairdf): Likewise.
11746 (store_pairdf): Likewise.
11747 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
11748 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
11749 (ldr_got_small_<mode>): Likewise.
11750 (ldr_got_small_28k_<mode>): Likewise.
11751 (ldr_got_tiny): Likewise.
11752 * config/aarch64/iterators.md (ldst_sz): New.
11753 (ldpstp_sz): Likewise.
11754 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
11756 (thunderx_load): Split load_8 to load_16.
11757 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
11759 (thunderx2t99_storepair_basic): Split store_8 to store_16.
11760 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
11761 (xgene1_store_pair): Split store_8 to store_16.
11762 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
11763 (falkor_st_0_st_sd): Split store_8 to store_16.
11765 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11767 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
11768 and store1/2/3/4 to store_4/8/12/16.
11769 * config/aarch64/aarch64.md: Update for rename.
11770 * config/arm/arm.md: Likewise.: Likewise.
11771 * config/arm/arm.c: Likewise.
11772 * config/arm/thumb1.md: Likewise.
11773 * config/arm/thumb2.md: Likewise.
11774 * config/arm/vfp.md: Likewise.
11775 * config/arm/arm-generic.md: Likewise.
11776 * config/arm/arm1020e.md: Likewise.
11777 * config/arm/arm1026ejs.md: Likewise.
11778 * config/arm/arm1136jfs.md: Likewise.
11779 * config/arm/arm926ejs.md: Likewise.
11780 * config/arm/cortex-a15.md: Likewise.
11781 * config/arm/cortex-a17.md: Likewise.
11782 * config/arm/cortex-a5.md: Likewise.
11783 * config/arm/cortex-a53.md: Likewise.
11784 * config/arm/cortex-a57.md: Likewise.
11785 * config/arm/cortex-a7.md: Likewise.
11786 * config/arm/cortex-a8.md: Likewise.
11787 * config/arm/cortex-a9.md: Likewise.
11788 * config/arm/cortex-m4.md: Likewise.
11789 * config/arm/cortex-m7.md: Likewise.
11790 * config/arm/cortex-r4.md: Likewise.
11791 * config/arm/exynos-m1.md: Likewise.
11792 * config/arm/fa526.md: Likewise.
11793 * config/arm/fa606te.md: Likewise.
11794 * config/arm/fa626te.md: Likewise.
11795 * config/arm/fa726te.md: Likewise.
11796 * config/arm/fmp626.md: Likewise.
11797 * config/arm/iwmmxt.md: Likewise.
11798 * config/arm/ldmstm.md: Likewise.
11799 * config/arm/marvell-pj4.md: Likewise.
11800 * config/arm/xgene1.md: Likewise.
11801 * config/aarch64/thunderx.md: Likewise.
11802 * config/aarch64/thunderx2t99.md: Likewise.
11803 * config/aarch64/falkor.md: Likewise.
11805 2017-09-12 Martin Liska <mliska@suse.cz>
11807 * attribs.c (private_lookup_attribute): New function.
11808 * attribs.h (private_lookup_attribute): Declared here.
11809 (lookup_attribute): Called from this place.
11811 2017-09-12 Richard Biener <rguenther@suse.de>
11813 PR tree-optimization/82157
11814 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
11815 stmts with side-effects.
11817 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11818 Alan Hayward <alan.hayward@arm.com>
11819 David Sherwood <david.sherwood@arm.com>
11821 * target.def (hard_regno_nregs): New hook.
11822 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
11823 * targhooks.h (default_hard_regno_nregs): Declare.
11824 * targhooks.c (default_hard_regno_nregs): New function.
11825 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
11826 (TARGET_HARD_REGNO_NREGS): ...this hook.
11827 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
11828 (CLASS_MAX_NREGS): Likewise.
11829 * doc/tm.texi: Regenerate.
11830 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
11831 instead of HARD_REGNO_NREGS.
11832 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
11833 HARD_REGNO_NREGS in the comment.
11834 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
11835 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
11836 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
11837 Return an unsigned int.
11838 (TARGET_HARD_REGNO_NREGS): Redefine.
11839 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
11840 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
11841 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11842 (arc_hard_regno_nregs): New function.
11843 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
11844 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
11845 (arm_hard_regno_nregs): New function.
11846 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
11847 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
11848 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
11849 (TARGET_HARD_REGNO_NREGS): Redefine.
11850 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
11851 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
11852 (HARD_REGNO_NREGS): Delete.
11853 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
11854 (cr16_hard_regno_nregs): New function.
11855 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
11856 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
11857 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
11858 (cris_hard_regno_nregs): New function.
11859 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
11860 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
11861 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11862 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
11863 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
11864 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
11865 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
11866 (frv_hard_regno_nregs): Make static. Take and return an
11868 (frv_class_max_nregs): Remove outdated copy of documentation.
11869 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
11870 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
11871 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
11872 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
11873 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
11874 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
11875 (TARGET_HARD_REGNO_NREGS): Redefine.
11876 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
11877 (CLASS_MAX_NREGS): Update comment.
11878 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
11879 (ia64_hard_regno_nregs): New function.
11880 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
11881 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
11882 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
11883 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
11884 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
11886 (m32c_hard_regno_nregs): Likewise. Make static.
11887 (TARGET_HARD_REGNO_NREGS): Redefine.
11888 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
11889 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
11890 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
11891 (m68k_hard_regno_nregs): New function.
11892 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
11893 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
11894 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
11895 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
11896 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
11897 Take and return an unsigned int.
11898 (TARGET_HARD_REGNO_NREGS): Redefine.
11899 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
11900 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11901 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
11902 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
11903 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
11904 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
11905 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
11906 (msp430_hard_regno_nregs): Make static. Take and return an
11908 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
11909 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
11910 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
11911 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
11912 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
11913 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
11914 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
11915 (TARGET_HARD_REGNO_NREGS): Redefine.
11916 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
11917 (PA_HARD_REGNO_NREGS): ...this.
11918 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
11919 (PA_HARD_REGNO_NREGS): ...this.
11920 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11921 (pa_hard_regno_nregs): New function.
11922 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
11923 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
11924 (pdp11_hard_regno_nregs): New function.
11925 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
11926 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
11927 (rs6000_hard_regno_nregs_hook): New function.
11928 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
11929 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
11930 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
11931 Take and return an unsigned int. Move earlier in file.
11932 (TARGET_HARD_REGNO_NREGS): Redefine.
11933 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
11934 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
11935 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
11936 (rl78_hard_regno_nregs): Make static. Take and return an
11938 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
11939 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
11940 (rs6000_hard_regno_nregs_hook): New function.
11941 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
11942 * config/rx/rx.c (rx_hard_regno_nregs): New function.
11943 (TARGET_HARD_REGNO_NREGS): Redefine.
11944 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
11945 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
11946 instead of HARD_REGNO_NREGS.
11947 (s390_hard_regno_nregs): New function.
11948 (s390_hard_regno_mode_ok): Add comment from s390.h.
11949 (TARGET_HARD_REGNO_NREGS): Redefine.
11950 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
11951 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
11952 (sh_hard_regno_nregs): New function.
11953 (sh_pass_in_reg_p): Use it.
11954 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
11955 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11956 (sparc_hard_regno_nregs): New function.
11957 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
11958 * config/spu/spu.c (spu_hard_regno_nregs): New function.
11959 (spu_function_arg_advance): Use it, supplying a valid register number.
11960 (TARGET_HARD_REGNO_NREGS): Redefine.
11961 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
11962 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
11963 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
11964 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
11965 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
11966 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
11967 (CLASS_MAX_NREGS): Remove copy of old documentation.
11968 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
11969 (visium_hard_regno_nregs): New function.
11970 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
11971 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
11972 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11973 xtensa_hard_regno_nregs): New function.
11974 * system.h (HARD_REGNO_NREGS): Poison.
11976 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11978 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
11979 hard_regno_nregs instead of HARD_REGNO_NREGS.
11980 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
11981 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
11982 (c6x_expand_epilogue): Likewise.
11983 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
11984 (frv_read_iacc_argument): Likewise.
11985 * config/sh/sh.c: Include regs.h.
11986 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11987 (regs_used): Likewise.
11988 (output_stack_adjust): Likewise.
11989 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
11990 * expmed.c: Include regs.h.
11991 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
11992 * ree.c: Include regs.h.
11993 (combine_reaching_defs): Use hard_regno_nregs instead of
11995 (add_removable_extension): Likewise.
11997 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11999 * regs.h (hard_regno_nregs): Turn into a function.
12000 (end_hard_regno): Update accordingly.
12001 * caller-save.c (setup_save_areas): Likewise.
12002 (save_call_clobbered_regs): Likewise.
12003 (replace_reg_with_saved_mem): Likewise.
12004 (insert_restore): Likewise.
12005 (insert_save): Likewise.
12006 * combine.c (can_change_dest_mode): Likewise.
12007 (move_deaths): Likewise.
12008 (distribute_notes): Likewise.
12009 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12010 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12011 (rs6000_split_multireg_move): Likewise.
12012 (rs6000_register_move_cost): Likewise.
12013 (rs6000_memory_move_cost): Likewise.
12014 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12015 (rs6000_split_multireg_move): Likewise.
12016 (rs6000_register_move_cost): Likewise.
12017 (rs6000_memory_move_cost): Likewise.
12018 * cselib.c (cselib_reset_table): Likewise.
12019 (cselib_lookup_1): Likewise.
12020 * emit-rtl.c (set_mode_and_regno): Likewise.
12021 * function.c (aggregate_value_p): Likewise.
12022 * ira-color.c (setup_profitable_hard_regs): Likewise.
12023 (check_hard_reg_p): Likewise.
12024 (calculate_saved_nregs): Likewise.
12025 (assign_hard_reg): Likewise.
12026 (improve_allocation): Likewise.
12027 (calculate_spill_cost): Likewise.
12028 * ira-emit.c (modify_move_list): Likewise.
12029 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12030 (ira_hard_reg_in_set_p): Likewise.
12031 * ira.c (setup_reg_mode_hard_regset): Likewise.
12032 (clarify_prohibited_class_mode_regs): Likewise.
12033 (check_allocation): Likewise.
12034 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12035 (lra_setup_reg_renumber): Likewise.
12036 (setup_try_hard_regno_pseudos): Likewise.
12037 (spill_for): Likewise.
12038 (assign_hard_regno): Likewise.
12039 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12040 * lra-constraints.c (in_class_p): Likewise.
12041 (lra_constraint_offset): Likewise.
12042 (simplify_operand_subreg): Likewise.
12043 (lra_constraints): Likewise.
12044 (split_reg): Likewise.
12045 (split_if_necessary): Likewise.
12046 (invariant_p): Likewise.
12047 (inherit_in_ebb): Likewise.
12048 * lra-lives.c (process_bb_lives): Likewise.
12049 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12050 (get_hard_regs): Likewise.
12051 (do_remat): Likewise.
12052 * lra-spills.c (assign_spill_hard_regs): Likewise.
12053 * mode-switching.c (create_pre_exit): Likewise.
12054 * postreload.c (reload_combine_recognize_pattern): Likewise.
12055 * recog.c (peep2_find_free_register): Likewise.
12056 * regcprop.c (kill_value_regno): Likewise.
12057 (set_value_regno): Likewise.
12058 (copy_value): Likewise.
12059 (maybe_mode_change): Likewise.
12060 (find_oldest_value_reg): Likewise.
12061 (copyprop_hardreg_forward_1): Likewise.
12062 * regrename.c (check_new_reg_p): Likewise.
12063 (regrename_do_replace): Likewise.
12064 * reload.c (push_reload): Likewise.
12065 (combine_reloads): Likewise.
12066 (find_dummy_reload): Likewise.
12067 (operands_match_p): Likewise.
12068 (find_reloads): Likewise.
12069 (find_equiv_reg): Likewise.
12070 (reload_adjust_reg_for_mode): Likewise.
12071 * reload1.c (count_pseudo): Likewise.
12072 (count_spilled_pseudo): Likewise.
12073 (find_reg): Likewise.
12074 (clear_reload_reg_in_use): Likewise.
12075 (free_for_value_p): Likewise.
12076 (allocate_reload_reg): Likewise.
12077 (choose_reload_regs): Likewise.
12078 (reload_adjust_reg_for_temp): Likewise.
12079 (emit_reload_insns): Likewise.
12080 (delete_output_reload): Likewise.
12081 * rtlanal.c (subreg_get_info): Likewise.
12082 * sched-deps.c (sched_analyze_reg): Likewise.
12083 * sel-sched.c (init_regs_for_mode): Likewise.
12084 (mark_unavailable_hard_regs): Likewise.
12085 (choose_best_reg_1): Likewise.
12086 (verify_target_availability): Likewise.
12087 * valtrack.c (dead_debug_insert_temp): Likewise.
12088 * var-tracking.c (track_loc_p): Likewise.
12089 (emit_note_insn_var_location): Likewise.
12090 * varasm.c (make_decl_rtl): Likewise.
12091 * reginfo.c (choose_hard_reg_mode): Likewise.
12092 (init_reg_modes_target): Refer directly to
12093 this_target_regs->x_hard_regno_nregs.
12095 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12097 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
12098 instead of hard_regno_nregs.
12100 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12102 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
12103 end_hard_regno instead of hard_regno_nregs.
12104 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
12105 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
12106 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
12107 * ira-color.c (improve_allocation): Likewise.
12108 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12109 * lra-lives.c (mark_regno_live): Likewise.
12110 (mark_regno_dead): Likewise.
12111 * lra-remat.c (operand_to_remat): Likewise.
12112 * lra.c (collect_non_operand_hard_regs): Likewise.
12113 * postreload.c (reload_combine_note_store): Likewise.
12114 (move2add_valid_value_p): Likewise.
12115 * reload.c (regno_clobbered_p): Likewise.
12117 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12119 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
12121 * config/v850/v850.c (v850_reorg): Likewise.
12122 * reload.c (refers_to_regno_for_reload_p): Likewise.
12123 (find_equiv_reg): Likewise.
12124 * reload1.c (reload_reg_reaches_end_p): Likewise.
12126 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12128 * caller-save.c (add_used_regs): Use REG_NREGS instead of
12130 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
12131 * config/arm/arm.c (output_move_neon): Likewise.
12132 (arm_attr_length_move_neon): Likewise.
12133 (neon_split_vcombine): Likewise.
12134 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
12135 (c6x_mark_reg_written): Likewise.
12136 (c6x_dwarf_register_span): Likewise.
12137 * config/i386/i386.c (ix86_save_reg): Likewise.
12138 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
12139 (rws_access_reg): Likewise.
12140 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
12141 * mode-switching.c (create_pre_exit): Likewise.
12142 * ree.c (combine_reaching_defs): Likewise.
12143 (add_removable_extension): Likewise.
12144 * regcprop.c (find_oldest_value_reg): Likewise.
12145 (copyprop_hardreg_forward_1): Likewise.
12146 * reload.c (reload_inner_reg_of_subreg): Likewise.
12147 (push_reload): Likewise.
12148 (combine_reloads): Likewise.
12149 (find_dummy_reload): Likewise.
12150 (reload_adjust_reg_for_mode): Likewise.
12151 * reload1.c (find_reload_regs): Likewise.
12152 (forget_old_reloads_1): Likewise.
12153 (reload_reg_free_for_value_p): Likewise.
12154 (reload_adjust_reg_for_temp): Likewise.
12155 (emit_reload_insns): Likewise.
12156 (delete_output_reload): Likewise.
12157 * sel-sched.c (choose_best_reg_1): Likewise.
12158 (choose_best_pseudo_reg): Likewise.
12160 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12161 Alan Hayward <alan.hayward@arm.com>
12162 David Sherwood <david.sherwood@arm.com>
12164 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
12165 * target.def (slow_unaligned_access): New hook.
12166 * targhooks.h (default_slow_unaligned_access): Declare.
12167 * targhooks.c (default_slow_unaligned_access): New function.
12168 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
12169 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
12170 * doc/tm.texi: Regenerate.
12171 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
12172 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
12173 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
12175 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
12176 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
12178 (rs6000_slow_unaligned_access): New function.
12179 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12180 (expand_block_compare): Likewise.
12181 (expand_strn_compare): Likewise.
12182 (rs6000_rtx_costs): Likewise.
12183 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
12184 (riscv_slow_unaligned_access): Likewise.
12185 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
12186 (riscv_slow_unaligned_access_p): ...this and make static.
12187 (riscv_option_override): Update accordingly.
12188 (riscv_slow_unaligned_access): New function.
12189 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12190 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
12191 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12192 (rs6000_slow_unaligned_access): New function.
12193 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12194 (rs6000_rtx_costs): Likewise.
12195 * config/rs6000/rs6000-string.c (expand_block_compare)
12196 (expand_strn_compare): Use targetm.slow_unaligned_access instead
12197 of SLOW_UNALIGNED_ACCESS.
12198 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
12199 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
12200 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
12201 of SLOW_UNALIGNED_ACCESS.
12202 * expmed.c (simple_mem_bitfield_p): Likewise.
12203 * expr.c (alignment_for_piecewise_move): Likewise.
12204 (emit_group_load_1): Likewise.
12205 (emit_group_store): Likewise.
12206 (copy_blkmode_from_reg): Likewise.
12207 (emit_push_insn): Likewise.
12208 (expand_assignment): Likewise.
12209 (store_field): Likewise.
12210 (expand_expr_real_1): Likewise.
12211 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
12212 * lra-constraints.c (simplify_operand_subreg): Likewise.
12213 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12214 * gimple-ssa-store-merging.c: Likewise in block comment at start
12216 * tree-ssa-strlen.c: Include target.h.
12217 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
12218 of SLOW_UNALIGNED_ACCESS.
12219 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
12221 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12223 PR rtl-optimization/82185
12224 * expmed.c (emit_store_flag_int): Only test tem if it has been
12227 2017-09-12 Richard Biener <rguenther@suse.de>
12229 PR middle-end/82149
12230 * match.pd ((FTYPE) N CMP CST): Fix typo.
12232 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
12234 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
12236 (mips_near_type_p): Add 'short_call' attribute as a synonym
12238 * doc/extend.texi (short_call): Document new function attribute.
12240 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12243 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
12244 assertion check that in the condition.
12245 (get_atomic_generic_size): Likewise. Before testing if parameter
12246 has pointer type, if it has array type, call for C++
12247 default_conversion to perform array-to-pointer conversion.
12249 2017-09-12 Richard Biener <rguenther@suse.de>
12251 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
12252 for operations we cannot scalarize.
12254 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
12256 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
12257 vectors heap vectors. Clean up comments.
12258 Make visited_bbs a reference.
12259 (profitable_jump_thread_path): Make GC
12260 vectors heap vectors. Clean up comments.
12262 (convert_and_register_jump_thread_path): Make GC vectors heap
12264 (check_subpath_and_update_thread_path): Same. Clean up comments.
12265 Make visited_bbs a reference.
12266 (handle_phi): Abstract common code to to
12267 register_jump_thread_path_if_profitable.
12268 Rename VAR_BB to DEF_BB.
12270 Make GC vectors heap vectors.
12271 Make visited_bbs a reference.
12272 (handle_assignment): Same.
12273 (register_jump_thread_path_if_profitable): New.
12274 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
12276 Make GC vectors heap vectors. Clean up comments.
12277 Make visited_bbs a reference.
12278 (find_jump_threads_backwards): Make visited_bbs live in the stack.
12279 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
12282 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
12285 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
12286 words of E_DImode object are reachable by xtensa_uimm8x4 access.
12288 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
12290 Revert r251800 and r251799.
12292 2017-09-11 Martin Jambor <mjambor@suse.cz>
12295 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
12296 arguments in advance.
12297 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
12298 use it to find predecessor edges.
12299 (naive_outof_ssa): Collect vector of predecessors.
12301 2017-09-08 Jason Merrill <jason@redhat.com>
12303 PR c++/70029 - ICE with ref-qualifier and -flto
12304 * langhooks.h (struct lang_hooks_for_types): Add
12305 copy_lang_qualifiers.
12306 * attribs.c (build_type_attribute_qual_variant): Use it.
12307 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
12309 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
12310 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
12312 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
12315 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
12316 (*mulsi3_sp64): New instruction.
12317 (mulsi3): New expander.
12319 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
12321 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
12323 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12325 * sancov.c: Include memmodel.h.
12327 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
12330 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
12333 2017-09-07 Carl Love <cel@us.ibm.com>
12335 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
12336 the sldi instruction.
12338 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
12340 * sancov.c: Include tm_p.h.
12342 2017-09-07 Jakub Jelinek <jakub@redhat.com>
12345 * output.h (switch_to_other_text_partition): New declaration.
12346 * varasm.c (switch_to_other_text_partition): New function.
12347 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
12348 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
12349 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
12350 to the other text partition before emitting LCL label and switch back
12351 after emitting the word after it.
12353 2017-09-07 Richard Biener <rguenther@suse.de>
12355 * passes.def (pass_split_crit_edges): Remove instance before PRE.
12356 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
12357 critical edges here, after loop init.
12358 (pass_data_pre): Remove PROP_no_crit_edges flags.
12359 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
12360 for valueization of call args to avoid leaking VN_TOP.
12361 (visit_use): Assert we do not visit default defs.
12362 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
12363 Use error_mark_node to more easily detect leaking VN_TOP.
12364 All default-defs are varying, not VN_TOP. Mark them visited.
12365 (run_scc_vn): Make code match comment.
12367 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12369 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
12370 OPTION_MASK_FLOAT128_KEYWORD.
12371 (POWERPC_MASKS): Likewise.
12372 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
12373 support for the -mfloat128-type option, and make -mfloat128
12374 default on PowerPC Linux systems. Define or undefine
12375 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
12376 Define __float128 to be __ieee128 if IEEE 128-bit support is
12377 enabled, or undefine it.
12378 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
12379 Delete defining __FLOAT128_TYPE__.
12380 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
12381 -mfloat128-type option and make -mfloat128 default on PowerPC
12383 (TARGET_FLOAT128_TYPE): Likewise.
12384 (-mfloat128-type): Likewise.
12385 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12386 Delete the -mfloat128-type option and make -mfloat128 default on
12387 PowerPC Linux systems. Always use __ieee128 to be the keyword for
12388 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
12389 128-bit floating point is enabled. Change tests from using
12390 -mfloat128-type to -mfloat128.
12391 (rs6000_mangle_type): Use the correct mangling for the __float128
12392 type even if normal long double is restricted to 64-bits.
12393 (floatn_mode): Enable the _Float128 type by default on VSX Linux
12395 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
12396 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
12397 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
12399 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
12400 documentation for -mfloat128.
12402 2017-09-06 Olivier Hainque <hainque@adacore.com>
12404 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
12406 2017-09-06 Wish Wu <wishwu007@gmail.com>
12407 Jakub Jelinek <jakub@redhat.com>
12409 * asan.c (initialize_sanitizer_builtins): Add
12410 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
12411 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
12412 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
12413 BT_FN_VOID_UINT64_PTR variables.
12414 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
12415 (BT_FN_VOID_UINT16_UINT16): Likewise.
12416 (BT_FN_VOID_UINT32_UINT32): Likewise.
12417 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
12418 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
12419 (BT_FN_VOID_UINT64_PTR): Likewise.
12420 * common.opt (flag_sanitize_coverage): New variable.
12421 (fsanitize-coverage=trace-pc): Remove.
12422 (fsanitize-coverage=): Add.
12423 * flag-types.h (enum sanitize_coverage_code): New enum.
12424 * fold-const.c (fold_range_test): Disable non-short-circuit
12425 optimization if flag_sanitize_coverage.
12426 (fold_truth_andor): Likewise.
12427 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
12428 * opts.c (COVERAGE_SANITIZER_OPT): Define.
12429 (coverage_sanitizer_opts): New array.
12430 (get_closest_sanitizer_option): Add OPTS argument, handle also
12431 OPT_fsanitize_coverage_.
12432 (parse_sanitizer_options): Adjusted to also handle
12433 OPT_fsanitize_coverage_.
12434 (common_handle_option): Add OPT_fsanitize_coverage_.
12435 * sancov.c (instrument_comparison, instrument_switch): New function.
12436 (sancov_pass): Add trace-cmp support.
12437 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
12438 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
12439 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
12440 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
12441 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
12442 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
12443 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
12444 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
12445 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
12446 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
12448 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12450 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12451 error. Only quit immediately if parsing is complete.
12452 (BEGIN): Initialize fatal_err and parse_done.
12453 (begin fpu, end fpu): Check number of arguments.
12454 (begin arch, end arch): Likewise.
12455 (begin cpu, end cpu): Likewise.
12456 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12457 (optalias): Likewise.
12459 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12461 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12462 * config/arm/arm-isa.h: Delete. Move definitions to ...
12463 * arm-cpus.in: ... here. Use new feature and fgroup values.
12464 * config/arm/arm.c (arm_option_override): Use lower case for feature
12466 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12467 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12468 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12470 (print_isa_bits_for): New function.
12471 (gen_isa): New function.
12472 (gen_comm_data): Use print_isa_bits_for.
12473 (define feature): New keyword.
12474 (define fgroup): New keyword.
12475 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
12476 (arm-isa.h): Add rule to generate file.
12477 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12478 case for feature bit names.
12480 2017-09-06 Richard Biener <rguenther@suse.de>
12482 * tree-ssa-pre.c (NECESSARY): Remove.
12483 (create_expression_by_pieces): Do not touch pass-local flags.
12484 (insert_into_preds_of_block): Likewise.
12485 (do_pre_regular_insertion): Likewise.
12486 (eliminate_insert): Likewise.
12487 (eliminate_dom_walker::before_dom_children): Likewise.
12488 (fini_eliminate): Do not look at inserted_exprs.
12489 (mark_operand_necessary): Remove.
12490 (remove_dead_inserted_code): Replace with simple work-list
12491 algorithm based on inserted_exprs and SSA uses.
12492 (pass_pre::execute): Re-order fini_eliminate and
12493 remove_dead_inserted_code.
12495 2017-09-06 Olivier Hainque <hainque@adacore.com>
12497 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
12498 for VxWorks 7. Adjust surrounding comments.
12500 2017-09-06 Richard Biener <rguenther@suse.de>
12502 * gimple-ssa-strength-reduction.c
12503 (find_candidates_dom_walker::before_dom_children): Also allow
12506 2017-09-06 Richard Biener <rguenther@suse.de>
12508 PR tree-optimization/82108
12509 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
12510 for gap in the non-permutation SLP case.
12512 2017-09-06 Martin Jambor <mjambor@suse.cz>
12514 PR tree-optimization/82078
12515 * tree-sra.c (sort_and_splice_var_accesses): Move call to
12516 add_access_to_work_queue...
12517 (build_accesses_from_assign): ...here.
12518 (propagate_all_subaccesses): Make sure racc is the group
12519 representative, if there is one.
12521 2017-09-06 Jakub Jelinek <jakub@redhat.com>
12523 PR middle-end/82095
12524 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
12527 2017-09-06 Richard Biener <rguenther@suse.de>
12529 * gimple-ssa-strength-reduction.c
12530 (find_candidates_dom_walker::before_doom_children): Use a
12531 type and not a mode check.
12533 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
12536 * config/arm/predicates.md (arm_general_adddi_operand): Create new
12538 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
12540 2017-09-05 Jeff Law <law@redhat.com>
12542 PR tree-optimization/64910
12543 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
12544 cases where we have 3 or more operands.
12546 2017-09-05 Jakub Jelinek <jakub@redhat.com>
12548 PR middle-end/81768
12549 * omp-low.c (lower_omp_for): Recompute tree invariant if
12550 gimple_omp_for_initial/final is ADDR_EXPR.
12552 PR middle-end/81768
12553 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
12554 into gimple val before gimplification fo the COND_EXPR.
12556 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
12558 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
12559 REGION_COPY argument.
12560 (thread_through_all_blocks): Remove unused argument to
12561 duplicate_thread_path.
12563 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12564 Alan Hayward <alan.hayward@arm.com>
12565 David Sherwood <david.sherwood@arm.com>
12567 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
12568 Take a scalar_mode rather than a machine_mode.
12569 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12570 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
12571 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12572 (aarch64_gen_adjusted_ldpstp): Likewise.
12573 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
12575 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12576 Alan Hayward <alan.hayward@arm.com>
12577 David Sherwood <david.sherwood@arm.com>
12579 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
12580 Take a scalar_int_mode instead of a machine_mode.
12581 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
12582 (aarch64_output_scalar_simd_mov_immediate): Likewise.
12583 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
12584 (aarch64_simd_attr_length_rglist): Delete.
12585 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
12586 a scalar_int_mode instead of a machine_mode.
12587 (aarch64_add_offset): Likewise.
12588 (aarch64_internal_mov_immediate): Likewise
12589 (aarch64_add_constant_internal): Likewise.
12590 (aarch64_add_constant): Likewise.
12591 (aarch64_movw_imm): Likewise.
12592 (aarch64_rtx_arith_op_extract_p): Likewise.
12593 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
12594 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
12595 Remove assert that the mode isn't a vector.
12596 (aarch64_output_scalar_simd_mov_immediate): Likewise.
12597 (aarch64_expand_mov_immediate): Update calls after above changes.
12598 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
12599 (aarch64_and_bitmask_imm): Check for scalar integer modes.
12600 (aarch64_move_imm): Likewise.
12601 (aarch64_can_const_movi_rtx_p): Likewise.
12602 (aarch64_strip_extend): Likewise.
12603 (aarch64_extr_rtx_p): Likewise.
12604 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
12605 a CONST_INT when the mode parameter is VOIDmode.
12606 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
12608 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12610 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
12611 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12612 (bitwise_type_for_mode): Update accordingly.
12614 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12616 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
12617 * stor-layout.c (mode_for_size_tree): Likewise.
12618 (mode_for_array): Update accordingly.
12619 (layout_decl): Likewise.
12620 (compute_record_mode): Likewise. Only set the mode once.
12622 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12624 * target.def (get_mask_mode): Change return type to opt_mode.
12626 * doc/tm.texi: Regenerate.
12627 * targhooks.h (default_get_mask_mode): Return an opt_mode.
12628 * targhooks.c (default_get_mask_mode): Likewise.
12629 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12630 * optabs-query.c (can_vec_mask_load_store_p): Update use of
12631 targetm.get_mask_mode.
12632 * tree.c (build_truth_vector_type): Likewise.
12634 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12636 * machmode.h (mode_for_vector): Return an opt_mode.
12637 * stor-layout.c (mode_for_vector): Likewise.
12638 (mode_for_int_vector): Update accordingly.
12639 (layout_type): Likewise.
12640 * config/i386/i386.c (emit_memmov): Likewise.
12641 (ix86_expand_set_or_movmem): Likewise.
12642 (ix86_expand_vector_init): Likewise.
12643 (ix86_get_mask_mode): Likewise.
12644 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
12646 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
12647 * expmed.c (extract_bit_field_1): Likewise.
12648 * expr.c (expand_expr_real_2): Likewise.
12649 * optabs-query.c (can_vec_perm_p): Likewise.
12650 (can_vec_mask_load_store_p): Likewise.
12651 * optabs.c (expand_vec_perm): Likewise.
12652 * targhooks.c (default_get_mask_mode): Likewise.
12653 * tree-vect-stmts.c (vectorizable_store): Likewise.
12654 (vectorizable_load): Likewise.
12655 (get_vectype_for_scalar_type_and_size): Likewise.
12657 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12659 * machmode.h (mode_for_int_vector): New function.
12660 * stor-layout.c (mode_for_int_vector): Likewise.
12661 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
12662 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
12663 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12664 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12665 (s390_expand_vcond): Likewise.
12667 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12669 * machmode.h (opt_machine_mode): New type.
12670 (opt_mode<T>): Allow construction from anything that can be
12672 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
12673 (mode_for_size): Return an opt_machine_mode.
12674 * stor-layout.c (mode_for_size): Likewise.
12675 (mode_for_size_tree): Update call accordingly.
12676 (bitwise_mode_for_mode): Likewise.
12677 (make_fract_type): Likewise.
12678 (make_accum_type): Likewise.
12679 * caller-save.c (replace_reg_with_saved_mem): Update call
12681 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12682 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12683 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12684 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12685 * expmed.c (extract_bit_field_1): Likewise.
12686 * reload.c (get_secondary_mem): Likewise.
12687 * varasm.c (assemble_integer): Likewise.
12688 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
12691 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12693 * machmode.h (decimal_float_mode_for_size): New function.
12694 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
12695 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
12696 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
12697 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
12698 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
12699 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
12701 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12703 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
12704 (get_builtin_sync_mode): Likewise.
12705 (expand_ifn_atomic_compare_exchange): Likewise.
12706 (expand_builtin_atomic_clear): Likewise.
12707 (expand_builtin_atomic_test_and_set): Likewise.
12708 (fold_builtin_atomic_always_lock_free): Likewise.
12709 * calls.c (compute_argument_addresses): Likewise.
12710 (emit_library_call_value_1): Likewise.
12711 (store_one_arg): Likewise.
12712 * combine.c (combine_instructions): Likewise.
12713 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
12714 * config/arm/arm.c (arm_function_value): Likewise.
12715 (aapcs_allocate_return_reg): Likewise.
12716 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
12717 * config/i386/i386.c (construct_container): Likewise.
12718 (ix86_gimplify_va_arg): Likewise.
12719 (ix86_expand_sse_cmp): Likewise.
12720 (emit_memmov): Likewise.
12721 (emit_memset): Likewise.
12722 (expand_small_movmem_or_setmem): Likewise.
12723 (ix86_expand_pextr): Likewise.
12724 (ix86_expand_pinsr): Likewise.
12725 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
12726 * config/microblaze/microblaze.c (microblaze_block_move_straight):
12728 * config/mips/mips.c (mips_function_value_1) Likewise.
12729 (mips_block_move_straight): Likewise.
12730 (mips_expand_ins_as_unaligned_store): Likewise.
12731 * config/powerpcspe/powerpcspe.c
12732 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12733 (rs6000_darwin64_record_arg_flush): Likewise.
12734 * config/rs6000/rs6000.c
12735 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12736 (rs6000_darwin64_record_arg_flush): Likewise.
12737 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
12738 (sparc_function_value_1): Likewise.
12739 * config/spu/spu.c (adjust_operand): Likewise.
12740 (spu_emit_branch_or_set): Likewise.
12741 (arith_immediate_p): Likewise.
12742 * emit-rtl.c (gen_lowpart_common): Likewise.
12743 * expr.c (expand_expr_real_1): Likewise.
12744 * function.c (assign_parm_setup_block): Likewise.
12745 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
12746 * reload1.c (alter_reg): Likewise.
12747 * stor-layout.c (mode_for_vector): Likewise.
12748 (layout_type): Likewise.
12750 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12752 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
12753 (spu_convert_move): Likewise.
12754 * lower-subreg.c (resolve_simple_move): Likewise.
12756 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12759 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
12760 define_insn to a define_expand.
12761 (altivec_vsum2sws_direct): New define_insn.
12762 (altivec_vsumsws): Convert from a define_insn to a define_expand.
12764 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12766 * config/arm/arm.c (arm_option_params_internal): Improve setting of
12769 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
12774 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
12775 realign stack if stack alignment needed is less than incoming
12778 2017-09-05 Marek Polacek <polacek@redhat.com>
12781 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
12784 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12786 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
12788 2017-09-05 Richard Biener <rguenther@suse.de>
12790 PR tree-optimization/82084
12791 * fold-const.c (can_native_encode_string_p): Handle wide characters.
12793 2017-09-05 Richard Biener <rguenther@suse.de>
12795 PR tree-optimization/82102
12796 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
12798 2017-09-05 Martin Liska <mliska@suse.cz>
12800 PR tree-optimization/82032
12801 * tree-cfg.c (generate_range_test): New function.
12802 * tree-cfg.h (generate_range_test): Declared here.
12803 * tree-cfgcleanup.c (convert_single_case_switch): New function.
12804 (cleanup_control_expr_graph): Use it.
12805 * tree-switch-conversion.c (try_switch_expansion): Remove
12807 (emit_case_nodes): Use generate_range_test.
12809 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12812 * config/i386/i386.md (*<btsc><mode>_mask): Add
12813 TARGET_USE_BT to insn constraint.
12814 (*btr<mode>_mask): Ditto.
12816 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
12818 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
12819 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
12821 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
12824 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
12825 TARGET_NEON and TARGET_IWMMXT.
12826 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
12827 TARGET_NEON and TARGET_IWMMXT.
12828 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
12830 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12832 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
12833 (ix86_rewrite_tls_address): Ditto.
12834 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
12835 (ix86_rewrite_tls_address_1): Ditto.
12836 (ix86_rewrite_tls_address): Ditto.
12837 * config/i386/predicates.md (tls_address_pattern): New predicate.
12838 * config/i386/i386.md (TLS address splitter): New splitter.
12840 2017-09-04 Richard Biener <rguenther@suse.de>
12842 PR tree-optimization/82084
12843 * fold-const.h (can_native_encode_string_p): Declare.
12844 * fold-const.c (can_native_encode_string_p): Factor out from ...
12845 (native_encode_string): ... here.
12846 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
12847 vectorizing stores from constants we later cannot handle.
12849 2017-09-04 Marek Polacek <polacek@redhat.com>
12852 * doc/invoke.texi: Update -Wtautological-compare documentation.
12854 2017-09-04 Jeff Law <law@redhat.com>
12856 PR tree-optimization/64910
12857 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
12858 swap the first and last operand if the last is a constant.
12860 2017-09-04 Marek Polacek <polacek@redhat.com>
12863 * convert.c (do_narrow): When sanitizing signed integer overflows,
12864 bail out for signed types.
12865 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
12867 2017-09-04 Richard Biener <rguenther@suse.de>
12869 PR tree-optimization/82060
12870 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12871 Move devirtualization after stmt folding and before EH/AB/noreturn
12872 cleanup to get the stmt refs canonicalized. Use a bool instead
12873 of gimple_modified_p since that doesn't work for NOPs. Schedule
12874 NOPs generated by folding for removal.
12876 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12877 Alan Hayward <alan.hayward@arm.com>
12878 David Sherwood <david.sherwood@arm.com>
12880 * coretypes.h (pad_direction): New enum.
12881 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
12882 (FUNCTION_ARG_PADDING): Likewise.
12883 * target.def (function_arg_padding): New hook.
12884 * targhooks.h (default_function_arg_padding): Declare.
12885 * targhooks.c (default_function_arg_padding): New function.
12886 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
12887 (TARGET_FUNCTION_ARG_PADDING): ...this.
12888 * doc/tm.texi: Regenerate.
12889 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
12890 instead of direction.
12891 (compute_argument_addresses): Likewise.
12892 (load_register_parameters): Likewise.
12893 (emit_library_call_value_1): Likewise.
12894 (store_one_arg): Use targetm.calls.function_arg_padding instead
12895 of FUNCTION_ARG_PADDING.
12896 (must_pass_in_stack_var_size_or_pad): Likewise.
12897 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
12898 (emit_group_store): Likewise.
12899 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
12900 instead of FUNCTION_ARG_PADDING.
12901 (emit_push_insn): Likewise, and propagate enum change throughout
12903 * function.h (direction): Delete.
12904 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
12906 * function.c (assign_parm_find_stack_rtl): Likewise.
12907 (assign_parm_setup_block_p): Likewise.
12908 (assign_parm_setup_block): Likewise.
12909 (gimplify_parameters): Likewise.
12910 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
12911 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
12913 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
12914 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12915 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
12916 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
12917 (aarch64_function_arg_padding): ...this new function.
12918 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
12919 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12920 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
12921 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12922 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
12923 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12924 (arm_pad_arg_upward): Replace with...
12925 (arm_function_arg_padding): ...this new function.
12926 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
12928 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
12929 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
12930 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12931 (ia64_hpux_function_arg_padding): Replace with...
12932 (ia64_function_arg_padding): ...this new function. Use pad_direction
12933 instead of direction. Check for TARGET_HPUX.
12934 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
12935 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12936 (iq2000_function_arg_padding): New function.
12937 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
12938 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
12939 (mips_function_arg_padding): ...this new function.
12940 (mips_pad_reg_upward): Update accordingly.
12941 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12942 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
12943 targetm.calls.function_arg_padding.
12944 (FUNCTION_ARG_PADDING): Delete.
12945 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12946 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
12947 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12948 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
12949 (nios2_block_reg_padding): Return pad_direction instead of direction.
12950 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
12951 instead of direction.
12952 (nios2_function_arg_padding): Likewise. Make static.
12953 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12954 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
12955 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
12956 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
12957 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12958 (pa_function_arg_padding): Make static. Return pad_direction instead
12960 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
12961 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12962 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
12963 instead of direction. Use targetm.calls.function_arg_padding.
12964 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
12965 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
12966 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
12967 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
12968 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
12970 (function_arg_padding): Rename to...
12971 (rs6000_function_arg_padding): ...this. Make static. Return
12972 pad_direction instead of direction.
12973 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12974 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
12975 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12976 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
12977 instead of direction. Use targetm.calls.function_arg_padding.
12978 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
12979 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
12980 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
12981 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
12982 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12983 (function_arg_padding): Rename to...
12984 (rs6000_function_arg_padding): ...this. Make static. Return
12985 pad_direction instead of direction.
12986 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12987 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
12988 * config/s390/s390.c (s390_function_arg_padding): New function.
12989 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12990 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
12991 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
12992 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12993 (function_arg_padding): Rename to...
12994 (sparc_function_arg_padding): ...this. Make static. Return
12995 pad_direction instead of direction.
12996 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
12997 * config/spu/spu.c (spu_function_arg_padding): New function.
12998 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12999 * system.h (FUNCTION_ARG_PADDING): Poison.
13001 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13002 Alan Hayward <alan.hayward@arm.com>
13003 David Sherwood <david.sherwood@arm.com>
13005 * target.def (modes_tieable_p): New hook.
13006 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13007 (TARGET_MODES_TIEABLE_P): ...this.
13008 * doc/tm.texi.in: Regenerate.
13009 * hooks.h (hook_bool_mode_mode_true): Declare.
13010 * hooks.c (hook_bool_mode_mode_true): New function.
13011 * combine.c (subst): Use targetm.modes_tieable_p instead of
13013 * dse.c (find_shift_sequence): Likewise.
13014 * expmed.c (extract_low_bits): Likewise.
13015 * lower-subreg.c: Include target.h.
13016 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13018 * rtlanal.c (rtx_cost): Likewise.
13019 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13020 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13021 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13022 (TARGET_MODES_TIEABLE_P): Redefine.
13023 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13024 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13025 (TARGET_MODES_TIEABLE_P): Redefine.
13026 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13027 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13028 (arc_modes_tieable_p): New function.
13029 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13030 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13031 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13032 (arm_modes_tieable_p): Make static.
13033 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13034 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13035 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13036 (TARGET_MODES_TIEABLE_P): Redefine.
13037 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13038 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13039 (TARGET_MODES_TIEABLE_P): Redefine.
13040 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13041 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13042 (cr16_modes_tieable_p): New function.
13043 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13044 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13045 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13046 (TRULY_NOOP_TRUNCATION): Update comment.
13047 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13048 (TRULY_NOOP_TRUNCATION): Update comment.
13049 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13050 (frv_modes_tieable_p): New function.
13051 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13052 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13053 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13054 (TARGET_MODES_TIEABLE_P): Redefine.
13055 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13056 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13057 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13058 (TARGET_MODES_TIEABLE_P): Redefine.
13059 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
13060 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
13061 (ia64_modes_tieable_p): New function.
13062 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
13063 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
13064 (iq2000_modes_tieable_p): New function.
13065 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
13066 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
13067 (lm32_modes_tieable_p): New function.
13068 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
13069 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
13070 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
13071 (TARGET_MODES_TIEABLE_P): Redefine.
13072 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
13073 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
13074 (m32r_modes_tieable_p): New function.
13075 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
13076 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
13077 (m68k_modes_tieable_p): New function.
13078 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
13079 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
13080 (mcore_modes_tieable_p): New function.
13081 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
13082 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
13084 (TARGET_MODES_TIEABLE_P): Redefine.
13085 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
13086 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
13087 * config/mips/mips.c (mips_modes_tieable_p): Make static.
13088 (TARGET_MODES_TIEABLE_P): Redefine.
13089 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
13090 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
13091 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
13092 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
13093 (mn10300_modes_tieable_p): ...this and make static.
13094 (TARGET_MODES_TIEABLE_P): Redefine.
13095 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
13096 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
13097 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
13098 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
13099 (msp430_modes_tieable_p): Make static.
13100 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
13101 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
13102 (TARGET_MODES_TIEABLE_P): Redefine.
13103 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
13104 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
13105 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
13106 (TARGET_MODES_TIEABLE_P): Redefine.
13107 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
13108 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
13109 * config/pa/pa.c (pa_modes_tieable_p): Make static.
13110 (TARGET_MODES_TIEABLE_P): Redefine.
13111 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
13112 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
13113 (pdp11_modes_tieable_p): New function.
13114 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
13115 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
13116 (rs6000_modes_tieable_p): New function.
13117 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13118 * config/powerpcspe/powerpcspe.md: Update comment.
13119 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
13120 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
13121 (TARGET_MODES_TIEABLE_P): Redefine.
13122 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
13123 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
13124 (rl78_modes_tieable_p): New function.
13125 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
13126 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
13127 (rs6000_modes_tieable_p): New function.
13128 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13129 * config/rs6000/rs6000.md: Update comment.
13130 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
13131 * config/rx/rx.c (rx_modes_tieable_p): New function.
13132 (TARGET_MODES_TIEABLE_P): Redefine.
13133 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
13134 * config/s390/s390.c (s390_modes_tieable_p): New function.
13135 (TARGET_MODES_TIEABLE_P): Redefine.
13136 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
13137 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
13138 (sh_modes_tieable_p): New function.
13139 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
13140 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
13141 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
13142 (sparc_modes_tieable_p): Make static.
13143 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
13144 * config/spu/spu.c (spu_modes_tieable_p): New function.
13145 (TARGET_MODES_TIEABLE_P): Redefine.
13146 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
13147 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
13148 (TARGET_MODES_TIEABLE_P): Redefine.
13149 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
13150 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
13151 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
13152 * config/v850/v850.c (v850_modes_tieable_p): New function.
13153 (TARGET_MODES_TIEABLE_P): Redefine.
13154 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
13155 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
13156 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
13157 (visium_modes_tieable_p): New function.
13158 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
13159 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
13160 (xtensa_modes_tieable_p): New function.
13161 * system.h (MODES_TIEABLE_P): Poison.
13163 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13164 Alan Hayward <alan.hayward@arm.com>
13165 David Sherwood <david.sherwood@arm.com>
13167 * target.def (hard_regno_mode_ok): New hook.
13168 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
13169 (TARGET_HARD_REGNO_MODE_OK): ...this.
13170 * doc/tm.texi.in: Regenerate.
13171 * hooks.h (hook_bool_uint_mode_true): Declare.
13172 * hooks.c (hook_bool_uint_mode_true): New function.
13173 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
13174 HARD_REGNO_MODE_OK.
13175 * genpreds.c (write_insn_preds_c): Add an include of target.h.
13176 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
13177 instead of HARD_REGNO_MODE_OK.
13178 * caller-save.c: Include target.h.
13179 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
13180 HARD_REGNO_MODE_OK.
13181 * combine.c (can_combine_p): Likewise.
13182 (combinable_i3pat): Likewise.
13183 (can_change_dest_mode): Likewise.
13184 * expr.c (init_expr_target): Likewise.
13185 (convert_move): Likewise.
13186 (convert_modes): Likewise.
13187 * ira.c (setup_prohibited_class_mode_regs): Likewise.
13188 (setup_prohibited_mode_move_regs): Likewise.
13189 * ira.h (target_ira): Likewise.
13190 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13191 * lra-constraints.c (process_alt_operands): Likewise.
13192 (split_reg): Likewise.
13193 * recog.c (peep2_find_free_register): Likewise.
13194 * ree.c (combine_reaching_defs): Likewise.
13195 * regcprop.c (maybe_mode_change): Likewise.
13196 * reginfo.c (init_reg_sets_1): Likewise.
13197 (choose_hard_reg_mode): Likewise.
13198 (simplifiable_subregs): Likewise.
13199 * regrename.c (check_new_reg_p): Likewise.
13200 * reload.c (find_valid_class): Likewise.
13201 (find_valid_class_1): Likewise.
13202 (reload_inner_reg_of_subreg): Likewise.
13203 (push_reload): Likewise.
13204 (combine_reloads): Likewise.
13205 (find_dummy_reload): Likewise.
13206 (find_reloads): Likewise.
13207 * reload1.c (find_reg): Likewise.
13208 (set_reload_reg): Likewise.
13209 (allocate_reload_reg): Likewise.
13210 (choose_reload_regs): Likewise.
13211 (reload_adjust_reg_for_temp): Likewise.
13212 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
13213 (simplify_subreg_regno): Likewise.
13214 * sel-sched.c (init_regs_for_mode): Likewise.
13215 * varasm.c (make_decl_rtl): Likewise.
13216 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
13217 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
13218 HARD_REGNO_MODE_OK.
13219 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
13220 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
13221 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13222 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
13223 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
13224 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13225 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
13226 (arc_mode_class): Delete.
13227 (HARD_REGNO_MODE_OK): Delete.
13228 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13229 (arc_hard_regno_mode_ok): Rename old array to...
13230 (arc_hard_regno_mode_ok_modes): ...this.
13231 (arc_conditional_register_usage): Update accordingly.
13232 (arc_mode_class): Make static.
13233 (arc_hard_regno_mode_ok): New function.
13234 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
13235 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
13236 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13237 (arm_hard_regno_mode_ok): Make static.
13238 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
13239 HARD_REGNO_MODE_OK.
13240 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
13241 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
13242 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
13244 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13245 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
13246 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
13247 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
13248 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
13249 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13250 * config/bfin/predicates.md (valid_reg_operand): Use
13251 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
13252 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
13253 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
13254 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13255 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
13256 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
13257 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13258 (cr16_hard_regno_mode_ok): Make static and return a bool.
13259 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
13260 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13261 (cris_hard_regno_mode_ok): New function.
13262 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
13263 (epiphany_mode_class): Delete.
13264 (HARD_REGNO_MODE_OK): Delete.
13265 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
13266 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13267 (hard_regno_mode_ok): Rename to...
13268 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
13269 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
13270 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
13271 HARD_REGNO_MODE_OK.
13272 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
13273 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
13274 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13275 (frv_hard_regno_mode_ok): Make static and return a bool.
13276 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
13277 HARD_REGNO_MODE_OK.
13278 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
13279 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
13280 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
13281 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
13283 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13284 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
13285 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
13286 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
13288 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13289 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
13290 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13291 (ia64_hard_regno_mode_ok): New function.
13292 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
13293 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13294 (iq2000_hard_regno_mode_ok): New function.
13295 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
13296 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13297 (lm32_hard_regno_mode_ok): New function.
13298 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
13299 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
13300 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
13301 instead of HARD_REGNO_MODE_OK.
13302 (m32c_hard_regno_ok): Rename to...
13303 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
13304 (m32c_cannot_change_mode_class): Update accordingly.
13305 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13306 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
13307 (m32r_mode_class): Delete.
13308 (HARD_REGNO_MODE_OK): Delete.
13309 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13310 (m32r_hard_regno_mode_ok): Rename to...
13311 (m32r_hard_regno_modes): ...this.
13312 (m32r_mode_class): Make static.
13313 (m32r_hard_regno_mode_ok): New function.
13314 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
13315 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
13316 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13317 (m68k_hard_regno_mode_ok): Make static.
13318 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
13319 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13320 (mcore_hard_regno_mode_ok): New function.
13321 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
13322 (HARD_REGNO_MODE_OK): Delete.
13323 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
13325 (microblaze_hard_regno_mode_ok_p): ...this and make static.
13326 (microblaze_hard_regno_mode_ok): New function.
13327 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13328 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
13329 (mips_hard_regno_mode_ok): Delete.
13330 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
13331 (mips_hard_regno_mode_ok_p): ...this and make static.
13332 (mips_hard_regno_mode_ok_p): Rename to...
13333 (mips_hard_regno_mode_ok_uncached): ...this.
13334 (mips_hard_regno_mode_ok): New function.
13335 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
13336 of HARD_REGNO_MODE_OK.
13337 (mips_option_override): Update after above name changes.
13338 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13339 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
13340 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
13341 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
13342 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
13343 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13344 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
13345 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
13346 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
13347 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13348 (msp430_hard_regno_mode_ok): Make static and return a bool.
13349 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
13350 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
13351 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
13353 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13354 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
13355 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
13356 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
13357 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
13358 (PA_HARD_REGNO_MODE_OK): ...this
13359 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
13360 (PA_HARD_REGNO_MODE_OK): ...this.
13361 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13362 (pa_hard_regno_mode_ok): New function.
13363 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
13364 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13365 (pdp11_hard_regno_mode_ok): New function.
13366 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
13367 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
13369 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
13371 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13372 (rs6000_hard_regno_mode_ok): Rename to...
13373 (rs6000_hard_regno_mode_ok_uncached): ...this.
13374 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13375 (rs6000_hard_regno_mode_ok): New function.
13376 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
13377 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
13378 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
13379 (riscv_hard_regno_mode_ok): ...this and make static.
13380 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13381 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
13382 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
13383 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13384 (rl78_hard_regno_mode_ok): Make static and return bool.
13385 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
13386 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
13388 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
13389 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13390 (rs6000_hard_regno_mode_ok): Rename to...
13391 (rs6000_hard_regno_mode_ok_uncached): ...this.
13392 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13393 (rs6000_hard_regno_mode_ok): New function.
13394 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
13395 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
13396 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13397 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
13398 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
13399 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
13400 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13401 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
13402 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
13403 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13404 (sh_hard_regno_mode_ok): Make static.
13405 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
13406 instead of HARD_REGNO_MODE_OK.
13407 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
13408 (sparc_mode_class): Delete.
13409 (HARD_REGNO_MODE_OK): Delete.
13410 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13411 (hard_regno_mode_classes): Make static.
13412 (sparc_mode_class): Likewise.
13413 (sparc_hard_regno_mode_ok): New function.
13414 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
13415 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
13416 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
13418 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13419 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
13420 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
13421 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
13422 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
13423 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13424 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
13425 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
13426 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13427 (visium_hard_regno_mode_ok): New function.
13428 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
13429 instead of HARD_REGNO_MODE_OK.
13430 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
13431 (HARD_REGNO_MODE_OK): Delete.
13432 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
13433 (xtensa_hard_regno_mode_ok_p): ...this and make static.
13434 (xtensa_option_override): Update accordingly.
13435 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13436 (xtensa_hard_regno_mode_ok): New function.
13437 * system.h (HARD_REGNO_MODE_OK): Poison.
13439 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13440 Alan Hayward <alan.hayward@arm.com>
13441 David Sherwood <david.sherwood@arm.com>
13443 * target.def (hard_regno_call_part_clobbered): New hook.
13444 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
13445 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
13446 * doc/tm.texi: Regenerate.
13447 * hooks.h (hook_bool_uint_mode_false): Declare.
13448 * hooks.c (hook_bool_uint_mode_false): New function.
13449 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13450 * cselib.c (cselib_process_insn): Use
13451 targetm.hard_regno_call_part_clobbered instead of
13452 HARD_REGNO_CALL_PART_CLOBBERED.
13453 * ira-conflicts.c (ira_build_conflicts): Likewise.
13454 * ira-costs.c (ira_tune_allocno_costs): Likewise.
13455 * lra-constraints.c (need_for_call_save_p): Likewise.
13456 * lra-lives.c: Include target.h.
13457 (check_pseudos_live_through_calls): Use
13458 targetm.hard_regno_call_part_clobbered instead of
13459 HARD_REGNO_CALL_PART_CLOBBERED.
13460 * regcprop.c: Include target.h.
13461 (copyprop_hardreg_forward_1): Use
13462 targetm.hard_regno_call_part_clobbered instead of
13463 HARD_REGNO_CALL_PART_CLOBBERED.
13464 * reginfo.c (choose_hard_reg_mode): Likewise.
13465 * regrename.c (check_new_reg_p): Likewise.
13466 * reload.c (find_equiv_reg): Likewise.
13467 * reload1.c (emit_reload_insns): Likewise.
13468 * sched-deps.c (deps_analyze_insn): Likewise.
13469 * sel-sched.c (init_regs_for_mode): Likewise.
13470 (mark_unavailable_hard_regs): Likewise.
13471 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
13472 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13473 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
13475 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13476 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13477 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
13479 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
13481 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13482 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13483 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
13485 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13486 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13487 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
13489 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13490 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
13492 * config/powerpcspe/powerpcspe.c
13493 (rs6000_hard_regno_call_part_clobbered): New function.
13494 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13495 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13496 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
13498 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13499 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13500 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
13502 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13503 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13504 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
13506 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13507 Alan Hayward <alan.hayward@arm.com>
13508 David Sherwood <david.sherwood@arm.com>
13510 * rtl.h (subreg_memory_offset): Declare.
13511 * emit-rtl.c (subreg_memory_offset): New function.
13512 * expmed.c (store_bit_field_1): Use it.
13513 * expr.c (undefined_operand_subword_p): Likewise.
13514 * simplify-rtx.c (simplify_subreg): Likewise.
13516 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
13518 PR rtl-optimization/57448
13521 * optabs.c (expand_atomic_load): Place compiler memory barriers if
13522 using atomic_load pattern.
13523 (expand_atomic_store): Likewise.
13525 2017-09-04 Jakub Jelinek <jakub@redhat.com>
13528 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
13529 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
13530 handling. Use replace_call_with_value with NULL instead of
13531 gsi_replace, unlink_stmt_vdef and release_defs.
13533 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
13536 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
13538 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13541 * rtl.h (emit_library_call_value_1): Declare.
13542 (emit_library_call): Replace declaration with a series of overloads.
13543 Remove the parameter count argument.
13544 (emit_library_call_value): Likewise.
13545 * calls.c (emit_library_call_value_1): Make global. Replace varargs
13546 with an "rtx_mode_t *".
13547 (emit_library_call_value): Delete.
13548 (emit_library_call): Likewise.
13549 * asan.c (asan_emit_stack_protection): Update calls accordingly.
13550 (asan_emit_allocas_unpoison): Likewise.
13551 * builtins.c (expand_builtin_powi): Likewise.
13552 (expand_asan_emit_allocas_unpoison): Likewise.
13553 * cfgexpand.c (expand_main_function): Likewise.
13554 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
13555 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
13556 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
13557 * config/arm/arm.c (arm_trampoline_init): Likewise.
13558 (arm_call_tls_get_addr): Likewise.
13559 (arm_expand_divmod_libfunc): Likewise.
13560 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
13561 (smulsi3_highpart): Likewise.
13562 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
13563 (c6x_expand_compare): Likewise.
13564 (c6x_expand_movmem): Likewise.
13565 * config/frv/frv.c (frv_trampoline_init): Likewise.
13566 * config/i386/i386.c (ix86_trampoline_init): Likewise.
13567 (ix86_expand_divmod_libfunc): Likewise.
13568 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
13569 (ia64_expand_compare): Likewise.
13570 (ia64_profile_hook): Likewise.
13571 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
13572 (nonlocal_goto): Likewise.
13573 (restore_stack_nonlocal): Likewise.
13574 * config/m32r/m32r.c (block_move_call): Likewise.
13575 (m32r_trampoline_init): Likewise.
13576 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
13577 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
13578 (m68k_call_m68k_read_tp): Likewise.
13579 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
13580 (microblaze_expand_divide): Likewise.
13581 * config/mips/mips.h (mips_args): Likewise.
13582 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
13583 (MIPS_ICACHE_SYNC): Likewise.
13584 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
13585 (nios2_trampoline_init): Likewise.
13586 * config/pa/pa.c (hppa_tls_call): Likewise.
13587 (pa_trampoline_init): Likewise.
13588 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
13589 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
13590 (expand_strn_compare): Likewise.
13591 (rs6000_generate_compare): Likewise.
13592 (rs6000_expand_float128_convert): Likewise.
13593 (output_profile_hook): Likewise.
13594 (rs6000_trampoline_init): Likewise.
13595 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
13596 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
13597 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
13598 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
13599 (rs6000_generate_compare): Likewise.
13600 (rs6000_expand_float128_convert): Likewise.
13601 (output_profile_hook): Likewise.
13602 (rs6000_trampoline_init): Likewise.
13603 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
13604 * config/sh/sh.c (sh_trampoline_init): Likewise.
13605 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
13606 (sparc_emit_float_lib_cmp): Likewise.
13607 (sparc32_initialize_trampoline): Likewise.
13608 (sparc64_initialize_trampoline): Likewise.
13609 (sparc_profile_hook): Likewise.
13610 * config/spu/spu.c (ea_load_store): Likewise.
13611 * config/spu/spu.md (floatunssidf2): Likewise.
13612 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
13613 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
13614 * config/visium/visium.c (expand_block_move_4): Likewise.
13615 (expand_block_move_2): Likewise.
13616 (expand_block_move_1): Likewise.
13617 (expand_block_set_4): Likewise.
13618 (expand_block_set_2): Likewise.
13619 (expand_block_set_1): Likewise.
13620 (visium_trampoline_init): Likewise.
13621 (visium_profile_hook): Likewise.
13622 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
13623 (xtensa_setup_frame_addresses): Likewise.
13624 (xtensa_trampoline_init): Likewise.
13625 * except.c (sjlj_emit_function_enter): Likewise.
13626 (sjlj_emit_function_exit): Likewise.
13627 * explow.c (allocate_dynamic_stack_space): Likewise.
13628 (probe_stack_range): Likewise.
13629 * expr.c (convert_mode_scalar): Likewise.
13630 * optabs.c (expand_binop): Likewise.
13631 (expand_twoval_binop_libfunc): Likewise.
13632 (expand_unop): Likewise.
13633 (prepare_cmp_insn): Likewise.
13634 (prepare_float_lib_cmp): Likewise.
13635 (expand_float): Likewise.
13636 (expand_fix): Likewise.
13637 (expand_fixed_convert): Likewise.
13638 (maybe_emit_sync_lock_test_and_set): Likewise.
13639 (expand_atomic_compare_and_swap): Likewise.
13640 (expand_mem_thread_fence): Likewise.
13641 (expand_atomic_fetch_op): Likewise.
13643 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
13645 * doc/generic.texi (OpenACC): Adjust URL.
13646 * doc/invoke.texi (C Dialect Options): Ditto.
13648 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
13650 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
13651 predicate for operand 1. Add (m,<S>) constraint.
13652 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
13653 Prevent memory operand 1 with register operand 2.
13655 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
13657 PR rtl-optimization/82024
13658 * combine.c (try_combine): If the combination result is a PARALLEL,
13659 and we only need to retain the SET in there that would be placed
13660 at I2, check that we can place that at I3 instead, before doing so.
13662 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13665 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
13667 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
13668 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
13671 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
13672 Jeff Law <law@redhat.com>
13674 * varasm.c (bss_initializer_p): Do not put constants into .bss
13675 (categorize_decl_for_section): Handle bss_initializer_p returning
13676 false when DECL_INITIAL is NULL.
13678 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13681 * config/s390/s390.c (s390_can_inline_p): New function.
13683 2017-09-01 Jeff Law <law@redhat.com>
13685 PR tree-optimization/82052
13686 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13687 Always initialize the returned slot after a hash table miss
13688 when INSERT is true.
13690 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
13692 * config/s390/s390.md (mem_signal_fence): Remove.
13693 * doc/md.texi (mem_signal_fence): Remove.
13694 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
13696 * target-insns.def (mem_signal_fence): Remove.
13698 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13701 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
13704 * asan.c (create_odr_indicator): Strip name encoding from assembler
13705 name before appending it after __odr_asan_.
13707 2017-09-01 Martin Liska <mliska@suse.cz>
13709 PR tree-optimization/82059
13710 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
13711 frequency only when an edge is redirected.
13713 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13715 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
13716 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
13717 (arc_conditional_register_usage): Remove ARC600 lp_count
13719 (arc_file_start): Emit Tag_ARC_CPU_variation.
13720 (arc_can_use_doloop_p): New conditions to use ZOLs.
13721 (hwloop_fail): New function.
13722 (hwloop_optimize): Likewise.
13723 (hwloop_pattern_reg): Likewise.
13724 (arc_doloop_hooks): New struct, to be used with reorg_loops.
13725 (arc_reorg_loops): New function, calls reorg_loops.
13726 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
13727 (arc600_corereg_hazard): Remove ZOL checking, case handled by
13729 (arc_loop_hazard): Remove function, functionality moved into
13731 (arc_hazard): Remove arc_loop_hazard call.
13732 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
13733 into hwloop_optimize.
13734 (arc_label_align): Remove ZOL handling.
13735 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
13736 * config/arc/arc.md (doloop_begin): Remove pattern.
13737 (doloop_begin_i): Likewise.
13738 (doloop_end_i): Likewise.
13739 (doloop_fallback): Likewise.
13740 (doloop_fallback_m): Likewise.
13741 (doloop_end): Reimplement expand.
13742 (arc_lp): New pattern for LP instruction.
13743 (loop_end): New pattern.
13744 (loop_fail): Likewise.
13745 (decrement_and_branch_until_zero): Likewise.
13746 * config/arc/arc.opt (mlpc-width): New option.
13747 * doc/invoke.texi (mlpc-width): Document option.
13749 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13751 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
13752 (arc_ccfsm_advance): Fix checking for delay slots.
13753 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
13755 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13757 * config/arc/arc.md (movqi_insn): Add stores to save constant long
13759 (movhi_insn): Update store instruction constraint which are saving
13760 6-bit short immediates.
13761 (movsi_insn): Consider also short scaled load operations.
13762 (zero_extendhisi2_i): Use Usd constraint instead of T.
13763 (extendhisi2_i): Add q constraint.
13764 (arc_clzsi2): Add type and length attributes.
13765 (arc_ctzsi2): Likewise.
13766 * config/arc/constraints.md (Usc): Update constraint, the
13767 assembler can parse two relocations for a single instruction.
13769 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13771 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
13772 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
13774 2017-08-31 Olivier Hainque <hainque@adacore.com>
13776 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
13777 match as powerpc-wrs-vxworks*.
13779 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
13781 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
13782 register constraint for by-element operand.
13783 (aarch64_mls_elt_merge<mode>): Likewise.
13785 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13787 * config/arc/arc.c (arc_can_follow_jump): Check for short
13790 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13792 * config.gcc: Use g.opt for arc.
13793 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
13794 functionality moved to ...
13795 (legitimate_scaled_address_p): New function, ...here.
13796 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
13797 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
13798 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
13800 (arc_override_options): Handle G option.
13801 (arc_output_pic_addr_const): Correct function definition.
13802 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
13803 (arc_decl_anon_ns_mem_p): Delete.
13804 (arc_in_small_data_p): Overhaul this function to take into
13805 consideration the value given via G option.
13806 (arc_rewrite_small_data_1): Renamed and corrected old
13807 arc_rewrite_small_data function.
13808 (arc_rewrite_small_data): New function.
13809 (small_data_pattern): Don't use pic_offset_table_rtx.
13810 * config/arc/arc.h (CC1_SPEC): Recognize G option.
13811 * config/arc/simdext.md (movmisalignv2hi): Use
13812 prepare_move_operands function.
13814 (movmisalign*): Likewise.
13815 * doc/invoke.texi (ARC options): Document -G option.
13817 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13819 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
13821 * config/arc/arc.c (arc_print_operand): Output scalled address for
13822 sdata whenever is possible.
13823 (arc_in_small_data_p): Allow sdata for 64bit datum when double
13824 load/stores are available.
13825 (compact_sda_memory_operand): Check for the alignment required by
13826 code density instructions.
13827 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
13829 * config/arc/constraints.md (Usd): Update constraint.
13830 (Us0): New constraint.
13831 (Usc): Update constraint.
13833 2017-08-31 Richard Biener <rguenther@suse.de>
13835 PR middle-end/82054
13836 * dwarf2out.c (dwarf2out_early_global_decl): Process each
13837 function only once.
13839 2017-08-31 Tamar Christina <tamar.christina@arm.com>
13841 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
13842 Resize type_signature.
13844 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13845 Alan Hayward <alan.hayward@arm.com>
13846 David Sherwood <david.sherwood@arm.com>
13848 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
13849 subregs whose inner modes can be stored in GPRs.
13850 (aarch64_classify_index): Likewise.
13852 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13853 Alan Hayward <alan.hayward@arm.com>
13854 David Sherwood <david.sherwood@arm.com>
13856 * config/aarch64/iterators.md (V_cmp_result): Rename to...
13857 (V_INT_EQUIV): ...this.
13858 (v_cmp_result): Rename to...
13859 (v_int_equiv): ...this.
13860 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
13861 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
13862 (copysign<mode>3): Likewise.
13863 (aarch64_simd_bsl<mode>_internal): Likewise.
13864 (aarch64_simd_bsl<mode>): Likewise.
13865 (vec_cmp<mode><mode>): Likewise.
13866 (vcond<mode><mode>): Likewise.
13867 (vcond<v_cmp_mixed><mode>): Likewise.
13868 (vcondu<mode><v_cmp_mixed>): Likewise.
13869 (aarch64_cm<optab><mode>): Likewise.
13870 (aarch64_cmtst<mode>): Likewise.
13871 (aarch64_fac<optab><mode>): Likewise.
13872 (vec_perm_const<mode>): Likewise.
13873 (vcond_mask_<mode><v_cmp_result>): Rename to...
13874 (vcond_mask_<mode><v_int_equiv>): ...this.
13875 (vec_cmp<mode><v_cmp_result>): Rename to...
13876 (vec_cmp<mode><v_int_equiv>): ...this.
13878 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13879 Alan Hayward <alan.hayward@arm.com>
13880 David Sherwood <david.sherwood@arm.com>
13882 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
13884 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
13885 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
13886 (UNSPEC_LD4_DREG): New unspecs.
13887 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
13888 (aarch64_ld2<mode>_dreg_be): Replace with...
13889 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
13891 (aarch64_ld3<mode>_dreg_le)
13892 (aarch64_ld3<mode>_dreg_be): Replace with...
13893 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
13895 (aarch64_ld4<mode>_dreg_le)
13896 (aarch64_ld4<mode>_dreg_be): Replace with...
13897 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
13900 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13902 PR tree-optimization/81987
13903 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
13904 insert an initializer in a location not dominated by the stride
13907 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13909 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
13910 on the entire header of the finally block in the fallthru case.
13912 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13914 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
13916 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
13918 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
13919 rs6000_emit_move_from_cr and call renamed function.
13920 (rs6000_emit_prologue): Call renamed functions.
13921 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
13922 movesi_from_cr, remove volatile CRs.
13924 2017-08-30 Jon Beniston <jon@beniston.com>
13925 Richard Biener <rguenther@suse.de>
13927 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
13928 of VECTOR_MODE_P check.
13929 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
13930 element vector types.
13932 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13934 * df.h (df_read_modify_subreg_p): Remove in favor of...
13935 * rtl.h (read_modify_subreg_p): ...this new function. Take a
13936 const_rtx instead of an rtx.
13937 * cprop.c (local_cprop_find_used_regs): Update accordingly.
13938 * df-problems.c (df_word_lr_mark_ref): Likewise.
13939 * ira-lives.c (mark_pseudo_reg_live): Likewise.
13940 (mark_pseudo_reg_dead): Likewise.
13941 (mark_ref_dead): Likewise.
13942 * reginfo.c (init_subregs_of_mode): Likewise.
13943 * sched-deps.c (sched_analyze_1): Likewise.
13944 * df-scan.c (df_def_record_1): Likewise.
13945 (df_uses_record): Likewise.
13946 (df_read_modify_subreg_p): Remove in favor of...
13947 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
13948 const_rtx instead of an rtx.
13950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13951 Alan Hayward <alan.hayward@arm.com>
13952 David Sherwood <david.sherwood@arm.com>
13954 * rtl.h (partial_subreg_p): New function.
13955 * caller-save.c (save_call_clobbered_regs): Use it.
13956 * calls.c (expand_call): Likewise.
13957 * combine.c (combinable_i3pat): Likewise.
13958 (simplify_set): Likewise.
13959 (make_extraction): Likewise.
13960 (make_compound_operation_int): Likewise.
13961 (gen_lowpart_or_truncate): Likewise.
13962 (force_to_mode): Likewise.
13963 (make_field_assignment): Likewise.
13964 (reg_truncated_to_mode): Likewise.
13965 (record_truncated_value): Likewise.
13966 (move_deaths): Likewise.
13967 * cse.c (record_jump_cond): Likewise.
13968 (cse_insn): Likewise.
13969 * cselib.c (cselib_lookup_1): Likewise.
13970 * expmed.c (extract_bit_field_using_extv): Likewise.
13971 * function.c (assign_parm_setup_reg): Likewise.
13972 * ifcvt.c (noce_convert_multiple_sets): Likewise.
13973 * ira-build.c (create_insn_allocnos): Likewise.
13974 * lra-coalesce.c (merge_pseudos): Likewise.
13975 * lra-constraints.c (match_reload): Likewise.
13976 (simplify_operand_subreg): Likewise.
13977 (curr_insn_transform): Likewise.
13978 * lra-lives.c (process_bb_lives): Likewise.
13979 * lra.c (new_insn_reg): Likewise.
13980 (lra_substitute_pseudo): Likewise.
13981 * regcprop.c (mode_change_ok): Likewise.
13982 (maybe_mode_change): Likewise.
13983 (copyprop_hardreg_forward_1): Likewise.
13984 * reload.c (push_reload): Likewise.
13985 (find_reloads): Likewise.
13986 (find_reloads_subreg_address): Likewise.
13987 * reload1.c (alter_reg): Likewise.
13988 (eliminate_regs_1): Likewise.
13989 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13991 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
13993 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
13994 back to if statements, including unpack.
13996 2017-08-30 Martin Liska <mliska@suse.cz>
13998 PR inline-asm/82001
13999 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14001 (func_checker::compare_asm_inputs_outputs): ... this function.
14002 (func_checker::compare_gimple_asm): Use the function to compare
14003 also ASM constrains.
14004 * ipa-icf-gimple.h: Rename the function.
14006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14007 Alan Hayward <alan.hayward@arm.com>
14008 David Sherwood <david.sherwood@arm.com>
14010 * coretypes.h (complex_mode): New type.
14011 * gdbhooks.py (build_pretty_printer): Handle it.
14012 * machmode.h (complex_mode): New class.
14013 (complex_mode::includes_p): New function.
14014 (is_complex_int_mode): Likewise.
14015 (is_complex_float_mode): Likewise.
14016 * genmodes.c (get_mode_class): Handle complex mode classes.
14017 * function.c (expand_function_end): Use is_complex_int_mode.
14019 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14020 Alan Hayward <alan.hayward@arm.com>
14021 David Sherwood <david.sherwood@arm.com>
14023 * coretypes.h (scalar_mode_pod): New typedef.
14024 * gdbhooks.py (build_pretty_printer): Handle it.
14025 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14026 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14027 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14028 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14029 as_a <scalar_mode>.
14031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14032 Alan Hayward <alan.hayward@arm.com>
14033 David Sherwood <david.sherwood@arm.com>
14035 * machmode.h (mode_for_vector): Take a scalar_mode instead
14037 * stor-layout.c (mode_for_vector): Likewise.
14038 * explow.c (promote_mode): Use as_a <scalar_mode>.
14039 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14042 Alan Hayward <alan.hayward@arm.com>
14043 David Sherwood <david.sherwood@arm.com>
14045 * target.def (preferred_simd_mode): Take a scalar_mode
14046 instead of a machine_mode.
14047 * targhooks.h (default_preferred_simd_mode): Likewise.
14048 * targhooks.c (default_preferred_simd_mode): Likewise.
14049 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14050 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14051 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14052 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14053 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14054 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14055 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14056 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14058 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14059 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
14060 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
14061 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
14062 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
14063 * doc/tm.texi: Regenerate.
14064 * optabs-query.c (can_vec_mask_load_store_p): Return false for
14067 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14068 Alan Hayward <alan.hayward@arm.com>
14069 David Sherwood <david.sherwood@arm.com>
14071 * target.def (scalar_mode_supported_p): Take a scalar_mode
14072 instead of a machine_mode.
14073 * targhooks.h (default_scalar_mode_supported_p): Likewise.
14074 * targhooks.c (default_scalar_mode_supported_p): Likewise.
14075 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
14076 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14077 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
14078 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
14079 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
14080 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
14081 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
14082 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
14083 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
14084 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
14085 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
14086 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
14088 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
14089 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
14090 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14091 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14092 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
14094 * doc/tm.texi: Regenerate.
14096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14097 Alan Hayward <alan.hayward@arm.com>
14098 David Sherwood <david.sherwood@arm.com>
14100 * coretypes.h (opt_scalar_mode): New typedef.
14101 * gdbhooks.py (build_pretty_printers): Handle it.
14102 * machmode.h (mode_iterator::get_2xwider): Add overload for
14104 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
14106 * expr.c (convert_mode_scalar): Likewise.
14107 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14108 * optabs.c (expand_float): Likewise.
14109 (expand_fix): Likewise.
14110 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14112 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14113 Alan Hayward <alan.hayward@arm.com>
14114 David Sherwood <david.sherwood@arm.com>
14116 * optabs.c (expand_float): Explicitly check for scalars before
14117 using a branching expansion.
14118 (expand_fix): Likewise.
14120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14121 Alan Hayward <alan.hayward@arm.com>
14122 David Sherwood <david.sherwood@arm.com>
14124 * expr.c (convert_mode): Split scalar handling out into...
14125 (convert_mode_scalar): ...this new function. Treat the modes
14128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14129 Alan Hayward <alan.hayward@arm.com>
14130 David Sherwood <david.sherwood@arm.com>
14132 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
14134 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
14136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14137 Alan Hayward <alan.hayward@arm.com>
14138 David Sherwood <david.sherwood@arm.com>
14140 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
14141 rather than a machine_mode.
14142 (fixed_from_string): Likewise.
14143 (fixed_convert): Likewise.
14144 (fixed_convert_from_int): Likewise.
14145 (fixed_convert_from_real): Likewise.
14146 (real_convert_from_fixed): Likewise.
14147 * fixed-value.c (fixed_from_double_int): Likewise.
14148 (fixed_from_string): Likewise.
14149 (fixed_convert): Likewise.
14150 (fixed_convert_from_int): Likewise.
14151 (fixed_convert_from_real): Likewise.
14152 (real_convert_from_fixed): Likewise.
14153 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
14155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14156 Alan Hayward <alan.hayward@arm.com>
14157 David Sherwood <david.sherwood@arm.com>
14159 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
14160 of separate mode class checks. Do not allow vector modes here.
14161 (immed_wide_int_const): Use as_a <scalar_mode>.
14162 * explow.c (trunc_int_for_mode): Likewise.
14163 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
14164 (wi::shwi): Likewise.
14165 (wi::min_value): Likewise.
14166 (wi::max_value): Likewise.
14167 * dwarf2out.c (loc_descriptor): Likewise.
14168 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
14169 for CONST_WIDE_INT.
14171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14172 Alan Hayward <alan.hayward@arm.com>
14173 David Sherwood <david.sherwood@arm.com>
14175 * tree.h (SCALAR_TYPE_MODE): New macro.
14176 * expr.c (expand_expr_addr_expr_1): Use it.
14177 (expand_expr_real_2): Likewise.
14178 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
14179 (fold_convert_const_fixed_from_int): Likewise.
14180 (fold_convert_const_fixed_from_real): Likewise.
14181 (native_encode_fixed): Likewise
14182 (native_encode_complex): Likewise
14183 (native_encode_vector): Likewise.
14184 (native_interpret_fixed): Likewise.
14185 (native_interpret_real): Likewise.
14186 (native_interpret_complex): Likewise.
14187 (native_interpret_vector): Likewise.
14188 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
14189 (simd_clone_adjust_argument_types): Likewise.
14190 (simd_clone_init_simd_arrays): Likewise.
14191 (simd_clone_adjust): Likewise.
14192 * stor-layout.c (layout_type): Likewise.
14193 * tree.c (build_minus_one_cst): Likewise.
14194 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14195 * tree-inline.c (estimate_move_cost): Likewise.
14196 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
14197 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
14198 (vectorizable_reduction): Likewise.
14199 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
14200 (vect_recog_mixed_size_cond_pattern): Likewise.
14201 (check_bool_pattern): Likewise.
14202 (adjust_bool_pattern): Likewise.
14203 (search_type_for_mask_1): Likewise.
14204 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
14205 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14206 (vectorizable_load): Likewise.
14207 (vectorizable_store): Likewise.
14208 * ubsan.c (ubsan_encode_value): Likewise.
14209 * varasm.c (output_constant): Likewise.
14211 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14212 Alan Hayward <alan.hayward@arm.com>
14213 David Sherwood <david.sherwood@arm.com>
14215 * coretypes.h (scalar_mode): New class.
14216 * machmode.h (scalar_mode): Likewise.
14217 (scalar_mode::includes_p): New function.
14218 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
14219 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
14220 * genmodes.c (get_mode_class): Handle remaining scalar modes.
14221 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
14222 * expmed.c (store_bit_field_1): Likewise.
14223 (extract_bit_field_1): Likewise.
14224 * expr.c (write_complex_part): Likewise.
14225 (read_complex_part): Likewise.
14226 (emit_move_complex_push): Likewise.
14227 (expand_expr_real_2): Likewise.
14228 * function.c (assign_parm_setup_reg): Likewise.
14229 (assign_parms_unsplit_complex): Likewise.
14230 * optabs.c (expand_binop): Likewise.
14231 * rtlanal.c (subreg_get_info): Likewise.
14232 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14233 * varasm.c (output_constant_pool_2): Likewise.
14235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14236 Alan Hayward <alan.hayward@arm.com>
14237 David Sherwood <david.sherwood@arm.com>
14239 * expmed.c (extract_high_half): Use scalar_int_mode and remove
14241 (expmed_mult_highpart_optab): Likewise.
14242 (expmed_mult_highpart): Likewise.
14244 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14245 Alan Hayward <alan.hayward@arm.com>
14246 David Sherwood <david.sherwood@arm.com>
14248 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
14249 instead of a machine_mode.
14250 (builtin_memset_read_str): Likewise.
14251 * builtins.c (c_readstr): Likewise.
14252 (builtin_memcpy_read_str): Likewise.
14253 (builtin_strncpy_read_str): Likewise.
14254 (builtin_memset_read_str): Likewise.
14255 (builtin_memset_gen_str): Likewise.
14256 (expand_builtin_signbit): Use scalar_int_mode for local variables.
14257 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
14258 instead of a machine_mode.
14259 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
14261 (make_extraction): Likewise.
14262 (try_widen_shift_mode): Take and return scalar_int_modes instead
14264 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
14265 a scalar_int_mode instead of a machine_mode.
14266 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
14267 (avr_addr_space_pointer_mode): Likewise.
14268 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
14269 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
14270 (msp430_unwind_word_mode): Likewise.
14271 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
14272 (spu_addr_space_pointer_mode): Likewise.
14273 (spu_addr_space_address_mode): Likewise.
14274 (spu_libgcc_cmp_return_mode): Likewise.
14275 (spu_libgcc_shift_count_mode): Likewise.
14276 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
14277 (rl78_addr_space_pointer_mode): Likewise.
14278 (fl78_unwind_word_mode): Likewise.
14279 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
14281 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
14282 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
14283 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
14284 (mips_valid_pointer_mode): Likewise.
14285 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
14286 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
14287 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
14289 (ft32_addr_space_address_mode): Likewise.
14290 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
14291 scalar_int_mode instead of a machine_mode.
14292 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
14294 (m32c_addr_space_address_mode): Likewise.
14295 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
14296 (rs6000_eh_return_filter_mode): Likewise.
14297 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
14298 (rs6000_eh_return_filter_mode): Likewise.
14299 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
14300 (s390_libgcc_shift_count_mode): Likewise.
14301 (s390_unwind_word_mode): Likewise.
14302 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
14304 * target.def (mode_rep_extended): Likewise.
14305 (valid_pointer_mode): Likewise.
14306 (addr_space.valid_pointer_mode): Likewise.
14307 (eh_return_filter_mode): Return a scalar_int_mode rather than
14309 (libgcc_cmp_return_mode): Likewise.
14310 (libgcc_shift_count_mode): Likewise.
14311 (unwind_word_mode): Likewise.
14312 (addr_space.pointer_mode): Likewise.
14313 (addr_space.address_mode): Likewise.
14314 * doc/tm.texi: Regenerate.
14315 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
14317 (do_jump): Use scalar_int_mode for local variables.
14318 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
14319 rather than a machine_mode.
14320 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
14321 (scompare_loc_descriptor_wide): Likewise.
14322 (scompare_loc_descriptor_narrow): Likewise.
14323 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
14325 * except.c (sjlj_emit_dispatch_table): Likewise.
14326 (expand_builtin_eh_copy_values): Likewise.
14327 * explow.c (convert_memory_address_addr_space_1): Likewise.
14328 Take a scalar_int_mode rather than a machine_mode.
14329 (convert_memory_address_addr_space): Take a scalar_int_mode rather
14330 than a machine_mode.
14331 (memory_address_addr_space): Use scalar_int_mode for local variables.
14332 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
14333 rather than a machine_mode.
14334 * expmed.c (mask_rtx): Likewise.
14335 (init_expmed_one_conv): Likewise.
14336 (expand_mult_highpart_adjust): Likewise.
14337 (extract_high_half): Likewise.
14338 (expmed_mult_highpart_optab): Likewise.
14339 (expmed_mult_highpart): Likewise.
14340 (expand_smod_pow2): Likewise.
14341 (expand_sdiv_pow2): Likewise.
14342 (emit_store_flag_int): Likewise.
14343 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
14345 (extract_low_bits): Likewise.
14346 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
14348 * expr.c (pieces_addr::adjust): Likewise.
14349 (can_store_by_pieces): Likewise.
14350 (store_by_pieces): Likewise.
14351 (clear_by_pieces_1): Likewise.
14352 (expand_expr_addr_expr_1): Likewise.
14353 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
14354 (expand_expr_real_1): Likewise.
14355 (try_casesi): Likewise.
14356 * final.c (shorten_branches): Likewise.
14357 * fold-const.c (fold_convert_const_int_from_fixed): Change the
14358 type of "mode" to machine_mode.
14359 * internal-fn.c (expand_arith_overflow_result_store): Take a
14360 scalar_int_mode rather than a machine_mode.
14361 (expand_mul_overflow): Use scalar_int_mode for local variables.
14362 * loop-doloop.c (doloop_modify): Likewise.
14363 (doloop_optimize): Likewise.
14364 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
14365 than a machine_mode.
14366 (expand_doubleword_shift_condmove): Likewise.
14367 (expand_doubleword_shift): Likewise.
14368 (expand_doubleword_clz): Likewise.
14369 (expand_doubleword_popcount): Likewise.
14370 (expand_doubleword_parity): Likewise.
14371 (expand_absneg_bit): Use scalar_int_mode for local variables.
14372 (prepare_float_lib_cmp): Likewise.
14373 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
14374 rather than a machine_mode.
14375 (convert_memory_address_addr_space): Likewise.
14376 (get_mode_bounds): Likewise.
14377 (get_address_mode): Return a scalar_int_mode rather than a
14379 * rtlanal.c (get_address_mode): Likewise.
14380 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
14381 than a machine_mode.
14382 * targhooks.c (default_mode_rep_extended): Likewise.
14383 (default_valid_pointer_mode): Likewise.
14384 (default_addr_space_valid_pointer_mode): Likewise.
14385 (default_eh_return_filter_mode): Return a scalar_int_mode rather
14386 than a machine_mode.
14387 (default_libgcc_cmp_return_mode): Likewise.
14388 (default_libgcc_shift_count_mode): Likewise.
14389 (default_unwind_word_mode): Likewise.
14390 (default_addr_space_pointer_mode): Likewise.
14391 (default_addr_space_address_mode): Likewise.
14392 * targhooks.h (default_eh_return_filter_mode): Likewise.
14393 (default_libgcc_cmp_return_mode): Likewise.
14394 (default_libgcc_shift_count_mode): Likewise.
14395 (default_unwind_word_mode): Likewise.
14396 (default_addr_space_pointer_mode): Likewise.
14397 (default_addr_space_address_mode): Likewise.
14398 (default_mode_rep_extended): Take a scalar_int_mode rather than
14400 (default_valid_pointer_mode): Likewise.
14401 (default_addr_space_valid_pointer_mode): Likewise.
14402 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
14404 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
14405 rather than a machine_mode.
14406 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
14407 for local variables.
14408 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14409 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
14410 than a machine_mode.
14412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14413 Alan Hayward <alan.hayward@arm.com>
14414 David Sherwood <david.sherwood@arm.com>
14416 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
14417 the mode argument to scalar_int_mode.
14418 (do_jump_by_parts_zero_rtx): Likewise.
14419 (do_jump_by_parts_equality_rtx): Likewise.
14420 (do_jump_by_parts_greater): Take a mode argument.
14421 (do_jump_by_parts_equality): Likewise.
14422 (do_jump_1): Update calls accordingly.
14424 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14425 Alan Hayward <alan.hayward@arm.com>
14426 David Sherwood <david.sherwood@arm.com>
14428 * is-a.h (safe_dyn_cast): New function.
14429 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
14430 (jump_table_for_label): Likewise.
14431 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
14432 instead of an rtx_insn *.
14433 (shorten_branches): Use dyn_cast instead of LABEL_P and
14434 JUMP_TABLE_DATA_P. Use jump_table_for_label and
14435 rtx_jump_table_data::get_data_mode.
14436 (final_scan_insn): Likewise.
14438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14439 Alan Hayward <alan.hayward@arm.com>
14440 David Sherwood <david.sherwood@arm.com>
14442 * combine.c (try_combine): Use is_a <scalar_int_mode> when
14443 trying to combine a full-register integer set with a subreg
14446 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14447 Alan Hayward <alan.hayward@arm.com>
14448 David Sherwood <david.sherwood@arm.com>
14450 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
14451 that is always either address_mode or pointer_mode.
14453 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14454 Alan Hayward <alan.hayward@arm.com>
14455 David Sherwood <david.sherwood@arm.com>
14457 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
14458 when the two are known to be equal.
14460 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14461 Alan Hayward <alan.hayward@arm.com>
14462 David Sherwood <david.sherwood@arm.com>
14464 * simplify-rtx.c (simplify_const_unary_operation): Use
14465 is_a <scalar_int_mode> instead of checking for a nonzero
14466 precision. Forcibly convert op_mode to a scalar_int_mode
14467 in that case. More clearly differentiate the operand and
14468 result modes and use the former when deciding what the value
14469 of a count-bits operation should be. Use is_int_mode instead
14470 of checking for a MODE_INT. Remove redundant check for whether
14471 this mode has a zero precision.
14473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14474 Alan Hayward <alan.hayward@arm.com>
14475 David Sherwood <david.sherwood@arm.com>
14477 * optabs.c (widen_leading): Change the type of the mode argument
14478 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
14479 (widen_bswap): Likewise.
14480 (expand_parity): Likewise.
14481 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
14482 (expand_ffs): Likewise.
14483 (epand_unop): Check for scalar integer modes before calling the
14486 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14487 Alan Hayward <alan.hayward@arm.com>
14488 David Sherwood <david.sherwood@arm.com>
14490 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
14492 (expand_expr_real_1): Update call accordingly.
14494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14495 Alan Hayward <alan.hayward@arm.com>
14496 David Sherwood <david.sherwood@arm.com>
14498 * expmed.c (store_bit_field_using_insv): Add op0_mode and
14499 value_mode arguments. Use scalar_int_mode internally.
14500 (store_bit_field_1): Rename the new integer mode from imode
14501 to op0_mode and use it instead of GET_MODE (op0). Update calls
14502 to store_split_bit_field, store_bit_field_using_insv and
14503 store_fixed_bit_field.
14504 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
14505 Use scalar_int_mode internally. Use a bit count rather than a mode
14506 when calculating the largest bit size for get_best_mode.
14507 Update calls to store_split_bit_field and store_fixed_bit_field_1.
14508 (store_fixed_bit_field_1): Add mode and value_mode arguments.
14509 Remove assertion that OP0 has a scalar integer mode.
14510 (store_split_bit_field): Add op0_mode and value_mode arguments.
14511 Update calls to extract_fixed_bit_field.
14512 (extract_bit_field_using_extv): Add an op0_mode argument.
14513 Use scalar_int_mode internally.
14514 (extract_bit_field_1): Rename the new integer mode from imode to
14515 op0_mode and use it instead of GET_MODE (op0). Update calls to
14516 extract_split_bit_field, extract_bit_field_using_extv and
14517 extract_fixed_bit_field.
14518 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
14519 to extract_split_bit_field and extract_fixed_bit_field_1.
14520 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
14521 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
14522 on the target mode.
14523 (extract_split_bit_field): Add an op0_mode argument. Update call
14524 to extract_fixed_bit_field.
14526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14527 Alan Hayward <alan.hayward@arm.com>
14528 David Sherwood <david.sherwood@arm.com>
14530 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
14531 * explow.c (hard_function_value): Likewise.
14532 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
14533 convert_to_mode call outside the loop.
14534 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
14535 for the mode iterator. Require the mode specified by max_pieces
14537 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
14539 (copy_blkmode_to_reg): Likewise.
14540 (set_storage_via_setmem): Likewise.
14541 * optabs.c (prepare_cmp_insn): Likewise.
14542 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14543 * stor-layout.c (finish_bitfield_representative): Likewise.
14545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14546 Alan Hayward <alan.hayward@arm.com>
14547 David Sherwood <david.sherwood@arm.com>
14549 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
14550 * expr.c (convert_move): Use them.
14551 (convert_modes): Likewise.
14552 (store_expr_with_bounds): Likewise.
14554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14555 Alan Hayward <alan.hayward@arm.com>
14556 David Sherwood <david.sherwood@arm.com>
14558 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
14559 parameter for the mode of "x". Remove the "known_x", "known_mode"
14560 and "known_ret" arguments. Change the type of the mode argument
14561 to scalar_int_mode.
14562 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
14563 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
14564 (reg_num_sign_bit_copies_for_combine): Likewise.
14565 * rtlanal.c (nonzero_bits1): Likewise.
14566 (num_sign_bit_copies1): Likewise.
14567 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
14568 (reg_num_sign_bit_copies_general): Likewise.
14569 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
14570 (reg_nonzero_bits_general): Likewise.
14572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14573 Alan Hayward <alan.hayward@arm.com>
14574 David Sherwood <david.sherwood@arm.com>
14576 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
14577 than in subroutines. Return 1 for non-integer modes.
14578 (cached_num_sign_bit_copies): Change the type of the mode parameter
14579 to scalar_int_mode.
14580 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
14581 classes. Handle CONST_INT_P first and then check whether X also
14582 has a scalar integer mode. Check the same thing for inner registers
14583 of a SUBREG and for values that are being extended or truncated.
14585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14586 Alan Hayward <alan.hayward@arm.com>
14587 David Sherwood <david.sherwood@arm.com>
14589 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
14590 in subroutines. Return the mode mask for non-integer modes.
14591 (cached_nonzero_bits): Change the type of the mode parameter
14592 to scalar_int_mode.
14593 (nonzero_bits1): Likewise. Remove early exit for other mode
14594 classes. Handle CONST_INT_P first and then check whether X
14595 also has a scalar integer mode.
14597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14598 Alan Hayward <alan.hayward@arm.com>
14599 David Sherwood <david.sherwood@arm.com>
14601 * expr.c (widest_int_mode_for_size): Make the comment match the code.
14602 Return a scalar_int_mode and assert that the size is greater than
14604 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
14605 (op_by_pieces_d::op_by_pieces_d): Likewise.
14606 (op_by_pieces_d::run): Likewise.
14607 (can_store_by_pieces): Likewise.
14609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14610 Alan Hayward <alan.hayward@arm.com>
14611 David Sherwood <david.sherwood@arm.com>
14613 * combine.c (extract_left_shift): Add a mode argument and update
14615 (make_compound_operation_int): Change the type of the mode parameter
14616 to scalar_int_mode and update the call to extract_left_shift.
14618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14619 Alan Hayward <alan.hayward@arm.com>
14620 David Sherwood <david.sherwood@arm.com>
14622 * combine.c (simplify_and_const_int): Change the type of the mode
14623 parameter to scalar_int_mode.
14624 (simplify_and_const_int_1): Likewise. Update recursive call.
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 * combine.c (simplify_compare_const): Check that the mode is a
14631 scalar_int_mode (rather than VOIDmode) before testing its
14633 (simplify_comparison): Move COMPARISON_P handling out of the
14634 loop and restrict the latter part of the loop to scalar_int_modes.
14635 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
14636 and when considering SUBREG_REGs. Use is_int_mode instead of
14637 checking GET_MODE_CLASS against MODE_INT.
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 * combine.c (try_widen_shift_mode): Move check for equal modes to...
14644 (simplify_shift_const_1): ...here. Use scalar_int_mode for
14645 shift_unit_mode and for modes involved in scalar shifts.
14647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14648 Alan Hayward <alan.hayward@arm.com>
14649 David Sherwood <david.sherwood@arm.com>
14651 * combine.c (force_int_to_mode): New function, split out from...
14652 (force_to_mode): ...here. Keep xmode up-to-date and use it
14653 instead of GET_MODE (x).
14655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14656 Alan Hayward <alan.hayward@arm.com>
14657 David Sherwood <david.sherwood@arm.com>
14659 * optabs-query.h (extraction_insn::struct_mode): Change type to
14660 opt_scalar_int_mode and update comment.
14661 (extraction_insn::field_mode): Change type to scalar_int_mode.
14662 (extraction_insn::pos_mode): Likewise.
14663 * combine.c (make_extraction): Update accordingly.
14664 * optabs-query.c (get_traditional_extraction_insn): Likewise.
14665 (get_optab_extraction_insn): Likewise.
14666 * recog.c (simplify_while_replacing): Likewise.
14667 * expmed.c (narrow_bit_field_mem): Change the type of the mode
14668 parameter to opt_scalar_int_mode.
14670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14671 Alan Hayward <alan.hayward@arm.com>
14672 David Sherwood <david.sherwood@arm.com>
14674 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
14675 to a scalar_int_mode instead of a machine_mode.
14676 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
14677 (get_best_mode): Return a boolean and use a pointer argument to store
14678 the selected mode. Replace the limit mode parameter with a bit limit.
14679 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
14680 for the values returned by bit_field_mode_iterator::next_mode.
14681 (store_bit_field): Update call to get_best_mode.
14682 (store_fixed_bit_field): Likewise.
14683 (extract_fixed_bit_field): Likewise.
14684 * expr.c (optimize_bitfield_assignment_op): Likewise.
14685 * fold-const.c (optimize_bit_field_compare): Likewise.
14686 (fold_truth_andor_1): Likewise.
14687 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
14688 Update for new type of m_mode.
14689 (get_best_mode): As above.
14691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14692 Alan Hayward <alan.hayward@arm.com>
14693 David Sherwood <david.sherwood@arm.com>
14695 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
14696 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
14697 (store_bit_field): Check is_a <scalar_int_mode> before calling
14698 strict_volatile_bitfield_p.
14699 (extract_bit_field): Likewise.
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 * target.def (cstore_mode): Return a scalar_int_mode.
14706 * doc/tm.texi: Regenerate.
14707 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
14708 * targhooks.h (default_cstore_mode): Likewise.
14709 * targhooks.c (default_cstore_mode): Likewise, using a forced
14711 * expmed.c (emit_cstore): Expect the target of the cstore to be
14714 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14715 Alan Hayward <alan.hayward@arm.com>
14716 David Sherwood <david.sherwood@arm.com>
14718 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
14720 (niter_desc): Likewise mode.
14721 (iv_analyze): Add a mode parameter.
14723 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
14724 and change its type to scalar_int_mode.
14725 * loop-iv.c: Update commentary at head of file.
14726 (iv_constant): Pass the mode paraeter before the rtx it describes
14727 and change its type to scalar_int_mode. Remove VOIDmode handling.
14728 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
14729 (iv_extend): Likewise.
14730 (shorten_into_mode): Likewise.
14731 (iv_add): Use scalar_int_mode.
14732 (iv_mult): Likewise.
14733 (iv_shift): Likewise.
14734 (canonicalize_iv_subregs): Likewise.
14735 (get_biv_step_1): Pass the outer_mode parameter before the rtx
14736 it describes and change its mode to scalar_int_mode. Also change
14737 the type of the returned inner_mode to scalar_int_mode.
14738 (get_biv_step): Likewise, turning outer_mode from a pointer
14739 into a direct parameter. Update call to get_biv_step_1.
14740 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
14741 iv_constant and get_biv_step.
14742 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
14743 and change its type to scalar_int_mode. Don't initialise iv->mode
14744 to VOIDmode and remove later checks for its still being VOIDmode.
14745 Update calls to iv_analyze_op and iv_analyze_expr. Check
14746 is_a <scalar_int_mode> when changing the mode under consideration.
14747 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
14748 Update call to iv_analyze_expr.
14749 (iv_analyze_op): Add a mode parameter. Reject subregs whose
14750 inner register is not also a scalar_int_mode. Update call to
14752 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
14753 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
14754 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
14755 separate mode class checks. Update calls to iv_analyze. Remove
14756 fix-up of VOIDmodes after iv_analyze_biv.
14757 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
14758 don't have a scalar_int_mode. Update call to biv_p.
14760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14761 Alan Hayward <alan.hayward@arm.com>
14762 David Sherwood <david.sherwood@arm.com>
14764 * cfgexpand.c (convert_debug_memory_address): Use
14765 as_a <scalar_int_mode>.
14766 * combine.c (expand_compound_operation): Likewise.
14767 (make_extraction): Likewise.
14768 (change_zero_ext): Likewise.
14769 (simplify_comparison): Likewise.
14770 * cse.c (cse_insn): Likewise.
14771 * dwarf2out.c (minmax_loc_descriptor): Likewise.
14772 (mem_loc_descriptor): Likewise.
14773 (loc_descriptor): Likewise.
14774 * expmed.c (init_expmed_one_mode): Likewise.
14775 (synth_mult): Likewise.
14776 (emit_store_flag_1): Likewise.
14777 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
14778 of a comparison with size.
14779 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
14780 (reduce_to_bit_field_precision): Likewise.
14781 * function.c (expand_function_end): Likewise.
14782 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
14783 * loop-doloop.c (doloop_modify): Likewise.
14784 * optabs.c (expand_binop): Likewise.
14785 (expand_unop): Likewise.
14786 (expand_copysign_absneg): Likewise.
14787 (prepare_cmp_insn): Likewise.
14788 (maybe_legitimize_operand): Likewise.
14789 * recog.c (const_scalar_int_operand): Likewise.
14790 * rtlanal.c (get_address_mode): Likewise.
14791 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14792 (simplify_cond_clz_ctz): Likewise.
14793 * tree-nested.c (get_nl_goto_field): Likewise.
14794 * tree.c (build_vector_type_for_mode): Likewise.
14795 * var-tracking.c (use_narrower_mode): Likewise.
14797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14798 Alan Hayward <alan.hayward@arm.com>
14799 David Sherwood <david.sherwood@arm.com>
14801 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
14802 * builtins.c (expand_builtin_signbit): Use it.
14803 * cfgexpand.c (expand_debug_expr): Likewise.
14804 * dojump.c (do_jump): Likewise.
14805 (do_compare_and_jump): Likewise.
14806 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
14807 * expmed.c (make_tree): Likewise.
14808 * expr.c (expand_expr_real_2): Likewise.
14809 (expand_expr_real_1): Likewise.
14810 (try_casesi): Likewise.
14811 * fold-const-call.c (fold_const_call_ss): Likewise.
14812 * fold-const.c (unextend): Likewise.
14813 (extract_muldiv_1): Likewise.
14814 (fold_single_bit_test): Likewise.
14815 (native_encode_int): Likewise.
14816 (native_encode_string): Likewise.
14817 (native_interpret_int): Likewise.
14818 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
14819 * internal-fn.c (expand_addsub_overflow): Likewise.
14820 (expand_neg_overflow): Likewise.
14821 (expand_mul_overflow): Likewise.
14822 (expand_arith_overflow): Likewise.
14823 * match.pd: Likewise.
14824 * stor-layout.c (layout_type): Likewise.
14825 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14826 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
14827 * tree-ssanames.c (get_range_info): Likewise.
14828 * tree-switch-conversion.c (array_value_type) Likewise.
14829 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
14830 (vect_recog_divmod_pattern): Likewise.
14831 (vect_recog_mixed_size_cond_pattern): Likewise.
14832 * tree-vrp.c (extract_range_basic): Likewise.
14833 (simplify_float_conversion_using_ranges): Likewise.
14834 * tree.c (int_fits_type_p): Likewise.
14835 * ubsan.c (instrument_bool_enum_load): Likewise.
14836 * varasm.c (mergeable_string_section): Likewise.
14837 (narrowing_initializer_constant_valid_p): Likewise.
14838 (output_constant): Likewise.
14840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14841 Alan Hayward <alan.hayward@arm.com>
14842 David Sherwood <david.sherwood@arm.com>
14844 * machmode.h (NARROWEST_INT_MODE): New macro.
14845 * expr.c (alignment_for_piecewise_move): Use it instead of
14846 GET_CLASS_NARROWEST_MODE (MODE_INT).
14847 (push_block): Likewise.
14848 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
14850 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14852 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14853 Alan Hayward <alan.hayward@arm.com>
14854 David Sherwood <david.sherwood@arm.com>
14856 * postreload.c (move2add_valid_value_p): Change the type of the
14857 mode parameter to scalar_int_mode.
14858 (move2add_use_add2_insn): Add a mode parameter and use it instead
14860 (move2add_use_add3_insn): Likewise.
14861 (reload_cse_move2add): Update accordingly.
14863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14864 Alan Hayward <alan.hayward@arm.com>
14865 David Sherwood <david.sherwood@arm.com>
14867 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
14869 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
14870 * optabs.c (expand_unop): Likewise.
14872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14873 Alan Hayward <alan.hayward@arm.com>
14874 David Sherwood <david.sherwood@arm.com>
14876 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
14877 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
14878 (popcount_loc_descriptor): Likewise.
14879 (bswap_loc_descriptor): Likewise.
14880 (rotate_loc_descriptor): Likewise.
14881 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
14882 calling the functions above.
14884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14885 Alan Hayward <alan.hayward@arm.com>
14886 David Sherwood <david.sherwood@arm.com>
14888 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
14890 (try_combine): Likewise.
14891 (simplify_if_then_else): Likewise.
14892 * cse.c (cse_insn): Likewise.
14893 * dwarf2out.c (mem_loc_descriptor): Likewise.
14894 * emit-rtl.c (gen_lowpart_common): Likewise.
14895 * simplify-rtx.c (simplify_truncation): Likewise.
14896 (simplify_binary_operation_1): Likewise.
14897 (simplify_const_relational_operation): Likewise.
14898 (simplify_ternary_operation): Likewise.
14899 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
14901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14902 Alan Hayward <alan.hayward@arm.com>
14903 David Sherwood <david.sherwood@arm.com>
14905 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
14906 * reload.c (push_reload): Likewise.
14907 (find_reloads): Likewise.
14909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14910 Alan Hayward <alan.hayward@arm.com>
14911 David Sherwood <david.sherwood@arm.com>
14913 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
14914 (make_compound_operation_int): Likewise.
14915 (change_zero_ext): Likewise.
14916 * expr.c (convert_move): Likewise.
14917 (convert_modes): Likewise.
14918 * fwprop.c (forward_propagate_subreg): Likewise.
14919 * loop-iv.c (get_biv_step_1): Likewise.
14920 * optabs.c (widen_operand): Likewise.
14921 * postreload.c (move2add_valid_value_p): Likewise.
14922 * recog.c (simplify_while_replacing): Likewise.
14923 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14924 (simplify_binary_operation_1): Likewise. Remove redundant
14925 mode equality check.
14927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14928 Alan Hayward <alan.hayward@arm.com>
14929 David Sherwood <david.sherwood@arm.com>
14931 * combine.c (combine_simplify_rtx): Add checks for
14932 is_a <scalar_int_mode>.
14933 (simplify_if_then_else): Likewise.
14934 (make_field_assignment): Likewise.
14935 (simplify_comparison): Likewise.
14936 * ifcvt.c (noce_try_bitop): Likewise.
14937 * loop-invariant.c (canonicalize_address_mult): Likewise.
14938 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14940 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14941 Alan Hayward <alan.hayward@arm.com>
14942 David Sherwood <david.sherwood@arm.com>
14944 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
14945 is_a <scalar_int_mode> instead of != BLKmode.
14947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14948 Alan Hayward <alan.hayward@arm.com>
14949 David Sherwood <david.sherwood@arm.com>
14951 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
14952 instead of != VOIDmode.
14953 * combine.c (if_then_else_cond): Likewise.
14954 (change_zero_ext): Likewise.
14955 * dwarf2out.c (mem_loc_descriptor): Likewise.
14956 (loc_descriptor): Likewise.
14957 * rtlanal.c (canonicalize_condition): Likewise.
14958 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
14960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14961 Alan Hayward <alan.hayward@arm.com>
14962 David Sherwood <david.sherwood@arm.com>
14964 * simplify-rtx.c (simplify_binary_operation_1): Use
14965 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
14967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14968 Alan Hayward <alan.hayward@arm.com>
14969 David Sherwood <david.sherwood@arm.com>
14971 * wide-int.h (int_traits<unsigned char>) New class.
14972 (int_traits<unsigned short>) Likewise.
14973 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
14974 Use GET_MODE_UNIT_PRECISION and remove redundant test for
14976 * combine.c (set_nonzero_bits_and_sign_copies): Use
14977 is_a <scalar_int_mode>.
14978 (find_split_point): Likewise.
14979 (combine_simplify_rtx): Likewise.
14980 (simplify_logical): Likewise.
14981 (expand_compound_operation): Likewise.
14982 (expand_field_assignment): Likewise.
14983 (make_compound_operation): Likewise.
14984 (extended_count): Likewise.
14985 (change_zero_ext): Likewise.
14986 (simplify_comparison): Likewise.
14987 * dwarf2out.c (scompare_loc_descriptor): Likewise.
14988 (ucompare_loc_descriptor): Likewise.
14989 (minmax_loc_descriptor): Likewise.
14990 (mem_loc_descriptor): Likewise.
14991 (loc_descriptor): Likewise.
14992 * expmed.c (init_expmed_one_mode): Likewise.
14993 * lra-constraints.c (lra_constraint_offset): Likewise.
14994 * optabs.c (prepare_libcall_arg): Likewise.
14995 * postreload.c (move2add_note_store): Likewise.
14996 * reload.c (operands_match_p): Likewise.
14997 * rtl.h (load_extend_op): Likewise.
14998 * rtlhooks.c (gen_lowpart_general): Likewise.
14999 * simplify-rtx.c (simplify_truncation): Likewise.
15000 (simplify_unary_operation_1): Likewise.
15001 (simplify_binary_operation_1): Likewise.
15002 (simplify_const_binary_operation): Likewise.
15003 (simplify_const_relational_operation): Likewise.
15004 (simplify_subreg): Likewise.
15005 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15006 * var-tracking.c (adjust_mems): Likewise.
15007 (prepare_call_arguments): Likewise.
15009 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15010 Alan Hayward <alan.hayward@arm.com>
15011 David Sherwood <david.sherwood@arm.com>
15013 * machmode.h (is_int_mode): New fuction.
15014 * combine.c (find_split_point): Use it.
15015 (combine_simplify_rtx): Likewise.
15016 (simplify_if_then_else): Likewise.
15017 (simplify_set): Likewise.
15018 (simplify_shift_const_1): Likewise.
15019 (simplify_comparison): Likewise.
15020 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15021 * cse.c (notreg_cost): Likewise.
15022 (cse_insn): Likewise.
15023 * cselib.c (cselib_lookup_1): Likewise.
15024 * dojump.c (do_jump_1): Likewise.
15025 (do_compare_rtx_and_jump): Likewise.
15026 * dse.c (get_call_args): Likewise.
15027 * dwarf2out.c (rtl_for_decl_init): Likewise.
15028 (native_encode_initializer): Likewise.
15029 * expmed.c (emit_store_flag_1): Likewise.
15030 (emit_store_flag): Likewise.
15031 * expr.c (convert_modes): Likewise.
15032 (store_field): Likewise.
15033 (expand_expr_real_1): Likewise.
15034 * fold-const.c (fold_read_from_constant_string): Likewise.
15035 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15036 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15037 * optabs.c (expand_binop): Likewise.
15038 (expand_unop): Likewise.
15039 (expand_abs_nojump): Likewise.
15040 (expand_one_cmpl_abs_nojump): Likewise.
15041 * simplify-rtx.c (mode_signbit_p): Likewise.
15042 (val_signbit_p): Likewise.
15043 (val_signbit_known_set_p): Likewise.
15044 (val_signbit_known_clear_p): Likewise.
15045 (simplify_relational_operation_1): Likewise.
15046 * tree.c (vector_type_mode): Likewise.
15048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15049 Alan Hayward <alan.hayward@arm.com>
15050 David Sherwood <david.sherwood@arm.com>
15052 * machmode.h (smallest_mode_for_size): Fix formatting.
15053 (smallest_int_mode_for_size): New function.
15054 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15055 instead of smallest_mode_for_size.
15056 * combine.c (make_extraction): Likewise.
15057 * config/arc/arc.c (arc_expand_movmem): Likewise.
15058 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15059 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15060 * config/s390/s390.c (s390_expand_insv): Likewise.
15061 * config/sparc/sparc.c (assign_int_registers): Likewise.
15062 * config/spu/spu.c (spu_function_value): Likewise.
15063 (spu_function_arg): Likewise.
15064 * coverage.c (get_gcov_type): Likewise.
15065 (get_gcov_unsigned_t): Likewise.
15066 * dse.c (find_shift_sequence): Likewise.
15067 * expmed.c (store_bit_field_1): Likewise.
15068 * expr.c (convert_move): Likewise.
15069 (store_field): Likewise.
15070 * internal-fn.c (expand_arith_overflow): Likewise.
15071 * optabs-query.c (get_best_extraction_insn): Likewise.
15072 * optabs.c (expand_twoval_binop_libfunc): Likewise.
15073 * stor-layout.c (layout_type): Likewise.
15074 (initialize_sizetypes): Likewise.
15075 * targhooks.c (default_get_mask_mode): Likewise.
15076 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
15078 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15079 Alan Hayward <alan.hayward@arm.com>
15080 David Sherwood <david.sherwood@arm.com>
15082 * machmode.h (opt_mode::else_blk): New function.
15083 (int_mode_for_mode): Declare.
15084 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
15085 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
15087 * cfgexpand.c (expand_debug_expr): Likewise.
15088 * combine.c (gen_lowpart_or_truncate): Likewise.
15089 (gen_lowpart_for_combine): Likewise.
15090 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
15091 * config/avr/avr.c (avr_to_int_mode): Likewise.
15092 (avr_out_plus_1): Likewise.
15093 (avr_out_plus): Likewise.
15094 (avr_out_round): Likewise.
15095 * config/i386/i386.c (ix86_split_to_parts): Likewise.
15096 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15097 (s390_expand_vcond): Likewise.
15098 * config/spu/spu.c (spu_split_immediate): Likewise.
15099 (spu_expand_mov): Likewise.
15100 * dse.c (get_stored_val): Likewise.
15101 * expmed.c (store_bit_field_1): Likewise.
15102 (convert_extracted_bit_field): Use int_mode_for_mode instead of
15104 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
15105 (extract_low_bits): Likewise.
15106 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
15107 handling rather than repeating the check.
15108 (emit_group_store): Likewise.
15109 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
15110 * optabs.c (expand_absneg_bit): Likewise.
15111 (expand_copysign_absneg): Likewise.
15112 (expand_copysign_bit): Likewise.
15113 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
15114 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
15115 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
15116 * var-tracking.c (prepare_call_arguments): Likewise.
15117 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
15118 int_mode_for_mode instead of mode_for_size.
15119 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15122 Alan Hayward <alan.hayward@arm.com>
15123 David Sherwood <david.sherwood@arm.com>
15125 * machmode.h (int_mode_for_size): New function.
15126 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
15127 instead of mode_for_size.
15128 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
15130 * combine.c (expand_field_assignment): Use int_mode_for_size
15131 instead of mode_for_size.
15132 (make_extraction): Likewise.
15133 (simplify_shift_const_1): Likewise.
15134 (simplify_comparison): Likewise.
15135 * dojump.c (do_jump): Likewise.
15136 * dwarf2out.c (mem_loc_descriptor): Likewise.
15137 * emit-rtl.c (init_derived_machine_modes): Likewise.
15138 * expmed.c (flip_storage_order): Likewise.
15139 (convert_extracted_bit_field): Likewise.
15140 * expr.c (copy_blkmode_from_reg): Likewise.
15141 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
15142 * internal-fn.c (expand_mul_overflow): Likewise.
15143 * lower-subreg.c (simple_move): Likewise.
15144 * optabs-libfuncs.c (init_optabs): Likewise.
15145 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15146 * tree.c (vector_type_mode): Likewise.
15147 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
15148 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
15149 * tree-vect-generic.c (expand_vector_parallel): Likewise.
15150 * tree-vect-stmts.c (vectorizable_load): Likewise.
15151 (vectorizable_store): Likewise.
15153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15154 Alan Hayward <alan.hayward@arm.com>
15155 David Sherwood <david.sherwood@arm.com>
15157 * coretypes.h (pod_mode): New type.
15158 (scalar_int_mode_pod): New typedef.
15159 * machmode.h (pod_mode): New class.
15160 (int_n_data_t::m): Change type to scalar_int_mode_pod.
15161 * genmodes.c (emit_mode_int_n): Update accordingly.
15162 * lower-subreg.h (target_lower_subreg): Change type to
15163 scalar_int_mode_pod.
15164 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
15165 scalar_int_mode_pod.
15167 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15168 Alan Hayward <alan.hayward@arm.com>
15169 David Sherwood <david.sherwood@arm.com>
15171 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
15172 machine_mode to scalar_int_mode.
15173 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
15174 (rs6000_option_override_internal): Remove cast to int.
15175 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
15176 machine_mode to scalar_int_mode.
15177 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
15178 (rs6000_option_override_internal): Remove cast to int.
15179 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
15180 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
15182 * config/s390/s390.c (s390_expand_builtin): Likewise.
15183 * coretypes.h (scalar_int_mode): New type.
15184 (opt_scalar_int_mode): New typedef.
15185 * machmode.h (scalar_int_mode): New class.
15186 (scalar_int_mode::includes_p): New function.
15187 (byte_mode): Change type to scalar_int_mode.
15188 (word_mode): Likewise.
15189 (ptr_mode): Likewise.
15190 * emit-rtl.c (byte_mode): Likewise.
15191 (word_mode): Likewise.
15192 (ptr_mode): Likewise.
15193 (init_derived_machine_modes): Update accordingly.
15194 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
15195 and MODE_PARTIAL_INT.
15196 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
15197 opt_scalar_int_mode.
15199 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15200 Alan Hayward <alan.hayward@arm.com>
15201 David Sherwood <david.sherwood@arm.com>
15203 * target.def (libgcc_floating_mode_supported_p): Take a
15205 * doc/tm.texi: Regenerate.
15206 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
15208 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15209 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
15212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15213 Alan Hayward <alan.hayward@arm.com>
15214 David Sherwood <david.sherwood@arm.com>
15216 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
15217 * doc/tm.texi: Regenerate.
15218 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
15219 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
15220 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
15221 * targhooks.h (default_floatn_mode): Likewise.
15222 * targhooks.c (default_floatn_mode): Likewise.
15223 * tree.c (build_common_tree_nodes): Update accordingly.
15225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15226 Alan Hayward <alan.hayward@arm.com>
15227 David Sherwood <david.sherwood@arm.com>
15229 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
15230 (mode_iterator::iterate_p): Likewise.
15231 (mode_iterator::get_wider): Likewise.
15232 * expr.c (init_expr_target): Use opt_scalar_float_mode.
15234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15235 Alan Hayward <alan.hayward@arm.com>
15236 David Sherwood <david.sherwood@arm.com>
15238 * coretypes.h (opt_scalar_float_mode): New typedef.
15239 * machmode.h (float_mode_for_size): New function.
15240 * emit-rtl.c (double_mode): Delete.
15241 (init_emit_once): Use float_mode_for_size.
15242 * stor-layout.c (layout_type): Likewise.
15243 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
15245 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15246 Alan Hayward <alan.hayward@arm.com>
15247 David Sherwood <david.sherwood@arm.com>
15249 * output.h (assemble_real): Take a scalar_float_mode.
15250 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
15251 * config/arm/arm.md (consttable_4): Likewise.
15252 (consttable_8): Likewise.
15253 (consttable_16): Likewise.
15254 * config/mips/mips.md (consttable_float): Likewise.
15255 * config/s390/s390.c (s390_output_pool_entry): Likewise.
15256 * varasm.c (assemble_real): Take a scalar_float_mode.
15257 (output_constant_pool_2): Update accordingly.
15258 (output_constant): Likewise.
15260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15261 Alan Hayward <alan.hayward@arm.com>
15262 David Sherwood <david.sherwood@arm.com>
15264 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
15265 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
15266 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
15267 (native_encode_real): Likewise.
15268 (native_interpret_real): Likewise.
15269 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
15270 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15273 Alan Hayward <alan.hayward@arm.com>
15274 David Sherwood <david.sherwood@arm.com>
15276 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
15277 <scalar_float_mode>. Simplify.
15278 (gen_extend_conv_libfunc): Likewise.
15280 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15281 Alan Hayward <alan.hayward@arm.com>
15282 David Sherwood <david.sherwood@arm.com>
15284 * coretypes.h (scalar_float_mode): New type.
15285 * machmode.h (mode_traits::from_int): Use machine_mode if
15286 USE_ENUM_MODES is defined.
15287 (is_a): New function.
15289 (dyn_cast): Likewise.
15290 (scalar_float_mode): New class.
15291 (scalar_float_mode::includes_p): New function.
15292 (is_float_mode): Likewise.
15293 * gdbhooks.py (MachineModePrinter): New class.
15294 (build_pretty_printer): Use it for scalar_float_mode.
15295 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
15296 (format_helper::format_helper): Turn into a template.
15297 * genmodes.c (get_mode_class): New function.
15298 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
15299 or machine_mode if none.
15300 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
15301 as_a <scalar_float_mode>.
15302 * dwarf2out.c (mem_loc_descriptor): Likewise.
15303 (insert_float): Likewise.
15304 (add_const_value_attribute): Likewise.
15305 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15306 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
15307 (expand_unop): Update accordingly.
15308 (expand_abs_nojump): Likewise.
15309 (expand_copysign_absneg): Take a scalar_float_mode.
15310 (expand_copysign_bit): Likewise.
15311 (expand_copysign): Update accordingly.
15313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15314 Alan Hayward <alan.hayward@arm.com>
15315 David Sherwood <david.sherwood@arm.com>
15317 * coretypes.h (opt_mode): New class.
15318 * machmode.h (opt_mode): Likewise.
15319 (opt_mode::else_void): New function.
15320 (opt_mode::require): Likewise.
15321 (opt_mode::exists): Likewise.
15322 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
15323 (GET_MODE_2XWIDER_MODE): Likewise.
15324 (mode_iterator::get_wider): Update accordingly.
15325 (mode_iterator::get_2xwider): Likewise.
15326 (mode_iterator::get_known_wider): Likewise, turning into a template.
15327 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
15328 forcing a wider mode to exist.
15329 * config/cr16/cr16.h (LONG_REG_P): Likewise.
15330 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15331 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
15332 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
15333 * lower-subreg.c (init_lower_subreg): Likewise.
15334 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
15335 on the final iteration.
15336 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
15337 a wider mode exists before asking for a move pattern.
15338 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
15339 forcing a wider mode to exist.
15340 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
15341 returning false if no such mode exists.
15342 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
15343 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
15344 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
15345 Avoid checking for a MODE_INT if we already know the mode is not a
15347 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
15348 forcing a wider mode to exist.
15349 (expmed_mult_highpart_optab): Likewise.
15350 (expmed_mult_highpart): Likewise.
15351 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
15353 * lto-streamer-in.c (lto_input_mode_table): Likewise.
15354 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
15355 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15356 * internal-fn.c (expand_mul_overflow): Update use of
15357 GET_MODE_2XWIDER_MODE.
15358 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15359 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
15360 GET_MODE_WIDER_MODE.
15361 (convert_plusminus_to_widen): Likewise.
15362 * tree-switch-conversion.c (array_value_type): Likewise.
15363 * var-tracking.c (emit_note_insn_var_location): Likewise.
15364 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15365 Return false inside rather than outside the loop if no wider mode
15367 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
15368 and GET_MODE_2XWIDER_MODE
15369 (can_compare_p): Use else_void.
15370 * gdbhooks.py (OptMachineModePrinter): New class.
15371 (build_pretty_printer): Use it for opt_mode.
15373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15374 Alan Hayward <alan.hayward@arm.com>
15375 David Sherwood <david.sherwood@arm.com>
15377 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
15378 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
15380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15381 Alan Hayward <alan.hayward@arm.com>
15382 David Sherwood <david.sherwood@arm.com>
15384 * machmode.h (mode_traits): New structure.
15385 (get_narrowest_mode): New function.
15386 (mode_iterator::start): Likewise.
15387 (mode_iterator::iterate_p): Likewise.
15388 (mode_iterator::get_wider): Likewise.
15389 (mode_iterator::get_known_wider): Likewise.
15390 (mode_iterator::get_2xwider): Likewise.
15391 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
15392 (FOR_EACH_MODE): Likewise.
15393 (FOR_EACH_MODE_FROM): Likewise.
15394 (FOR_EACH_MODE_UNTIL): Likewise.
15395 (FOR_EACH_WIDER_MODE): Likewise.
15396 (FOR_EACH_2XWIDER_MODE): Likewise.
15397 * builtins.c (expand_builtin_strlen): Use new mode iterators.
15398 * combine.c (simplify_comparison): Likewise
15399 * config/i386/i386.c (type_natural_mode): Likewise.
15400 * cse.c (cse_insn): Likewise.
15401 * dse.c (find_shift_sequence): Likewise.
15402 * emit-rtl.c (init_derived_machine_modes): Likewise.
15403 (init_emit_once): Likewise.
15404 * explow.c (hard_function_value): Likewise.
15405 * expmed.c (extract_fixed_bit_field_1): Likewise.
15406 (extract_bit_field_1): Likewise.
15407 (expand_divmod): Likewise.
15408 (emit_store_flag_1): Likewise.
15409 * expr.c (init_expr_target): Likewise.
15410 (convert_move): Likewise.
15411 (alignment_for_piecewise_move): Likewise.
15412 (widest_int_mode_for_size): Likewise.
15413 (emit_block_move_via_movmem): Likewise.
15414 (copy_blkmode_to_reg): Likewise.
15415 (set_storage_via_setmem): Likewise.
15416 (compress_float_constant): Likewise.
15417 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15418 * optabs-query.c (get_best_extraction_insn): Likewise.
15419 * optabs.c (expand_binop): Likewise.
15420 (expand_twoval_unop): Likewise.
15421 (expand_twoval_binop): Likewise.
15422 (widen_leading): Likewise.
15423 (widen_bswap): Likewise.
15424 (expand_parity): Likewise.
15425 (expand_unop): Likewise.
15426 (prepare_cmp_insn): Likewise.
15427 (prepare_float_lib_cmp): Likewise.
15428 (expand_float): Likewise.
15429 (expand_fix): Likewise.
15430 (expand_sfix_optab): Likewise.
15431 * postreload.c (move2add_use_add2_insn): Likewise.
15432 * reg-stack.c (reg_to_stack): Likewise.
15433 * reginfo.c (choose_hard_reg_mode): Likewise.
15434 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15435 * stor-layout.c (mode_for_size): Likewise.
15436 (smallest_mode_for_size): Likewise.
15437 (mode_for_vector): Likewise.
15438 (finish_bitfield_representative): Likewise.
15439 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
15440 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
15441 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15442 * var-tracking.c (prepare_call_arguments): Likewise.
15444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15445 Alan Hayward <alan.hayward@arm.com>
15446 David Sherwood <david.sherwood@arm.com>
15448 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
15449 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
15450 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
15451 * machmode.h (mode_size): Move earlier in file.
15452 (mode_precision): Likewise.
15453 (mode_inner): Likewise.
15454 (mode_nunits): Likewise.
15455 (mode_unit_size): Likewise.
15456 (unit_unit_precision): Likewise.
15457 (mode_wider): Likewise.
15458 (mode_2xwider): Likewise.
15459 (machine_mode): New class.
15460 (mode_to_bytes): New function.
15461 (mode_to_bits): Likewise.
15462 (mode_to_precision): Likewise.
15463 (mode_to_inner): Likewise.
15464 (mode_to_unit_size): Likewise.
15465 (mode_to_unit_precision): Likewise.
15466 (mode_to_nunits): Likewise.
15467 (GET_MODE_SIZE): Use mode_to_bytes.
15468 (GET_MODE_BITSIZE): Use mode_to_bits.
15469 (GET_MODE_PRECISION): Use mode_to_precision.
15470 (GET_MODE_INNER): Use mode_to_inner.
15471 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
15472 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
15473 (GET_MODE_NUNITS): Use mode_to_nunits.
15474 * system.h (ALWAYS_INLINE): New macro.
15475 * config/powerpcspe/powerpcspe-c.c
15476 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
15477 int for arg1_mode and arg2_mode.
15479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15480 Alan Hayward <alan.hayward@arm.com>
15481 David Sherwood <david.sherwood@arm.com>
15483 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15484 Prefix mode names with E_ in case statements.
15485 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15486 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
15487 (aarch64_split_simd_move): Likewise.
15488 (aarch64_gen_storewb_pair): Likewise.
15489 (aarch64_gen_loadwb_pair): Likewise.
15490 (aarch64_gen_store_pair): Likewise.
15491 (aarch64_gen_load_pair): Likewise.
15492 (aarch64_get_condition_code_1): Likewise.
15493 (aarch64_constant_pool_reload_icode): Likewise.
15494 (get_rsqrte_type): Likewise.
15495 (get_rsqrts_type): Likewise.
15496 (get_recpe_type): Likewise.
15497 (get_recps_type): Likewise.
15498 (aarch64_gimplify_va_arg_expr): Likewise.
15499 (aarch64_simd_container_mode): Likewise.
15500 (aarch64_emit_load_exclusive): Likewise.
15501 (aarch64_emit_store_exclusive): Likewise.
15502 (aarch64_expand_compare_and_swap): Likewise.
15503 (aarch64_gen_atomic_cas): Likewise.
15504 (aarch64_emit_bic): Likewise.
15505 (aarch64_emit_atomic_swap): Likewise.
15506 (aarch64_emit_atomic_load_op): Likewise.
15507 (aarch64_evpc_trn): Likewise.
15508 (aarch64_evpc_uzp): Likewise.
15509 (aarch64_evpc_zip): Likewise.
15510 (aarch64_evpc_ext): Likewise.
15511 (aarch64_evpc_rev): Likewise.
15512 (aarch64_evpc_dup): Likewise.
15513 (aarch64_gen_ccmp_first): Likewise.
15514 (aarch64_gen_ccmp_next): Likewise.
15515 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15516 (alpha_emit_xfloating_libcall): Likewise.
15517 (emit_insxl): Likewise.
15518 (alpha_arg_type): Likewise.
15519 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
15520 (arc_preferred_simd_mode): Likewise.
15521 (arc_secondary_reload): Likewise.
15522 (get_arc_condition_code): Likewise.
15523 (arc_print_operand): Likewise.
15524 (arc_legitimate_constant_p): Likewise.
15525 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15526 * config/arc/arc.md (casesi_load): Likewise.
15527 (casesi_compact_jump): Likewise.
15528 * config/arc/predicates.md (proper_comparison_operator): Likewise.
15529 (cc_use_register): Likewise.
15530 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15531 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
15532 (arm_init_iwmmxt_builtins): Likewise.
15533 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
15534 (neon_expand_vector_init): Likewise.
15535 (arm_attr_length_move_neon): Likewise.
15536 (maybe_get_arm_condition_code): Likewise.
15537 (arm_emit_vector_const): Likewise.
15538 (arm_preferred_simd_mode): Likewise.
15539 (arm_output_iwmmxt_tinsr): Likewise.
15540 (thumb1_output_casesi): Likewise.
15541 (thumb2_output_casesi): Likewise.
15542 (arm_emit_load_exclusive): Likewise.
15543 (arm_emit_store_exclusive): Likewise.
15544 (arm_expand_compare_and_swap): Likewise.
15545 (arm_evpc_neon_vuzp): Likewise.
15546 (arm_evpc_neon_vzip): Likewise.
15547 (arm_evpc_neon_vrev): Likewise.
15548 (arm_evpc_neon_vtrn): Likewise.
15549 (arm_evpc_neon_vext): Likewise.
15550 (arm_validize_comparison): Likewise.
15551 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
15552 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
15553 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
15554 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
15555 (c6x_preferred_simd_mode): Likewise.
15556 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
15557 (epiphany_rtx_costs): Likewise.
15558 * config/epiphany/predicates.md (proper_comparison_operator):
15560 * config/frv/frv.c (condexec_memory_operand): Likewise.
15561 (frv_emit_move): Likewise.
15562 (output_move_single): Likewise.
15563 (output_condmove_single): Likewise.
15564 (frv_hard_regno_mode_ok): Likewise.
15565 (frv_matching_accg_mode): Likewise.
15566 * config/h8300/h8300.c (split_adds_subs): Likewise.
15567 (h8300_rtx_costs): Likewise.
15568 (h8300_print_operand): Likewise.
15569 (compute_mov_length): Likewise.
15570 (output_logical_op): Likewise.
15571 (compute_logical_op_length): Likewise.
15572 (compute_logical_op_cc): Likewise.
15573 (h8300_shift_needs_scratch_p): Likewise.
15574 (output_a_shift): Likewise.
15575 (compute_a_shift_length): Likewise.
15576 (compute_a_shift_cc): Likewise.
15577 (expand_a_rotate): Likewise.
15578 (output_a_rotate): Likewise.
15579 * config/i386/i386.c (classify_argument): Likewise.
15580 (function_arg_advance_32): Likewise.
15581 (function_arg_32): Likewise.
15582 (function_arg_64): Likewise.
15583 (function_value_64): Likewise.
15584 (ix86_gimplify_va_arg): Likewise.
15585 (ix86_legitimate_constant_p): Likewise.
15586 (put_condition_code): Likewise.
15587 (split_double_mode): Likewise.
15588 (ix86_avx256_split_vector_move_misalign): Likewise.
15589 (ix86_expand_vector_logical_operator): Likewise.
15590 (ix86_split_idivmod): Likewise.
15591 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
15592 (ix86_build_const_vector): Likewise.
15593 (ix86_build_signbit_mask): Likewise.
15594 (ix86_match_ccmode): Likewise.
15595 (ix86_cc_modes_compatible): Likewise.
15596 (ix86_expand_branch): Likewise.
15597 (ix86_expand_sse_cmp): Likewise.
15598 (ix86_expand_sse_movcc): Likewise.
15599 (ix86_expand_int_sse_cmp): Likewise.
15600 (ix86_expand_vec_perm_vpermi2): Likewise.
15601 (ix86_expand_vec_perm): Likewise.
15602 (ix86_expand_sse_unpack): Likewise.
15603 (ix86_expand_int_addcc): Likewise.
15604 (ix86_split_to_parts): Likewise.
15605 (ix86_vectorize_builtin_gather): Likewise.
15606 (ix86_vectorize_builtin_scatter): Likewise.
15607 (avx_vpermilp_parallel): Likewise.
15608 (inline_memory_move_cost): Likewise.
15609 (ix86_tieable_integer_mode_p): Likewise.
15610 (x86_maybe_negate_const_int): Likewise.
15611 (ix86_expand_vector_init_duplicate): Likewise.
15612 (ix86_expand_vector_init_one_nonzero): Likewise.
15613 (ix86_expand_vector_init_one_var): Likewise.
15614 (ix86_expand_vector_init_concat): Likewise.
15615 (ix86_expand_vector_init_interleave): Likewise.
15616 (ix86_expand_vector_init_general): Likewise.
15617 (ix86_expand_vector_set): Likewise.
15618 (ix86_expand_vector_extract): Likewise.
15619 (emit_reduc_half): Likewise.
15620 (ix86_emit_i387_round): Likewise.
15621 (ix86_mangle_type): Likewise.
15622 (ix86_expand_round_sse4): Likewise.
15623 (expand_vec_perm_blend): Likewise.
15624 (canonicalize_vector_int_perm): Likewise.
15625 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
15626 (expand_vec_perm_1): Likewise.
15627 (expand_vec_perm_interleave3): Likewise.
15628 (expand_vec_perm_even_odd_pack): Likewise.
15629 (expand_vec_perm_even_odd_1): Likewise.
15630 (expand_vec_perm_broadcast_1): Likewise.
15631 (ix86_vectorize_vec_perm_const_ok): Likewise.
15632 (ix86_expand_vecop_qihi): Likewise.
15633 (ix86_expand_mul_widen_hilo): Likewise.
15634 (ix86_expand_sse2_abs): Likewise.
15635 (ix86_expand_pextr): Likewise.
15636 (ix86_expand_pinsr): Likewise.
15637 (ix86_preferred_simd_mode): Likewise.
15638 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
15639 * config/i386/sse.md (*andnot<mode>3): Likewise.
15640 (<mask_codefor><code><mode>3<mask_name>): Likewise.
15641 (*<code><mode>3): Likewise.
15642 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
15643 (ia64_expand_atomic_op): Likewise.
15644 (ia64_arg_type): Likewise.
15645 (ia64_mode_to_int): Likewise.
15646 (ia64_scalar_mode_supported_p): Likewise.
15647 (ia64_vector_mode_supported_p): Likewise.
15648 (expand_vec_perm_broadcast): Likewise.
15649 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
15650 (iq2000_function_arg_advance): Likewise.
15651 (iq2000_function_arg): Likewise.
15652 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
15653 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
15654 (m68k_libcall_value): Likewise.
15655 (m68k_function_value): Likewise.
15656 (sched_attr_op_type): Likewise.
15657 * config/mcore/mcore.c (mcore_output_move): Likewise.
15658 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
15660 (microblaze_function_arg): Likewise.
15661 * config/mips/mips.c (mips16_build_call_stub): Likewise.
15662 (mips_print_operand): Likewise.
15663 (mips_mode_ok_for_mov_fmt_p): Likewise.
15664 (mips_vector_mode_supported_p): Likewise.
15665 (mips_preferred_simd_mode): Likewise.
15666 (mips_expand_vpc_loongson_even_odd): Likewise.
15667 (mips_expand_vec_unpack): Likewise.
15668 (mips_expand_vi_broadcast): Likewise.
15669 (mips_expand_vector_init): Likewise.
15670 (mips_expand_vec_reduc): Likewise.
15671 (mips_expand_msa_cmp): Likewise.
15672 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
15673 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
15674 (cc_flags_for_mode): Likewise.
15675 * config/msp430/msp430.c (msp430_print_operand): Likewise.
15676 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
15677 (nds32_output_casesi_pc_relative): Likewise.
15678 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15679 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
15680 (nvptx_gen_unpack): Likewise.
15681 (nvptx_gen_pack): Likewise.
15682 (nvptx_gen_shuffle): Likewise.
15683 (nvptx_gen_wcast): Likewise.
15684 (nvptx_preferred_simd_mode): Likewise.
15685 * config/pa/pa.c (pa_secondary_reload): Likewise.
15686 * config/pa/predicates.md (base14_operand): Likewise.
15687 * config/powerpcspe/powerpcspe-c.c
15688 (altivec_resolve_overloaded_builtin): Likewise.
15689 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
15691 (rs6000_preferred_simd_mode): Likewise.
15692 (output_vec_const_move): Likewise.
15693 (rs6000_expand_vector_extract): Likewise.
15694 (rs6000_split_vec_extract_var): Likewise.
15695 (reg_offset_addressing_ok_p): Likewise.
15696 (rs6000_legitimate_offset_address_p): Likewise.
15697 (rs6000_legitimize_address): Likewise.
15698 (rs6000_emit_set_const): Likewise.
15699 (rs6000_const_vec): Likewise.
15700 (rs6000_emit_move): Likewise.
15701 (spe_build_register_parallel): Likewise.
15702 (rs6000_darwin64_record_arg_recurse): Likewise.
15703 (swap_selector_for_mode): Likewise.
15704 (spe_init_builtins): Likewise.
15705 (paired_init_builtins): Likewise.
15706 (altivec_init_builtins): Likewise.
15707 (do_load_for_compare): Likewise.
15708 (rs6000_generate_compare): Likewise.
15709 (rs6000_expand_float128_convert): Likewise.
15710 (emit_load_locked): Likewise.
15711 (emit_store_conditional): Likewise.
15712 (rs6000_output_function_epilogue): Likewise.
15713 (rs6000_handle_altivec_attribute): Likewise.
15714 (rs6000_function_value): Likewise.
15715 (emit_fusion_gpr_load): Likewise.
15716 (emit_fusion_p9_load): Likewise.
15717 (emit_fusion_p9_store): Likewise.
15718 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
15719 (fusion_gpr_mem_load): Likewise.
15720 (fusion_addis_mem_combo_load): Likewise.
15721 (fusion_addis_mem_combo_store): Likewise.
15722 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
15723 (fusion_gpr_mem_load): Likewise.
15724 (fusion_addis_mem_combo_load): Likewise.
15725 (fusion_addis_mem_combo_store): Likewise.
15726 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15728 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
15729 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15730 (rs6000_preferred_simd_mode): Likewise.
15731 (output_vec_const_move): Likewise.
15732 (rs6000_expand_vector_extract): Likewise.
15733 (rs6000_split_vec_extract_var): Likewise.
15734 (reg_offset_addressing_ok_p): Likewise.
15735 (rs6000_legitimate_offset_address_p): Likewise.
15736 (rs6000_legitimize_address): Likewise.
15737 (rs6000_emit_set_const): Likewise.
15738 (rs6000_const_vec): Likewise.
15739 (rs6000_emit_move): Likewise.
15740 (rs6000_darwin64_record_arg_recurse): Likewise.
15741 (swap_selector_for_mode): Likewise.
15742 (paired_init_builtins): Likewise.
15743 (altivec_init_builtins): Likewise.
15744 (rs6000_expand_float128_convert): Likewise.
15745 (emit_load_locked): Likewise.
15746 (emit_store_conditional): Likewise.
15747 (rs6000_output_function_epilogue): Likewise.
15748 (rs6000_handle_altivec_attribute): Likewise.
15749 (rs6000_function_value): Likewise.
15750 (emit_fusion_gpr_load): Likewise.
15751 (emit_fusion_p9_load): Likewise.
15752 (emit_fusion_p9_store): Likewise.
15753 * config/rx/rx.c (rx_gen_move_template): Likewise.
15754 (flags_from_mode): Likewise.
15755 * config/s390/predicates.md (s390_alc_comparison): Likewise.
15756 (s390_slb_comparison): Likewise.
15757 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
15758 (s390_vector_mode_supported_p): Likewise.
15759 (s390_cc_modes_compatible): Likewise.
15760 (s390_match_ccmode_set): Likewise.
15761 (s390_canonicalize_comparison): Likewise.
15762 (s390_emit_compare_and_swap): Likewise.
15763 (s390_branch_condition_mask): Likewise.
15764 (s390_rtx_costs): Likewise.
15765 (s390_secondary_reload): Likewise.
15766 (__SECONDARY_RELOAD_CASE): Likewise.
15767 (s390_expand_cs): Likewise.
15768 (s390_preferred_simd_mode): Likewise.
15769 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
15770 * config/sh/sh.c (sh_print_operand): Likewise.
15771 (dump_table): Likewise.
15772 (sh_secondary_reload): Likewise.
15773 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15774 * config/sh/sh.md (casesi_worker_1): Likewise.
15775 (casesi_worker_2): Likewise.
15776 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
15777 (fcc_comparison_operator): Likewise.
15778 * config/sparc/sparc.c (sparc_expand_move): Likewise.
15779 (emit_soft_tfmode_cvt): Likewise.
15780 (sparc_preferred_simd_mode): Likewise.
15781 (output_cbranch): Likewise.
15782 (sparc_print_operand): Likewise.
15783 (sparc_expand_vec_perm_bmask): Likewise.
15784 (vector_init_bshuffle): Likewise.
15785 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15786 (spu_vector_mode_supported_p): Likewise.
15787 (spu_expand_insv): Likewise.
15788 (spu_emit_branch_or_set): Likewise.
15789 (spu_handle_vector_attribute): Likewise.
15790 (spu_builtin_splats): Likewise.
15791 (spu_builtin_extract): Likewise.
15792 (spu_builtin_promote): Likewise.
15793 (spu_expand_sign_extend): Likewise.
15794 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15795 (tilegx_simd_int): Likewise.
15796 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
15797 (tilepro_simd_int): Likewise.
15798 * config/v850/v850.c (const_double_split): Likewise.
15799 (v850_print_operand): Likewise.
15800 (ep_memory_offset): Likewise.
15801 * config/vax/vax.c (vax_rtx_costs): Likewise.
15802 (vax_output_int_move): Likewise.
15803 (vax_output_int_add): Likewise.
15804 (vax_output_int_subtract): Likewise.
15805 * config/visium/predicates.md (visium_branch_operator): Likewise.
15806 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
15807 (visium_print_operand_address): Likewise.
15808 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15809 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
15810 (xtensa_expand_conditional_branch): Likewise.
15811 (xtensa_copy_incoming_a7): Likewise.
15812 (xtensa_output_literal): Likewise.
15813 * dfp.c (decimal_real_maxval): Likewise.
15814 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15816 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15817 Alan Hayward <alan.hayward@arm.com>
15818 David Sherwood <david.sherwood@arm.com>
15820 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
15821 (mode_nunits_inline): Likewise.
15822 (mode_inner_inline): Likewise.
15823 (mode_unit_size_inline): Likewise.
15824 (mode_unit_precision_inline): Likewise.
15825 (emit_insn_modes_h): Likewise. Also emit a #define of the
15827 (emit_mode_wider): Add an E_ prefix to mode names.
15828 (emit_mode_complex): Likewise.
15829 (emit_mode_inner): Likewise.
15830 (emit_mode_adjustments): Likewise.
15831 (emit_mode_int_n): Likewise.
15832 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
15833 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
15834 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
15835 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
15836 (CRC32_BUILTIN, ENTRY): Likewise.
15837 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
15838 (aarch64_pop_regs): Likewise.
15839 (aarch64_process_components): Likewise.
15840 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
15841 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
15842 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
15843 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
15844 * config/arm/arm.c (arm_init_libfuncs): Likewise.
15845 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
15847 * config/i386/i386-builtin.def (pcmpestr): Likewise.
15848 (pcmpistr): Likewise.
15849 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
15850 * config/mmix/mmix.c (mmix_output_condition): Likewise.
15851 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15853 * config/rl78/rl78.c (mduc_regs): Likewise.
15854 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15855 (htm_expand_builtin): Likewise.
15856 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
15857 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
15858 * config/xtensa/xtensa.c (print_operand): Likewise.
15859 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
15860 (NUM_MODE_VECTOR_INT): Likewise.
15861 * genoutput.c (null_operand): Likewise.
15862 (output_operand_data): Likewise.
15863 * genrecog.c (print_parameter_value): Likewise.
15864 * lra.c (debug_operand_data): Likewise.
15866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15868 * dwarf2out.c (scompare_loc_descriptor_wide)
15869 (scompare_loc_descriptor_narrow): New functions, split out from...
15870 (scompare_loc_descriptor): ...here.
15871 * expmed.c (emit_store_flag_int): New function, split out from...
15872 (emit_store_flag): ...here.
15874 2017-08-30 Richard Biener <rguenther@suse.de>
15876 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
15877 (dwarf2out_early_finish): Move setting of AT_pubnames from
15878 early debug output to early finish.
15880 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15882 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
15883 and -mdata-region to the assembler.
15885 2017-08-30 Richard Biener <rguenther@suse.de>
15887 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
15889 (gen_subprogram_die): Add DW_AT_object_pointer only early.
15890 (dwarf2out_early_global_decl): Only generate a DIE for the
15891 abstract origin if it doesn't already exist or is a declaration DIE.
15892 (resolve_addr): Do not add the linkage name twice when
15893 generating a stub DIE for the DW_TAG_GNU_call_site target.
15895 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15897 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15898 Use machine_mode rather than int for arg1_mode.
15900 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
15903 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
15904 that the second argument of the built-in functions to unpack
15905 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
15906 switch statement instead a lot of if statements.
15907 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
15908 Allow 64-bit values to be in Altivec registers as well as
15909 traditional floating point registers.
15910 (pack<mode>, FMOVE128_VSX iterator): Likewise.
15912 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
15914 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
15915 MAX_REGS_PER_ADDRESS == 1.
15917 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
15919 * config/i386/i386.opt (flag_fentry): Do not init to -1.
15920 * config/i386/i386.c (ix86_option_override_internal): Simplify
15921 setting of opts->x_flag_entry.
15923 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15924 Jakub Jelinek <jakub@redhat.com>
15925 Richard Biener <rguenther@suse.de>
15927 PR tree-optimization/81503
15928 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
15929 folded constant fits in the target type; reorder tests for clarity.
15931 2017-08-29 Martin Liska <mliska@suse.cz>
15933 * passes.def: Include pass_lower_switch.
15934 * stmt.c (dump_case_nodes): Remove and move to
15935 tree-switch-conversion.
15936 (case_values_threshold): Likewise.
15937 (expand_switch_as_decision_tree_p): Likewise.
15938 (emit_case_decision_tree): Likewise.
15939 (expand_case): Likewise.
15940 (balance_case_nodes): Likewise.
15941 (node_has_low_bound): Likewise.
15942 (node_has_high_bound): Likewise.
15943 (node_is_bounded): Likewise.
15944 (emit_case_nodes): Likewise.
15945 (struct simple_case_node): New struct.
15946 (add_case_node): Remove.
15947 (emit_case_dispatch_table): Use vector instead of case_list.
15948 (reset_out_edges_aux): Remove.
15949 (compute_cases_per_edge): Likewise.
15950 (expand_case): Build list of simple_case_node.
15951 (expand_sjlj_dispatch_table): Use it.
15952 * tree-switch-conversion.c (struct case_node): Moved from
15953 stmt.c and adjusted.
15954 (emit_case_nodes): Likewise.
15955 (node_has_low_bound): Likewise.
15956 (node_has_high_bound): Likewise.
15957 (node_is_bounded): Likewise.
15958 (case_values_threshold): Likewise.
15959 (reset_out_edges_aux): Likewise.
15960 (compute_cases_per_edge): Likewise.
15961 (add_case_node): Likewise.
15962 (dump_case_nodes): Likewise.
15963 (balance_case_nodes): Likewise.
15964 (expand_switch_as_decision_tree_p): Likewise.
15965 (emit_jump): Likewise.
15966 (emit_case_decision_tree): Likewise.
15967 (try_switch_expansion): Likewise.
15968 (do_jump_if_equal): Likewise.
15969 (emit_cmp_and_jump_insns): Likewise.
15970 (fix_phi_operands_for_edge): New function.
15971 (record_phi_operand_mapping): Likewise.
15972 (class pass_lower_switch): New pass.
15973 (pass_lower_switch::execute): New function.
15974 (make_pass_lower_switch): Likewise.
15975 (conditional_probability):
15976 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
15977 * tree-pass.h: Add make_pass_lower_switch.
15979 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15982 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
15985 2017-08-29 Richard Biener <rguenther@suse.de>
15987 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
15988 we do not add a DW_AT_inline attribute twice.
15989 (gen_subprogram_die): Remove code setting DW_AT_inline on
15990 DECL_ABSTRACT_P nodes.
15992 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
15994 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
15995 calls to internal functions.
15996 (gimplify_modify_expr): Likewise.
15997 * tree-call-cdce.c (use_internal_fn): Likewise.
15998 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
15999 (convert_to_divmod): Set the nothrow flag.
16000 * tree-if-conv.c (predicate_mem_writes): Likewise.
16001 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16002 (vectorizable_call): Likewise.
16003 (vectorizable_store): Likewise.
16004 (vectorizable_load): Likewise.
16005 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16006 (vect_recog_mask_conversion_pattern): Likewise.
16008 2017-08-29 Martin Liska <mliska@suse.cz>
16011 * gcc.c (driver_handle_option): Add new argument.
16012 * opts-common.c (handle_option): Pass
16013 target_option_override_hook.
16014 * opts-global.c (lang_handle_option): Add new option.
16015 (set_default_handlers): Add new argument.
16016 (decode_options): Likewise.
16017 * opts.c (target_handle_option): Likewise.
16018 (common_handle_option): Call target_option_override_hook.
16019 * opts.h (struct cl_option_handler_func): Add hook for
16020 target option override.
16021 (struct cl_option_handlers): Likewise.
16022 (set_default_handlers): Add new argument.
16023 (decode_options): Likewise.
16024 (common_handle_option): Likewise.
16025 (target_handle_option): Likewise.
16026 * toplev.c (toplev::main): Pass targetm.target_option.override
16029 2017-08-29 Richard Biener <rguenther@suse.de>
16030 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16032 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16033 life to the active subtree.
16035 2017-08-28 Jeff Law <law@redhat.com>
16037 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16038 derive_equivalences.
16039 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16040 Code moved into....
16041 (edge_info::derive_equivalences): New private member function
16043 * tree-ssa-dom.c (class edge_info): Changed from a struct
16044 to a class. Add ctor/dtor, methods and data members.
16045 (edge_info::edge_info): Renamed from allocate_edge_info.
16046 Initialize additional members.
16047 (edge_info::~edge_info): New.
16048 (free_dom_edge_info): Delete the edge info.
16049 (record_edge_info): Use new class & associated member functions.
16050 Tighten forms for testing for edge equivalences.
16051 (record_temporary_equivalences): Iterate over the simple
16052 equivalences rather than assuming there's only one per edge.
16053 (cprop_into_successor_phis): Iterate over the simple
16054 equivalences rather than assuming there's only one per edge.
16055 (optimize_stmt): Use operand_equal_p rather than pointer
16056 equality for mini-DSE code.
16058 2017-08-28 Nathan Sidwell <nathan@acm.org>
16060 * gcc.c (execute): Fold SIGPIPE handling into switch
16061 statement. Adjust internal error message.
16063 2017-08-28 Richard Biener <rguenther@suse.de>
16066 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
16067 Do nothing for removed DIEs.
16069 2017-08-28 Richard Biener <rguenther@suse.de>
16071 PR tree-optimization/81977
16072 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
16075 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
16078 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
16079 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
16080 using targetm.gen_mem_thread_fence.
16082 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
16085 * config/i386/i386.md (*<btsc><mode>): Change operand 2
16086 predicate to register_operand. Reorder operands.
16087 (*btr<mode>): Ditto.
16088 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
16089 (*btr<mode>_mask): Ditto.
16091 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
16093 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
16094 * config/rs6000/xmmintrin.h: New file.
16095 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
16097 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16100 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
16101 parameter and_insn and return it.
16102 (recombine_lvx_pattern): Insert a copy to ensure availability of
16103 the base register of the copied masking operation at the point of
16104 the instruction replacement.
16105 (recombine_stvx_pattern): Likewise.
16107 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16109 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
16110 undocumented switches.
16111 (-mpower9-dform-vector): Likewise.
16112 (-mpower9-dform): Likewise.
16113 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
16114 comments to delete references to -mpower9-dform* switches.
16115 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
16116 Delete reference to -mpower9-dform* switches, test for
16117 -mpower9-vector instead.
16118 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
16119 (OTHER_P9_VECTOR_MASKS): Likewise.
16120 (POWERPC_MASKS): Likewise.
16121 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
16122 tests against -mpower9-dform* to -mpower9-vector. Delete code
16123 that checked for -mpower9-dform* consistancy with other options.
16124 Add test for -mpower9-misc to enable other power9 switches.
16125 (rs6000_init_hard_regno_mode_ok): Likewise.
16126 (rs6000_option_override_internal): Likewise.
16127 (rs6000_emit_prologue): Likewise.
16128 (rs6000_emit_epilogue): Likewise.
16129 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
16130 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
16131 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
16133 (emit_fusion_p9_store): Likewise.
16134 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
16135 resetting these macros if the assembler does not support ISA 3.0
16137 (TARGET_P9_DFORM_VECTOR): Likewise.
16138 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
16139 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
16141 2017-08-25 Alan Modra <amodra@gmail.com>
16143 PR rtl-optimization/81747
16144 * cse.c (cse_extended_basic_block): Don't attempt to record
16145 equivalences for degenerate conditional branches that branch
16146 to their fall-through.
16148 2017-08-24 Martin Sebor <msebor@redhat.com>
16150 PR middle-end/81908
16151 * gimple-fold.c (size_must_be_zero_p): New function.
16152 (gimple_fold_builtin_memory_op): Call it.
16154 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
16156 * config/rs6000/mm_malloc.h: New file.
16158 2017-08-24 Bin Cheng <bin.cheng@arm.com>
16160 PR tree-optimization/81913
16161 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
16162 analysis when either IVs in condition can wrap.
16164 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
16166 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
16167 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
16169 2017-08-24 Richard Biener <rguenther@suse.de>
16172 * targhooks.c (default_target_can_inline_p): Properly
16173 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
16174 is present and always compare.
16175 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
16176 infer -mfpmath=sse from TARGET_SSE_P.
16177 (ix86_can_inline_p): Properly use target_option_default_node when
16178 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
16180 2017-08-24 Richard Biener <rguenther@suse.de>
16183 * dwarf2out.c (output_die): Handle flag_generate_offload like
16185 (output_comp_unit): Likewise.
16186 (gen_array_type_die): Likewise.
16187 (dwarf2out_early_finish): Likewise.
16188 (note_variable_value_in_expr): Likewise.
16189 (dwarf2out_finish): Likewise. Adjust assert.
16190 * cgraphunit.c (symbol_table::compile): Move setting of
16191 flag_generate_offload earlier ...
16192 (symbol_table::finalize_compilation_unit): ... here, before
16193 early debug finalization.
16195 2017-08-24 Richard Biener <rguenther@suse.de>
16197 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
16198 and ipa-fnsummary.h.
16199 (ix86_can_inline_p): When ix86_fpmath flags do not match
16200 check whether the callee uses FP math at all.
16202 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
16204 PR middle-end/81931
16205 * tree-ssanames.c (get_nonzero_bits): Use element_precision
16206 instead of TYPE_PRECISION.
16208 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
16209 Alan Hayward <alan.hayward@arm.com>
16210 David Sherwood <david.sherwood@arm.com>
16212 * combine.c (make_extraction): Use subreg_offset_from_lsb.
16214 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
16216 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
16218 (ix86_frame::stack_realign_allocate): New field.
16219 (struct machine_frame_state): Modify comments.
16220 (machine_frame_state::sp_realigned_fp_end): New field.
16221 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
16222 layout calculation.
16223 (sp_valid_at): Add assertion to assure no attempt to access invalid
16224 offset of a realigned stack.
16225 (fp_valid_at): Likewise.
16226 (choose_baseaddr): Modify comments.
16227 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
16228 ix86_expand_prologue.
16229 (ix86_expand_prologue): Modify stack realignment and allocation.
16230 (ix86_expand_epilogue): Modify comments.
16231 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
16232 avx2_runtime, avx512f, and avx512f_runtime.
16234 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
16236 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
16237 (mstackrealign): Do not init to -1.
16238 * config/i386/i386.c (ix86_option_override_internal):
16239 Check opts_set, not opts when setting default value of
16240 opts->x_ix86_force_align_arg_pointer.
16242 2017-08-23 Richard Biener <rguenther@suse.de>
16244 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
16245 lang_hooks.decl_printable_name.
16246 * print-rtl-function.c (print_rtx_function): Likewise.
16247 * tree-pretty-print.c (dump_function_header): Likewise.
16249 2017-08-23 Richard Biener <rguenther@suse.de>
16252 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
16253 -g0 at compile-time.
16255 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16257 PR middle-end/19706
16258 * doc/sourcebuild.texi (Other hardware attributes):
16261 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16263 PR middle-end/19706
16264 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
16265 Fix single-use check.
16267 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
16269 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
16271 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16273 * doc/install.texi: Modify to add more details on running selected
16276 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16278 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
16279 is combined with -mabi=ms.
16280 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
16283 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16285 PR tree-optimization/81488
16286 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
16287 and cached_basis fields.
16288 (MAX_SPREAD): New constant.
16289 (alloc_cand_and_find_basis): Initialize new fields.
16290 (clear_visited): New function.
16291 (create_phi_basis_1): Rename from create_phi_basis, set visited
16292 and cached_basis fields.
16293 (create_phi_basis): New wrapper function.
16294 (phi_add_costs_1): Rename from phi_add_costs, add spread
16295 parameter, set visited field, short-circuit when limits reached.
16296 (phi_add_costs): New wrapper function.
16297 (record_phi_increments_1): Rename from record_phi_increments, set
16299 (record_phi_increments): New wrapper function.
16300 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
16301 (phi_incr_cost): New wrapper function.
16302 (all_phi_incrs_profitable_1): Rename from
16303 all_phi_incrs_profitable, set visited field.
16304 (all_phi_incrs_profitable): New wrapper function.
16306 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16307 Alan Hayward <alan.hayward@arm.com>
16308 David Sherwood <david.sherwood@arm.com>
16310 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
16311 that takes the outer and inner modes.
16312 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
16313 comparison as the canonical test for a paradoxical subreg.
16314 * combine.c (simplify_set): Use paradoxical_subreg_p.
16315 (make_extraction): Likewise.
16316 (force_to_mode): Likewise.
16317 (rtx_equal_for_field_assignment_p): Likewise.
16318 (gen_lowpart_for_combine): Likewise.
16319 (simplify_comparison): Likewise.
16320 * cse.c (equiv_constant): Likewise.
16321 * expmed.c (store_bit_field_1): Likewise.
16322 * final.c (alter_subreg): Likewise.
16323 * fwprop.c (propagate_rtx): Likewise.
16324 (forward_propagate_subreg): Likewise.
16325 * ira-conflicts.c (ira_build_conflicts): Likewise.
16326 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
16327 * lra-constraints.c (curr_insn_transform): Likewise.
16328 (split_reg): Likewise.
16329 * lra-eliminations.c (move_plus_up): Likewise.
16330 (lra_eliminate_regs_1): Likewise.
16331 * recog.c (general_operand): Likewise.
16332 * ree.c (combine_reaching_defs): Likewise.
16333 * reload.c (push_reload): Likewise.
16334 (find_reloads): Likewise.
16335 * reload1.c (elimination_effects): Likewise.
16336 (compute_reload_subreg_offset): Likewise.
16337 (choose_reload_regs): Likewise.
16338 * rtlanal.c (subreg_lsb_1): Likewise.
16339 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16340 (simplify_subreg): Likewise.
16341 * var-tracking.c (track_loc_p): Likewise.
16342 * emit-rtl.c (byte_lowpart_offset): Likewise.
16343 (paradoxical_subreg_p): Delete out-of-line definition.
16345 2017-08-22 Jeff Law <law@redhat.com>
16347 PR tree-optimization/81741
16348 PR tree-optimization/71947
16349 * tree-ssa-dom.c: Include tree-inline.h.
16350 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
16351 equivalences if one is more expensive to compute than the other.
16352 * tree-ssa-scopedtables.h (class const_or_copies): Make
16353 record_const_or_copy_raw method private.
16354 (class avail_exprs_stack): New method simplify_binary_operation.
16355 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
16356 avail_exprs_stack::simplify_binary_operation as needed.
16357 (avail_exprs_stack::simplify_binary_operation): New function.
16359 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
16361 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
16362 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
16363 (DOT_SYMBOLS): Likewise.
16364 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
16365 (RELOCATABLE_NEEDS_FIXUP): Likewise.
16366 (RS6000_ABI_NAME): Likewise.
16367 (TARGET_CMODEL): Likewise.
16368 (TOC_SECTION_ASM_OP): Likewise.
16369 (SET_CMODEL): New macro.
16370 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
16372 2017-08-22 Richard Biener <rguenther@suse.de>
16374 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
16375 to allow for free-lang-data replacements similar to verify_type_variant.
16377 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16378 Alan Hayward <alan.hayward@arm.com>
16379 David Sherwood <david.sherwood@arm.com>
16381 * config/aarch64/aarch64.md (casesi): Use DImode rather than
16382 VOIDmode for the LABEL_REF.
16384 2017-08-22 Richard Biener <rguenther@suse.de>
16386 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
16388 2017-08-22 Richard Biener <rguenther@suse.de>
16390 * common.opt (feliminate-dwarf2-dups): Ignore.
16391 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
16392 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
16393 same_die_p_wrap, compute_section_prefix,
16394 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
16395 (comdat_symbol_id, comdat_symbol_number): Likewise.
16396 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
16398 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
16399 (output_die): Mark unreachable path unreachable.
16400 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
16401 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
16402 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
16403 (dwarf2out_early_finish): Likewise.
16405 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
16407 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
16409 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
16412 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
16413 not VAR_P. Filter attribute warnings with OPT_Wattributes.
16414 (avr_attribute_table) <io, io_low, address>: Initialize
16415 .decl_required with true.
16417 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
16419 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
16420 undocumented debugging options.
16421 (-mvsx-scalar-double): Likewise.
16422 (-mallow-df-permute): Likewise.
16423 (-mvectorize-builtins): Likewise.
16424 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16425 (rs6000_builtin_vectorized_function): Likewise.
16426 (rs6000_builtin_md_vectorized_function): Likewise.
16427 (rs6000_opt_vars): Likewise.
16429 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
16432 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
16433 (*btrq_imm): Rename from *btrq.
16434 (*btcq_imm): Rename from *btcq.
16435 (btsc): New code attribute.
16436 (*<btsc><mode>): New insn pattern.
16437 (*btr<mode>): Ditto.
16438 (*<btsc><mode>_mask): New insn_and_split pattern.
16439 (*btr<mode>_mask): Ditto.
16441 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16442 Alan Hayward <alan.hayward@arm.com>
16443 David Sherwood <david.sherwood@arm.com>
16445 * function.c (pad_below): Simplify padding calculation.
16447 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16448 Alan Hayward <alan.hayward@arm.com>
16449 David Sherwood <david.sherwood@arm.com>
16451 * target.def (function_prologue): Remove frame size argument.
16452 (function_epilogue): Likewise.
16453 * doc/tm.texi: Regenerate.
16454 * final.c (final_start_function): Update call to function_prologue.
16455 (final_end_function): Update call to function_epilogue.
16456 (default_function_pro_epilogue): Remove frame size argument.
16457 * output.h (default_function_pro_epilogue): Likewise.
16458 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
16459 (arm_output_function_prologue): Likewise.
16460 * config/frv/frv.c (frv_function_prologue): Likewise.
16461 (frv_function_epilogue): Likewise.
16462 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
16463 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
16464 (ia64_output_function_epilogue): Likewise.
16465 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
16466 (m32r_output_function_epilogue): Likewise.
16467 * config/microblaze/microblaze.c (microblaze_function_prologue)
16468 (microblaze_function_epilogue): Likewise.
16469 * config/mips/mips.c (mips_output_function_prologue): Likewise.
16470 (mips_output_function_epilogue): Likewise.
16471 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
16472 (mmix_target_asm_function_epilogue): Likewise.
16473 * config/msp430/msp430.c (msp430_start_function): Likewise.
16474 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
16475 (nds32_asm_function_epilogue): Likewise.
16476 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
16477 * config/pa/pa.c (pa_output_function_prologue): Likewise.
16478 (pa_output_function_epilogue): Likewise.
16479 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
16480 (rs6000_output_function_epilogue): Likewise.
16481 * config/rl78/rl78.c (rl78_start_function): Likewise.
16482 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
16483 (rs6000_output_function_epilogue): Likewise.
16484 * config/rx/rx.c (rx_output_function_prologue): Likewise.
16485 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
16486 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
16487 (sparc_asm_function_epilogue): Likewise.
16489 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16491 * tree.h (type_has_mode_precision_p): New function.
16492 * convert.c (convert_to_integer_1): Use it.
16493 * expr.c (expand_expr_real_2): Likewise.
16494 (expand_expr_real_1): Likewise.
16495 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
16496 * match.pd: Likewise.
16497 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
16498 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
16499 * tree-tailcall.c (process_assignment): Likewise.
16500 * tree-vect-loop.c (vectorizable_reduction): Likewise.
16501 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
16502 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
16503 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16504 (vectorizable_assignment): Likewise.
16505 (vectorizable_shift): Likewise.
16506 (vectorizable_operation): Likewise.
16507 * tree-vrp.c (register_edge_assert_for_2): Likewise.
16509 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
16511 * match.pd: Add pow (C, x) simplification.
16513 2017-08-21 Richard Biener <rguenther@suse.de>
16515 PR tree-optimization/81900
16516 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
16517 for blocks with abnormal predecessors.
16518 (compute_antic): Do not set visited flag prematurely.
16520 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
16523 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
16525 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16527 * stor-layout.h (vector_type_mode): Move to...
16528 * tree.h (vector_type_mode): ...here.
16529 * stor-layout.c (vector_type_mode): Move to...
16530 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
16532 2017-08-21 Richard Biener <rguenther@suse.de>
16534 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
16535 register_external_die hooks.
16536 (debug_false_tree_charstarstar_uhwistar): Declare.
16537 (debug_nothing_tree_charstar_uhwi): Likewise.
16538 * debug.c (do_nothing_debug_hooks): Adjust.
16539 (debug_false_tree_charstarstar_uhwistar): New do nothing.
16540 (debug_nothing_tree_charstar_uhwi): Likewise.
16541 * dbxout.c (dbx_debug_hooks): Adjust.
16542 (xcoff_debug_hooks): Likewise.
16543 * sdbout.c (sdb_debug_hooks): Likewise.
16544 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16545 * dwarf2out.c (macinfo_label_base): New global.
16546 (dwarf2out_register_external_die): New function for the
16547 register_external_die hook.
16548 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
16549 (dwarf2_debug_hooks): Use them.
16550 (dwarf2_lineno_debug_hooks): Adjust.
16551 (struct die_struct): Add with_offset flag.
16552 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
16553 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
16554 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
16555 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
16556 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
16557 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
16558 defining section names for the early LTO debug variants.
16559 (reset_indirect_string): New helper.
16560 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
16561 (print_dw_val): Add support for offsetted symbol references.
16562 (get_ultimate_context): Split out from is_cxx.
16563 (is_cxx): Use get_ultimate_context.
16564 (is_fortran): Add decl overload.
16565 (compute_comp_unit_symbol): Split out worker from
16566 compute_section_prefix.
16567 (compute_section_prefix): Call compute_comp_unit_symbol and
16568 set comdat_type_p here.
16569 (output_die): Skip DIE symbol output for the LTO added one.
16570 Handle DIE symbol references with offset.
16571 (output_comp_unit): Guard section name mangling properly.
16572 For LTO debug sections emit a symbol at the section beginning
16573 which we use to refer to its DIEs.
16574 (add_abstract_origin_attribute): For DIEs registered via
16575 dwarf2out_register_external_die directly refer to the early
16576 DIE rather than indirectly through the shadow one we created.
16577 Remove obsolete call to dwarf2out_abstract_function for
16578 non-function/block origins.
16579 (gen_array_type_die): When generating early LTO debug do
16580 not emit DW_AT_string_length.
16581 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
16582 late when in LTO. As suggested place a gcc_unreachable for
16583 the DECL_ABSTRACT_P case.
16584 (gen_subprogram_die): Avoid another specification DIE
16585 for early built declarations/definitions for the late LTO case.
16586 (gen_variable_die): Add type references for late duplicated VLA dies
16588 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
16589 we have the abstract instance already.
16590 (process_scope_var): Adjust decl DIE contexts in LTO which
16591 first puts them in limbo.
16592 (gen_decl_die): Do not generate type DIEs late apart from
16593 types for VLAs or for decls we do not yet have a DIE. Do not
16594 call dwarf2out_abstract_function late.
16595 (dwarf2out_early_global_decl): Make sure to create DIEs
16596 for abstract instances of a decl first.
16597 (dwarf2out_late_global_decl): Adjust comment.
16598 (output_macinfo_op): With multiple macro sections use
16599 macinfo_label_base to distinguish labels.
16600 (output_macinfo): Likewise. Update macinfo_label_base.
16601 Pass in the line info label.
16602 (note_variable_value_in_expr): When generating LTO resolve
16603 all variable values here by generating DIEs as needed.
16604 (init_sections_and_labels): Add early LTO debug flag parameter
16605 and generate different sections and names if set. Add generation
16606 counter for the labels so we can have multiple of them.
16607 (reset_dies): Helper to allow DIEs to be output multiple times.
16608 (dwarf2out_finish): When outputting DIEs to the fat part of an
16609 LTO object first reset DIEs.
16610 (dwarf2out_early_finish): Output early DIEs when generating LTO.
16611 (modified_type_die): Check for decl_ultimate_origin being self
16613 (gen_type_die_with_usage): Likewise.
16614 (gen_typedef_die): Allow decl_ultimate_origin being self.
16615 (set_decl_abstract_flags): Remove.
16616 (set_block_abstract_flags): Likewise.
16617 (dwarf2out_abstract_function): Treat the early generated DIEs
16618 as the abstract copy and only add DW_AT_inline and
16619 DW_AT_artificial here and call set_decl_origin_self.
16620 If the DIE has an abstract origin don't do anything.
16621 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
16622 if we have none yet (Go fails to build one, PR78628).
16623 (variably_modified_type_p): Prevent endless recursion for Ada
16624 cyclic pointer types.
16625 * lto-streamer-in.c: Include debug.h.
16626 (dref_queue): New global.
16627 (lto_read_tree_1): Stream in DIE references.
16628 (lto_input_tree): Register DIE references.
16629 (input_function): Stream DECL_DEBUG_ARGS.
16630 * lto-streamer-out.c: Include debug.h.
16631 (lto_write_tree_1): Output DIE references.
16632 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
16633 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
16634 (output_function): Stream DECL_DEBUG_ARGS.
16635 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
16636 Stream DECL_ABSTRACT_ORIGIN.
16637 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
16638 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
16639 DECL_CONTEXT for file-scope decls.
16640 * lto-streamer.h (struct dref_entry): Declare.
16641 (dref_queue): Likewise.
16642 * cfgexpand.c (pass_expand::execute): Do not call the
16643 outlining_inline_function hook here.
16644 * lto-wrapper.c (debug_obj): New global.
16645 (tool_cleanup): Unlink it if required.
16646 (debug_objcopy): New function.
16647 (run_gcc): Handle early debug sections in the IL files by
16648 extracting them to separate files, partially linkin them and
16649 feeding the result back as result to the linker.
16650 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
16651 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
16652 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
16653 sections into a separate segment.
16654 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
16656 (darwin_asm_dwarf_section): Likewise.
16657 (darwin_asm_output_dwarf_offset): Likewise.
16658 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
16660 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16661 Alan Hayward <alan.hayward@arm.com>
16662 David Sherwood <david.sherwood@arm.com>
16664 * read-md.h (md_reader::record_potential_iterator_use): Replace
16665 pointer argument with an rtx and an index.
16666 * read-rtl.c (iterator_group::apply_iterator): Likewise.
16667 (apply_mode_iterator): Likewise.
16668 (apply_code_iterator): Likewise.
16669 (apply_int_iterator): Likewise.
16670 (apply_subst_iterator): Likewise.
16671 (record_iterator_use): Likewise.
16672 (record_attribute_use): Likewise.
16673 (md_reader::record_potential_iterator_use): Likewise. Update calls
16674 to record_iterator_use and apply_iterator.
16675 (iterator_use): Replace ptr with x and index.
16676 (attribute_use): Likewise.
16677 (apply_attribute_uses): Update calls to apply_iterator.
16678 (apply_iterators): Likewise. Update initialization of iterator_use.
16679 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
16680 and record_potential_iterator_use.
16681 (rtx_reader::read_rtx_operand): Likewise.
16683 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16684 Alan Hayward <alan.hayward@arm.com>
16685 David Sherwood <david.sherwood@arm.com>
16687 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
16690 2017-08-21 Richard Biener <rguenther@suse.de>
16692 PR middle-end/81884
16693 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
16694 at struct end conservatively when comparing common bases.
16696 2017-08-21 Richard Biener <rguenther@suse.de>
16698 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
16699 (mem_ref_in_stmt): Remove.
16700 (determine_max_movement): Use ref index to get at the reference.
16701 (invariantness_dom_walker::before_dom_children): Deal with
16702 lim data already initialized.
16703 (gather_mem_refs_stmt): Initialize lim data and record ref index.
16705 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16707 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
16708 (TARGET_ISA_ROUND): Ditto.
16709 (TARGET_ROUND): Ditto.
16710 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
16711 * config/i386/i386.md: Ditto.
16712 * config/i386/sse.md: Ditto.
16713 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
16714 with OPTION_MASK_ISA_SSE4_1.
16716 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16719 * doc/extend.texi (x86 Built-in Functions): Correct the name of
16720 __builtin_ia32_lzcnt_u16.
16722 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
16725 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
16726 (rs6000_set_current_function): Rewrite function to use it.
16728 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
16731 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
16732 and TYPE_WARN_IF_NOT_ALIGN.
16733 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
16734 (handle_warn_if_not_align): New.
16735 (place_union_field): Call handle_warn_if_not_align.
16736 (place_field): Call handle_warn_if_not_align.
16737 Copy TYPE_WARN_IF_NOT_ALIGN.
16738 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
16739 (layout_type): Likewise.
16740 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
16742 (tree_decl_common): Add warn_if_not_align.
16743 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
16744 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
16745 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
16746 (DECL_WARN_IF_NOT_ALIGN): Likewise.
16747 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
16748 * doc/extend.texi: Document warn_if_not_aligned attribute.
16749 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
16751 2017-08-17 Martin Liska <mliska@suse.cz>
16754 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
16755 (get_data_dependence): Use it as pointer type.
16756 (distribute_loop): Likewise.
16758 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16760 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
16761 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
16762 (p8_vmrgow_<mode>_direct): New define_insn.
16763 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
16764 handle endianness for vmrgew and vmrgow permute patterns.
16766 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16768 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
16769 * config/rs6000/rs6000-cpus.def: Remove comment.
16770 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
16771 (POWERPC_MASKS): Likewise.
16772 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
16773 use of TARGET_VSX_TIMODE.
16774 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
16775 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
16776 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
16777 (rs6000_option_override_internal): Remove dead code.
16778 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
16779 (rs6000_legitimize_reload_address): Likewise.
16780 (rs6000_legitimate_address_p): Likewise.
16781 (rs6000_opt_masks): Delete "vsx-timode".
16782 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
16783 from function comment.
16784 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
16785 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
16786 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
16788 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
16789 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
16790 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
16791 (VSX_TI): Likewise.
16793 (define_peephole2): Likewise.
16795 2017-08-17 Martin Sebor <msebor@redhat.com>
16798 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
16799 past the end of an array.
16800 (test_pp_format): Add test cases.
16802 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
16804 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
16805 missing ECF_NOTHROW flags.
16807 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16810 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
16811 operands residing in integer registers.
16812 (*vsx_le_perm_load_<mode>): Likewise.
16813 (*vsx_le_perm_store_<mode>): Likewise.
16814 (define_peephole2): Add peepholes to optimize the above.
16816 2017-08-17 Marek Polacek <polacek@redhat.com>
16818 PR middle-end/81814
16819 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
16820 to mimic what shorten_compare did. Change the return type to bool.
16821 (fold_cond_expr_with_comparison): Update call to
16822 operand_equal_for_comparison_p.
16823 (fold_ternary_loc): Likewise.
16825 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
16827 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
16829 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
16831 2017-08-17 Richard Biener <rguenther@suse.de>
16833 * tree-ssa-structalias.c (solve_graph): When propagating
16834 to successors update the graphs succ edges and avoid duplicate work.
16836 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
16839 * config/i386/i386.c (ix86_option_override_internal): Save target
16840 specific options after ix86_stack_protector_guard_reg was changed.
16842 2017-08-17 Richard Biener <rguenther@suse.de>
16844 PR tree-optimization/81827
16845 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
16846 (new_var_info): Initialize it conservatively.
16847 (get_call_vi): Mark register vars.
16848 (new_scalar_tmp_constraint_exp): Likewise.
16849 (handle_rhs_call): Likewise.
16850 (handle_const_call): Likewise.
16851 (create_function_info_for): Likewise.
16852 (solve_constraints): Sort varinfos to separate register from
16853 non-register vars to pack points-to solution bitmaps during
16856 2017-08-17 Marek Polacek <polacek@redhat.com>
16858 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
16860 2017-08-17 Richard Biener <rguenther@suse.de>
16862 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
16863 to true when overflow is undefined and we saturated the result.
16865 2017-08-17 Alan Modra <amodra@gmail.com>
16868 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
16869 Don't use store multiple if only one reg needs saving.
16870 (interesting_frame_related_regno): New function.
16871 (rs6000_frame_related): Don't emit frame info for regs that
16873 (rs6000_emit_epilogue): Likewise.
16875 2017-08-16 Nathan Sidwell <nathan@acm.org>
16877 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
16878 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
16879 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
16880 * tree.c (free_lang_data_in_type): Use else-if chain. Always
16881 clear TYPE_LANG_1. Remove obsolete member-function stripping.
16882 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
16883 (verify_type): Adjust for TYPE_BINFO move.
16884 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
16885 process TYPE_BINFO directly.
16886 (hash_tree): Likewise.
16887 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16889 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16892 2017-08-16 David Malcolm <dmalcolm@redhat.com>
16894 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
16896 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
16899 * config/i386/i386.md (*anddi_1_btr): Change predicates of
16900 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
16901 Add ix86_binary_operator_ok to insn constraint.
16902 (*iordi_1_bts): Ditto.
16903 (*xordi_1_btc): Ditto.
16904 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
16905 Update corresponding peephole2 pattern.
16909 2017-08-16 Bin Cheng <bin.cheng@arm.com>
16911 PR tree-optimization/81832
16912 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
16913 copy loop header which has IFN_LOOP_DIST_ALIAS call.
16915 2017-08-16 Marek Polacek <polacek@redhat.com>
16918 * fold-const.c (fold_indirect_ref_1): Restore original behavior
16919 regarding size_zero_node.
16921 2017-08-16 Martin Liska <mliska@suse.cz>
16924 * config.gcc: Respect previously set extra_objs in case
16927 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
16929 PR tree-optimization/81835
16930 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
16931 the phi SSA_NAME. Check that the condition in a COND_EXPR does
16932 not depend on the phi.
16934 2017-08-16 Alan Modra <amodra@gmail.com>
16936 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
16939 2017-08-16 Alan Modra <amodra@gmail.com>
16941 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
16942 (save_reg_p): ..into this. Update all callers.
16943 (first_reg_to_save): Simplify.
16945 2017-08-16 Alan Modra <amodra@gmail.com>
16947 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
16950 2017-08-15 Joseph Myers <joseph@codesourcery.com>
16954 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
16955 constant count if that count is less than 32.
16957 2017-08-15 Nathan Sidwell <nathan@acm.org>
16959 * gcc.c (execute): Emit friendlier message if inferior is killed
16960 by an external cause.
16962 2017-08-15 Richard Biener <rguenther@suse.de>
16964 PR tree-optimization/81790
16965 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
16966 CONSTRUCTORs from simplifying and VN.
16968 2017-08-14 Martin Sebor <msebor@redhat.com>
16970 * builtin-attrs.def: Add comments.
16972 2017-08-14 Martin Sebor <msebor@redhat.com>
16975 * doc/extend.texi (attribute nonstring): Document new attribute.
16977 2017-08-14 Martin Sebor <msebor@redhat.com>
16980 * tree-diagnostic.c (default_tree_printer): Handle %G.
16981 * gimple-pretty-print.h (percent_G_format): Declare new function.
16982 * gimple-pretty-print.c (percent_G_format): Define.
16983 * tree-pretty-print.c (percent_K_format): Add argument.
16985 2017-08-14 Martin Sebor <msebor@redhat.com>
16987 PR translation/79998
16988 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
16989 Remove a stray space.
16991 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
16994 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
16995 (*iordi_1_bts): Ditto.
16996 (*xordi_1_btc): Ditto.
16998 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17001 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17002 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17004 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17005 quoted strings, and make more translator-friendly.
17006 (darwin_rs6000_override_options): Likewise.
17007 (rs6000_option_override_internal): Likewise.
17008 (rs6000_return_in_memory): Fix overlong line.
17009 (init_cmulative_args): Use quoted strings, and make more
17010 translator-friendly.
17011 (rs6000_pass_by_reference): Fix overlong line.
17012 (def_builtin): Use quoted strings.
17013 (altivec_expand_predicate_builtin): Use quoted strings, and make
17014 more translator-friendly.
17015 (htm_expand_builtin): Use quoted strings.
17016 (cpu_expand_builtin): Use quoted strings, and make more
17017 translator-friendly.
17018 (altivec_expand_builtin): Likewise.
17019 (paired_expand_predicate_builtin): Likewise.
17020 (rs6000_invalid_builtin): Likewise.
17021 (builtin_function_type): Use quoted strings.
17022 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17023 more translator-friendly.
17024 (rs6000_trampoline_init): Likewise.
17025 (rs6000_handle_altivec_attribute): Likewise.
17026 (rs6000_inner_target_options): Use quoted strings.
17027 (rs6000_disable_incompatible_switches): Likewise.
17028 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17029 strings, and make more translator-friendly.
17030 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17032 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17034 PR tree-optimization/81799
17035 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17036 cond_expr to simple gimple operand.
17038 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17040 PR middle-end/46932
17041 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17043 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17047 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17048 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17049 TARGET_GASISR_PROLOGUES.
17050 * config/avr/avr.c (avr_option_override): Same.
17051 (avr_pass_pre_proep::execute): Same.
17053 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17056 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17057 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17059 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
17061 * config/i386/i386.md (*load_tp_<mode>): Redefine as
17062 define_insn_and_split. Split to a memory load from 0 in
17063 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
17064 using PTR mode iterator.
17065 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
17066 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
17067 (*add_tp_<mode>): Redefine as define_insn_and_split.
17068 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
17069 address space. Merge with *add_tp_x32 using PTR mode iterator.
17070 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
17071 Split to an add with a memory load from 0 in
17072 DEFAULT_TLS_SEG_REG address space.
17074 2017-08-12 Andrew Pinski <apinski@cavium.com>
17076 * config/aarch64/aarch64-option-extensions.def (rdma):
17077 Fix feature string to what Linux prints out in /proc/cpuinfo.
17079 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
17082 * dwarf2out.c (modified_type_die): For C typedef types that have
17083 an ultimate origin, process the ultimate origin instead of the
17085 (gen_typedef_die): Assert that input DECLs have no ultimate
17087 (gen_type_die_with_usage): For typedef variants that have an
17088 ultimate origin, just call gen_decl_die on the original DECL.
17089 (process_scope_var): Avoid creating DIEs for local typedefs and
17090 concrete static variables.
17092 2017-08-12 Alan Modra <amodra@gmail.com>
17096 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
17097 match gnu-user.h startfile.
17098 (ENDFILE_LINUX_SPEC): Similarly.
17100 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
17103 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
17105 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
17107 2017-08-11 Tamar Christina <tamar.christina@arm.com>
17108 * config/aarch64/aarch64.md (mov<mode>): Change.
17109 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17110 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
17111 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
17113 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
17115 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
17116 for storage order barriers.
17118 2017-08-11 Martin Liska <mliska@suse.cz>
17121 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
17122 variables of void type.
17124 2017-08-11 Martin Liska <mliska@suse.cz>
17126 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
17127 TARGET_SUPPORTS_ALIASES.
17128 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
17129 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
17130 (optimize_weakref): Likewise.
17131 * symtab.c (symtab_node::noninterposable_alias): Likewise.
17132 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
17133 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
17135 2017-08-11 Martin Liska <mliska@suse.cz>
17138 * config/i386/i386.c (make_resolver_func): Do complete
17139 refactoring of the function.
17141 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17144 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
17145 * config/i386/i386.c (ix86_stack_protect_guard): Use
17146 ix86_stack_protect_guard_symbol_str to generate varible declaration.
17147 * doc/invoke.texi (x86 Options): Document
17148 -mstack-protector-guard-symbol= option.
17150 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17152 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
17153 * config/i386/i386.c (ix86_split_stack_guard): New function.
17154 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
17155 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
17156 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
17157 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
17158 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
17159 (split_stack_space_check): Call ix86_split_stack_guard.
17161 2017-08-10 Martin Sebor <msebor@redhat.com>
17163 * print-tree.c (print_node): Print location using the established
17165 Replace spaces with colons.
17166 (debug_raw, debug): Ditto.
17168 2017-08-10 Martin Sebor <msebor@redhat.com>
17171 * pretty-print.c (pp_format): Correct the handling of %s precision.
17173 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
17176 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17178 (ix86_finalize_stack_frame_flags): This. Also clear
17179 frame_pointer_needed if -fno-omit-frame-pointer is used without
17181 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17182 with ix86_finalize_stack_frame_flags.
17183 (ix86_expand_epilogue): Likewise.
17184 (ix86_expand_split_stack_prologue): Likewise.
17185 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
17187 2017-08-10 Martin Liska <mliska@suse.cz>
17190 * c-attribs.c (handle_target_attribute):
17191 Report warning for an empty string argument of target attribute.
17193 2017-08-09 Jakub Jelinek <jakub@redhat.com>
17196 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
17198 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
17199 or DECL_NONLOCAL labels.
17200 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
17201 or DECL_NONLOCAL labels here.
17203 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
17205 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
17206 to indicate when early gimple folding has been disabled.
17207 (rs6000_gimple_fold_builtin): Add debug content.
17208 (rs6000_invalid_builtin): Fix whitespace.
17209 (rs6000_expand_builtin): Fix whitespace.
17210 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
17212 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
17215 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
17216 SAVE_MULTIPLE if not all the registers that saves, should be saved.
17218 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
17220 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
17221 (qdf24xx): Likewise.
17222 * config/aarch64/aarch64.md: Include falkor.md.
17223 * config/aarch64/falkor.md: New.
17225 2017-08-09 Marek Polacek <polacek@redhat.com>
17228 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
17229 * diagnostic.c (emit_diagnostic): Add a comment.
17230 (emit_diagnostic_valist): New function.
17232 2017-08-09 Marek Polacek <polacek@redhat.com>
17235 * input.c (make_location): New overload.
17236 * input.h (make_location): Declare.
17238 2017-08-08 Alan Modra <amodra@gmail.com>
17239 H.J. Lu <hongjiu.lu@intel.com>
17242 * gcc.c (NO_PIE_SPEC): Delete.
17243 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
17245 (LINK_PIE_SPEC): ..to here.
17246 (LINK_COMMAND_SPEC): Support -no-pie.
17247 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
17248 chain of crtbegin*.o selection, update for PIE_SPEC changes and
17250 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
17251 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
17252 (ENDFILE_CRTEND_SPEC): Similarly.
17254 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
17257 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
17258 (mstack-protector-guard-offset=): Ditto.
17259 * config/i386/i386.c (ix86_option_override): Handle
17260 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
17262 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
17263 ix86_stack_protect_guard_offset variables.
17264 (TARGET_STACK_PROTECT_GUARD): Always define.
17265 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
17266 and -mstack-protector-guard-offset= options.
17268 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17270 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
17271 boundary case for the last candidate.
17273 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17275 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
17276 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
17278 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17280 PR middle-end/19706
17281 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
17282 * config/aarch64/aarch64-builtins.c
17283 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
17284 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
17285 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
17287 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17288 Andrew Pinski <pinskia@gmail.com>
17290 PR middle-end/19706
17291 * internal-fn.def (XORSIGN): New.
17292 * optabs.def (xorsign_optab): New.
17293 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
17294 (convert_expand_mult_copysign): New.
17295 (pass_optimize_widening_mul::execute): Call
17296 convert_expand_mult_copysign.
17298 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17300 PR tree-optimization/81354
17301 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
17302 Insert on edges rather than explicitly creating landing pads.
17303 (analyze_candidates_and_replace): Commit edge inserts.
17305 2017-08-08 Richard Biener <rguenther@suse.de>
17307 PR middle-end/81719
17308 * tree-ssa-loop-niter.c: Include tree-dfa.h.
17309 (expand_simple_operations): Also look through ADDR_EXPRs with
17310 MEM_REF bases treating them as POINTER_PLUS_EXPR.
17312 2017-08-08 Richard Biener <rguenther@suse.de>
17314 PR tree-optimization/81723
17315 * tree-vect-slp.c (struct bst_traits): New hash traits.
17316 (bst_fail): New global.
17317 (vect_build_slp_tree_2): New worker, split out from ...
17318 (vect_build_slp_tree): ... this now wrapping it with using
17319 bst_fail set to cache SLP tree build fails. Properly handle
17321 (vect_analyze_slp_instance): Allocate and free bst_fail.
17323 2017-08-08 Martin Liska <mliska@suse.cz>
17326 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
17327 LABEL_DECLs that can be from a different function.
17329 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17331 PR tree-optimization/81744
17332 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
17333 loop's number of iterations.
17335 2017-08-08 Martin Liska <mliska@suse.cz>
17337 * asan.c: Include header files.
17338 * attribs.c (build_decl_attribute_variant): New function moved
17340 (build_type_attribute_qual_variant): Likewise.
17341 (cmp_attrib_identifiers): Likewise.
17342 (simple_cst_list_equal): Likewise.
17343 (omp_declare_simd_clauses_equal): Likewise.
17344 (attribute_value_equal): Likewise.
17345 (comp_type_attributes): Likewise.
17346 (build_type_attribute_variant): Likewise.
17347 (lookup_ident_attribute): Likewise.
17348 (remove_attribute): Likewise.
17349 (merge_attributes): Likewise.
17350 (merge_type_attributes): Likewise.
17351 (merge_decl_attributes): Likewise.
17352 (merge_dllimport_decl_attributes): Likewise.
17353 (handle_dll_attribute): Likewise.
17354 (attribute_list_equal): Likewise.
17355 (attribute_list_contained): Likewise.
17356 * attribs.h (lookup_attribute): New function moved from tree.[ch].
17357 (lookup_attribute_by_prefix): Likewise.
17358 * bb-reorder.c: Include header files.
17359 * builtins.c: Likewise.
17360 * calls.c: Likewise.
17361 * cfgexpand.c: Likewise.
17362 * cgraph.c: Likewise.
17363 * cgraphunit.c: Likewise.
17364 * convert.c: Likewise.
17365 * dwarf2out.c: Likewise.
17366 * final.c: Likewise.
17367 * fold-const.c: Likewise.
17368 * function.c: Likewise.
17369 * gimple-expr.c: Likewise.
17370 * gimple-fold.c: Likewise.
17371 * gimple-pretty-print.c: Likewise.
17372 * gimple.c: Likewise.
17373 * gimplify.c: Likewise.
17374 * hsa-common.c: Likewise.
17375 * hsa-gen.c: Likewise.
17376 * internal-fn.c: Likewise.
17377 * ipa-chkp.c: Likewise.
17378 * ipa-cp.c: Likewise.
17379 * ipa-devirt.c: Likewise.
17380 * ipa-fnsummary.c: Likewise.
17381 * ipa-inline.c: Likewise.
17382 * ipa-visibility.c: Likewise.
17384 * lto-cgraph.c: Likewise.
17385 * omp-expand.c: Likewise.
17386 * omp-general.c: Likewise.
17387 * omp-low.c: Likewise.
17388 * omp-offload.c: Likewise.
17389 * omp-simd-clone.c: Likewise.
17390 * opts-global.c: Likewise.
17391 * passes.c: Likewise.
17392 * predict.c: Likewise.
17393 * sancov.c: Likewise.
17394 * sanopt.c: Likewise.
17395 * symtab.c: Likewise.
17396 * toplev.c: Likewise.
17397 * trans-mem.c: Likewise.
17398 * tree-chkp.c: Likewise.
17399 * tree-eh.c: Likewise.
17400 * tree-into-ssa.c: Likewise.
17401 * tree-object-size.c: Likewise.
17402 * tree-parloops.c: Likewise.
17403 * tree-profile.c: Likewise.
17404 * tree-ssa-ccp.c: Likewise.
17405 * tree-ssa-live.c: Likewise.
17406 * tree-ssa-loop.c: Likewise.
17407 * tree-ssa-sccvn.c: Likewise.
17408 * tree-ssa-structalias.c: Likewise.
17409 * tree-ssa.c: Likewise.
17410 * tree-streamer-in.c: Likewise.
17411 * tree-vectorizer.c: Likewise.
17412 * tree-vrp.c: Likewise.
17413 * tsan.c: Likewise.
17414 * ubsan.c: Likewise.
17415 * varasm.c: Likewise.
17416 * varpool.c: Likewise.
17417 * tree.c: Remove functions moved to attribs.[ch].
17418 * tree.h: Likewise.
17419 * config/aarch64/aarch64.c: Add attrs.h header file.
17420 * config/alpha/alpha.c: Likewise.
17421 * config/arc/arc.c: Likewise.
17422 * config/arm/arm.c: Likewise.
17423 * config/avr/avr.c: Likewise.
17424 * config/bfin/bfin.c: Likewise.
17425 * config/c6x/c6x.c: Likewise.
17426 * config/cr16/cr16.c: Likewise.
17427 * config/cris/cris.c: Likewise.
17428 * config/darwin.c: Likewise.
17429 * config/epiphany/epiphany.c: Likewise.
17430 * config/fr30/fr30.c: Likewise.
17431 * config/frv/frv.c: Likewise.
17432 * config/ft32/ft32.c: Likewise.
17433 * config/h8300/h8300.c: Likewise.
17434 * config/i386/winnt.c: Likewise.
17435 * config/ia64/ia64.c: Likewise.
17436 * config/iq2000/iq2000.c: Likewise.
17437 * config/lm32/lm32.c: Likewise.
17438 * config/m32c/m32c.c: Likewise.
17439 * config/m32r/m32r.c: Likewise.
17440 * config/m68k/m68k.c: Likewise.
17441 * config/mcore/mcore.c: Likewise.
17442 * config/microblaze/microblaze.c: Likewise.
17443 * config/mips/mips.c: Likewise.
17444 * config/mmix/mmix.c: Likewise.
17445 * config/mn10300/mn10300.c: Likewise.
17446 * config/moxie/moxie.c: Likewise.
17447 * config/msp430/msp430.c: Likewise.
17448 * config/nds32/nds32-isr.c: Likewise.
17449 * config/nds32/nds32.c: Likewise.
17450 * config/nios2/nios2.c: Likewise.
17451 * config/nvptx/nvptx.c: Likewise.
17452 * config/pa/pa.c: Likewise.
17453 * config/pdp11/pdp11.c: Likewise.
17454 * config/powerpcspe/powerpcspe.c: Likewise.
17455 * config/riscv/riscv.c: Likewise.
17456 * config/rl78/rl78.c: Likewise.
17457 * config/rx/rx.c: Likewise.
17458 * config/s390/s390.c: Likewise.
17459 * config/sh/sh.c: Likewise.
17460 * config/sol2.c: Likewise.
17461 * config/sparc/sparc.c: Likewise.
17462 * config/spu/spu.c: Likewise.
17463 * config/stormy16/stormy16.c: Likewise.
17464 * config/tilegx/tilegx.c: Likewise.
17465 * config/tilepro/tilepro.c: Likewise.
17466 * config/v850/v850.c: Likewise.
17467 * config/vax/vax.c: Likewise.
17468 * config/visium/visium.c: Likewise.
17469 * config/xtensa/xtensa.c: Likewise.
17471 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17474 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
17475 constraints since the -mupper-regs-* switches have been
17477 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
17478 into a vector from a double word element that was extracted from
17479 another vector, and eliminate extra XXPERMDI instructions.
17480 (vsx_concat_<mode>_2): Likewise.
17481 (vsx_concat_<mode>_3): Likewise.
17482 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
17483 concat to allow optimizing inserts from previous extracts.
17485 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
17487 * config/i386/i386.c (ix86_stack_protect_guard): Generate
17488 memory reference to a SSP offset in TLS address space.
17489 (ix86_print_operand) <case '@'>: Remove.
17490 (ix86_print_operand_punct_valid_p): Remove '@' code.
17491 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
17492 UNSPEC_SP_TLS_TEST.
17493 (stack_tls_protect_set_<mode>): Remove.
17494 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
17495 (stack_tls_protect_test_<mode>): Remove.
17496 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
17498 2017-08-07 Olivier Hainque <hainque@adacore.com>
17501 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
17503 2017-08-07 Douglas Rupp <rupp@adacore.com>
17505 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
17506 variable was referenced as multidir in command.
17508 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17511 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
17514 2017-08-07 Martin Liska <mliska@suse.cz>
17516 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
17517 * config/rl78/rl78.c: Add include of attribs.h.
17518 * config/sh/sh.c: Likewise.
17519 * config/v850/v850.c: Likewise.
17521 2017-08-07 Tom de Vries <tom@codesourcery.com>
17523 PR middle-end/78266
17524 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
17526 2017-08-07 Martin Liska <mliska@suse.cz>
17528 * config/mips/mips.c: Include attribs.h.
17530 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
17533 * doc/invoke.texi: Document change in behvaior for -Ofast for
17536 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
17538 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
17540 (aarch64_pop_regs): Likewise.
17541 (aarch64_gen_load_pair): Likewise.
17542 (aarch64_save_callee_saves): Likewise.
17543 (aarch64_restore_callee_saves): Likewise.
17545 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17547 * config/i386/i386.c: Revert the last change.
17549 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17552 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17554 (ix86_finalize_stack_frame_flags): This. Also clear
17555 frame_pointer_needed if -fno-omit-frame-pointer is used without
17557 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17558 with ix86_finalize_stack_frame_flags.
17559 (ix86_expand_epilogue): Likewise.
17560 (ix86_expand_split_stack_prologue): Likewise.
17562 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17565 * config/i386/i386.c (get_builtin_code_for_version): Set priority
17566 to P_AES for Westmere.
17568 2017-08-07 Jonathan Yong <10walls@gmail.com>
17570 * config/i386/mingw.opt (fset-stack-executable): Removed.
17571 * config/i386/cygming.opt (fset-stack-executable): Moved
17573 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
17575 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
17577 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
17579 2017-08-07 Marek Polacek <polacek@redhat.com>
17581 PR middle-end/81737
17582 * fold-const.c (fold_indirect_ref_1): Check type_domain.
17584 2017-08-07 Martin Liska <mliska@suse.cz>
17586 * attribs.h (canonicalize_attr_name): New function.
17587 (cmp_attribs): Move from c-format.c and adjusted.
17588 (is_attribute_p): Moved from tree.h.
17589 * tree-inline.c: Add new includes.
17590 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
17591 (private_is_attribute_p): Remove.
17592 (private_lookup_attribute): Likewise.
17593 (private_lookup_attribute_by_prefix): Simplify.
17594 (remove_attribute): Use is_attribute_p.
17595 * tree.h: Remove removed declarations.
17597 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17599 PR middle-end/81698
17600 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
17601 instead of computing it in the function. Formatting fix.
17602 (expand_case): Don't rely on default_edge being the first edge,
17603 clear it if removing it, pass default_edge to
17604 emit_case_dispatch_table.
17605 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
17608 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
17610 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
17611 insn in the function, emit NOP after the insn.
17613 2017-08-06 Tom de Vries <tom@codesourcery.com>
17615 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
17618 2017-08-06 Tom de Vries <tom@codesourcery.com>
17620 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
17623 2017-08-04 Yury Gribov <tetra2005@gmail.com>
17625 PR tree-optimization/57371
17626 * match.pd: New pattern.
17628 2017-08-04 Marek Polacek <polacek@redhat.com>
17630 PR middle-end/81695
17631 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
17632 perform the computation in offset_int.
17634 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17636 PR tree-optimization/81136
17637 * tree-vectorizer.h: Include tree-hash-traits.h.
17638 (vec_base_alignments): New typedef.
17639 (vec_info): Add a base_alignments field.
17640 (vect_record_base_alignments): Declare.
17641 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
17643 (DR_IS_CONDITIONAL_IN_STMT): New macro.
17644 (create_data_ref): Add an is_conditional_in_stmt argument.
17645 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
17646 the is_conditional_in_stmt field.
17647 (data_ref_loc): Add an is_conditional_in_stmt field.
17648 (get_references_in_stmt): Set the is_conditional_in_stmt field.
17649 (find_data_references_in_stmt): Update call to create_data_ref.
17650 (graphite_find_data_references_in_stmt): Likewise.
17651 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
17652 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
17653 (vect_record_base_alignment): New function.
17654 (vect_record_base_alignments): Likewise.
17655 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
17656 for nested statements even if we fail to compute a misalignment.
17657 Use pooled base alignments for unconditional references.
17658 (vect_find_same_alignment_drs): Compare base addresses instead
17660 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
17661 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
17663 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17665 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
17666 Add an explicit name for the enum. Use auto_vec for slp_instances
17667 and grouped_stores.
17668 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
17670 (_bb_vec_info): Add a constructor and destructor.
17671 (vinfo_for_stmt): Return NULL for uids of -1 as well.
17672 (destroy_loop_vec_info): Delete.
17673 (vect_destroy_datarefs): Likewise.
17674 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
17675 (vec_info::vec_info): New function.
17676 (vec_info::~vec_info): Likewise.
17677 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
17678 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
17679 destroy_loop_vec_info.
17680 * tree-vect-loop.c (new_loop_vec_info): Replace with...
17681 (_loop_vec_info::_loop_vec_info): ...this.
17682 (destroy_loop_vec_info): Replace with...
17683 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
17684 the stmt_vec_infos. Leave handling of vec_info information to its
17685 destructor. Remove explicit vector releases.
17686 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
17687 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
17688 * tree-vect-slp.c (new_bb_vec_info): Replace with...
17689 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
17690 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
17691 (destroy_bb_vec_info): Replace with...
17692 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
17693 information to its destructor.
17694 (vect_slp_analyze_bb_1): Use new and delete instead of
17695 new_bb_vec_info and destroy_bb_vec_info.
17696 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
17699 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17701 * tree-data-ref.h (subscript): Add access_fn field.
17702 (data_dependence_relation): Add could_be_independent_p.
17703 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
17704 (same_access_functions): Move to tree-data-ref.c.
17705 * tree-data-ref.c (ref_contains_union_access_p): New function.
17706 (access_fn_component_p): Likewise.
17707 (access_fn_components_comparable_p): Likewise.
17708 (dr_analyze_indices): Add a reference to access_fn_component_p.
17709 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
17711 (constant_access_functions): Likewise.
17712 (add_other_self_distances): Likewise.
17713 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
17714 (initialize_data_dependence_relation): Use XCNEW and remove
17715 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
17716 of access functions that have the same type. Allow the
17717 subsequence to end with different bases in some circumstances.
17718 Record the chosen access functions in SUB_ACCESS_FN.
17719 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
17720 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
17721 (subscript_dependence_tester_1): Likewise dra and drb.
17722 (build_classic_dist_vector): Update calls accordingly.
17723 (subscript_dependence_tester): Likewise.
17724 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
17725 DDR_COULD_BE_INDEPENDENT_P.
17726 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
17727 comp_alias_ddrs instead of may_alias_ddrs.
17728 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
17730 (vect_analyze_data_ref_dependence): Use it if
17731 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
17732 distance vectors if that fails.
17733 (dependence_distance_ge_vf): New function.
17734 (vect_prune_runtime_alias_test_list): Use it. Don't clear
17735 LOOP_VINFO_MAY_ALIAS_DDRS.
17737 2017-08-04 Richard Biener <rguenther@suse.de>
17739 PR middle-end/81705
17740 * fold-const.c (fold_binary_loc): Properly restrict
17741 minus_var0 && minus_var1 case when associating undefined overflow
17744 2017-08-04 Tom de Vries <tom@codesourcery.com>
17746 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
17748 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17750 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17751 Don't start diagnostic messages with a capital letter.
17752 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17754 (rs6000_invalid_builtin): Likewise.
17755 (rs6000_trampoline_init): Likewise.
17757 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17760 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
17761 after setting changeable df flags.
17763 2017-08-03 Richard Biener <rguenther@suse.de>
17765 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
17766 up if the use is in USE - X.
17768 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
17770 * toplev.c (dumpfile.h): New include.
17771 (internal_error_reentered): New static function. Use it...
17772 (internal_error_function): ...here to handle reentered internal_error.
17774 2017-08-03 Richard Biener <rguenther@suse.de>
17776 PR middle-end/81148
17777 * fold-const.c (split_tree): Add minus_var and minus_con
17778 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
17779 here but always use minus_*.
17780 (associate_trees): Assert we never associate with MINUS_EXPR
17781 and NULL first operand. Do not recurse for PLUS_EXPR operands
17782 when associating as MINUS_EXPR either.
17783 (fold_binary_loc): Track minus_var and minus_con.
17785 2017-08-03 Tom de Vries <tom@codesourcery.com>
17788 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
17789 ACCEL_COMPILER, apply finish_options on
17790 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
17792 2017-08-03 Tom de Vries <tom@codesourcery.com>
17795 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
17796 function_entry_patch_area_size > 0.
17798 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17801 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
17802 instead of 10??LU, perform unit multiplication in wide_int,
17803 don't change alloc_object_size_limit if the limit is larger
17806 PR tree-optimization/81655
17807 PR tree-optimization/81588
17808 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
17809 the case when ranges[i].low and high are 1 for unsigned type with
17812 PR middle-end/81052
17813 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
17814 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
17816 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17818 * tree-vrp.h: Add include guard.
17820 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
17823 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
17824 (ud2): New insn pattern.
17825 * config/i386/i386.c (ix86_expand_epilogue):
17826 For naked functions, generate ud2 instead of trap insn.
17828 2017-08-02 Marek Polacek <polacek@redhat.com>
17831 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
17833 2017-08-02 Tom de Vries <tom@codesourcery.com>
17834 Cesar Philippidis <cesar@codesourcery.com>
17836 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
17837 Add missing edge probabilities.
17839 2017-08-02 Tamar Christina <tamar.christina@arm.com>
17841 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
17842 Correct endianness.
17844 2017-08-02 Jakub Jelinek <jakub@redhat.com>
17846 PR middle-end/79499
17847 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
17848 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
17849 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
17850 prologue_seq sequences - if any.
17852 2017-08-02 Richard Biener <rguenther@suse.de>
17854 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
17855 via vectors if supported, integer extracts via punning if supported
17856 or otherwise vector extracts.
17858 2017-08-02 Richard Biener <rguenther@suse.de>
17860 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
17862 (bitmap_insert_into_set): ... this.
17864 2017-08-02 Richard Biener <rguenther@suse.de>
17866 PR tree-optimization/81633
17868 2015-08-17 Alan Hayward <alan.hayward@arm.com>
17870 PR tree-optimization/71752
17871 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
17873 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
17875 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
17876 (machine_function::call_ms2sysv_pad_out): Remove field.
17877 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
17878 (ix86_compute_frame_layout): Likewise.
17880 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
17883 * config/i386/i386.c (ix86_set_func_type): Disallow naked
17884 attribute with interrupt attribute.
17886 2017-08-01 Andrew Pinski <apinski@cavium.com>
17888 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
17889 BIT_INSERT_EXPR's operand 1
17890 to see if the types precision matches.
17892 2017-08-01 Martin Liska <mliska@suse.cz>
17894 PR middle-end/70140
17895 * builtins.c (expand_builtin_memcpy_args): Remove.
17896 (expand_builtin_memcpy): Call newly added function
17897 expand_builtin_memory_copy_args.
17898 (expand_builtin_memcpy_with_bounds): Likewise.
17899 (expand_builtin_mempcpy): Remove last argument.
17900 (expand_builtin_mempcpy_with_bounds): Likewise.
17901 (expand_builtin_memory_copy_args): New function created from
17902 expand_builtin_mempcpy_args with small modifications.
17903 (expand_builtin_mempcpy_args): Remove.
17904 (expand_builtin_stpcpy): Remove unused argument.
17905 (expand_builtin): Likewise.
17906 (expand_builtin_with_bounds): Likewise.
17908 2017-08-01 Martin Liska <mliska@suse.cz>
17911 Make mempcpy more optimal (PR middle-end/70140).
17913 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17916 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
17917 __builtin_vec_cmpne verify both arguments are compatible vectors
17918 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
17919 verify arg1_type is a pointer or array type. For __builtin_vec_st,
17920 move computation of aligned to after checking the argument types.
17924 * config/rs6000/vsx.md (vextract_fp_from_shorth,
17925 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
17928 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
17929 Doug Rupp <rupp@adacore.com>
17930 Olivier Hainque <hainque@adacore.com>
17932 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
17933 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
17935 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
17936 for TARGET_OS_CPP_BUILTIN.
17937 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
17938 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
17940 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
17941 passing required abi options to the assembler for EABI configurations.
17942 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
17943 of .text.hot and .text.unlikely sections for kernel modules when
17944 using ARM style exceptions.
17945 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
17946 options. Add EXTRA_CC1_SPEC.
17947 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
17949 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
17951 (ARM_TARGET2_DWARF_FORMAT): Define.
17952 * config/arm/t-vxworks: Adjust multilib control to removal of the
17953 Diab command line options.
17955 2017-08-01 Martin Liska <mliska@suse.cz>
17957 PR gcov-profile/81561
17958 * gcov.c (unblock): Make unblocking safe as we need to preserve
17959 index correspondence of blocks and block_lists.
17961 2017-08-01 Richard Biener <rguenther@suse.de>
17963 PR tree-optimization/81181
17964 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
17965 (compute_antic): ... end of iteration here.
17967 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
17969 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
17970 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
17971 (ftree-slp-vectorize): Likewise.
17972 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
17973 can no longer be set independent of flag_tree_loop_vectorize.
17974 * omp-general.c (emp_max_vf): Likewise.
17975 * opts.c (enable_fdo_optimizations): Remove references to
17976 flag_tree_vectorize, these are now implicit.
17977 (common_handle_option): Remove handling for OPT_ftree_vectorize,
17978 and leave it for the options machinery.
17980 2017-08-01 Martin Liska <mliska@suse.cz>
17982 PR middle-end/70140
17983 * builtins.c (expand_builtin_memcpy_args): Remove.
17984 (expand_builtin_memcpy): Call newly added function
17985 expand_builtin_memory_copy_args.
17986 (expand_builtin_memcpy_with_bounds): Likewise.
17987 (expand_builtin_mempcpy): Remove last argument.
17988 (expand_builtin_mempcpy_with_bounds): Likewise.
17989 (expand_builtin_memory_copy_args): New function created from
17990 expand_builtin_mempcpy_args with small modifications.
17991 (expand_builtin_mempcpy_args): Remove.
17992 (expand_builtin_stpcpy): Remove unused argument.
17993 (expand_builtin): Likewise.
17994 (expand_builtin_with_bounds): Likewise.
17996 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
17999 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18000 print "ds:" only for immediates in generic address space.
18002 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18005 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18006 (ix86_function_ok_for_sibcall): Return false for naked functions.
18008 2017-08-01 Richard Biener <rguenther@suse.de>
18010 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18011 (compute_antic): Seed worklist with exit block predecessors.
18012 * cfganal.c (dfs_find_deadend): For a cycle return the source
18013 of the edge closing it.
18015 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18017 * config/aarch64/aarch64.c
18018 (aarch64_can_const_movi_rtx_p): Move 0 check.
18020 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18022 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18023 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18025 * match.pd: Ditto in address comparison pattern.
18027 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18029 PR tree-optimization/81627
18030 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18031 closed ssa form for store-store chain.
18033 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18035 PR tree-optimization/81620
18036 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18037 for store-store chain.
18039 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18041 PR tree-optimization/81588
18042 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18043 ranges[i].in_p, invert comparison code ccode. For >/>=,
18044 swap rhs1 and rhs2 and comparison code unconditionally,
18045 for </<= don't do that. Don't swap rhs1/rhs2 again if
18046 ranges[i].in_p, instead invert comparison code ccode if
18047 opcode or oe->rank is BIT_IOR_EXPR.
18050 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18051 a direct optab to conversion optab.
18052 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18053 with GET_MODE_INNER as last argument instead of optab_handler.
18054 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18055 vector extraction if possible and optab is available.
18056 * expr.c (store_constructor): Use convert_optab_handler instead
18057 of optab_handler. Use vector initialization from smaller
18058 vectors if possible and optab is available.
18059 * tree-vect-stmts.c (vectorizable_load): Likewise.
18060 * doc/md.texi (vec_extract, vec_init): Document that the optabs
18061 now have two modes.
18062 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
18063 of vec_init from half-sized vectors with the same element mode.
18064 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
18065 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
18066 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
18067 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
18068 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
18069 after mode in gen_vec_extract* calls.
18070 (vec_extract<mode>): Renamed to ...
18071 (vec_extract<mode><ssescalarmodelower>): ... this.
18072 (vec_extract<mode><ssehalfvecmodelower>): New expander.
18073 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
18074 element mode after mode in gen_vec_init* calls.
18075 (VEC_INIT_HALF_MODE): New mode iterator.
18076 (vec_init<mode>): Renamed to ...
18077 (vec_init<mode><ssescalarmodelower>): ... this.
18078 (vec_init<mode><ssehalfvecmodelower>): New expander.
18079 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
18080 (vec_extractv2sfsf): ... this.
18081 (vec_initv2sf): Renamed to ...
18082 (vec_initv2sfsf): ... this.
18083 (vec_extractv2si): Renamed to ...
18084 (vec_extractv2sisi): ... this.
18085 (vec_initv2si): Renamed to ...
18086 (vec_initv2sisi): ... this.
18087 (vec_extractv4hi): Renamed to ...
18088 (vec_extractv4hihi): ... this.
18089 (vec_initv4hi): Renamed to ...
18090 (vec_initv4hihi): ... this.
18091 (vec_extractv8qi): Renamed to ...
18092 (vec_extractv8qiqi): ... this.
18093 (vec_initv8qi): Renamed to ...
18094 (vec_initv8qiqi): ... this.
18095 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
18096 (vec_init<mode>): Renamed to ...
18097 (vec_init<mode><VEC_base_l>): ... this.
18098 (vec_extract<mode>): Renamed to ...
18099 (vec_extract<mode><VEC_base_l>): ... this.
18100 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
18101 (vec_initv2sfsf): ... this.
18102 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
18103 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18104 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
18105 element mode after mode in gen_vec_init* calls.
18106 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
18107 (vec_init<mode><Vel>): ... this.
18108 (vec_extract<mode>): Renamed to ...
18109 (vec_extract<mode><Vel>): ... this.
18110 * config/aarch64/iterators.md (Vel): New mode attribute.
18111 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
18112 Add element mode after mode in gen_vec_extract* calls.
18113 * config/s390/vector.md (non_vec_l): New mode attribute.
18114 (vec_extract<mode>): Renamed to ...
18115 (vec_extract<mode><non_vec_l>): ... this.
18116 (vec_init<mode>): Renamed to ...
18117 (vec_init<mode><non_vec_l>): ... this.
18118 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
18119 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
18121 * config/arm/iterators.md (V_elem_l): New mode attribute.
18122 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
18123 (vec_extract<mode><V_elem_l>): ... this.
18124 (vec_extractv2di): Renamed to ...
18125 (vec_extractv2didi): ... this.
18126 (vec_init<mode>): Renamed to ...
18127 (vec_init<mode><V_elem_l>): ... this.
18128 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
18129 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
18130 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
18131 Add element mode after gen_vec_extract* calls.
18132 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
18133 (vec_init<mode><unitmode>): ... this.
18134 (vec_extract<mode>): Renamed to ...
18135 (vec_extract<mode><unitmode>): ... this.
18136 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
18137 (vec_init<mode><unitmode>): ... this.
18138 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
18139 (vec_initv2sfsf): ... this.
18140 (vec_extractv2sf): Renamed to ...
18141 (vec_extractv2sfsf): ... this.
18142 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
18143 Add element mode after gen_vec_extract* calls.
18144 * config/mips/mips.md (unitmode): New mode iterator.
18145 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
18146 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
18147 * config/spu/spu.md (inner_l): New mode attribute.
18148 (vec_init<mode>): Renamed to ...
18149 (vec_init<mode><inner_l>): ... this.
18150 (vec_extract<mode>): Renamed to ...
18151 (vec_extract<mode><inner_l>): ... this.
18152 * config/sparc/sparc.md (veltmode): New mode iterator.
18153 (vec_init<VMALL:mode>): Renamed to ...
18154 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
18155 * config/ia64/vect.md (vec_initv2si): Renamed to ...
18156 (vec_initv2sisi): ... this.
18157 (vec_initv2sf): Renamed to ...
18158 (vec_initv2sfsf): ... this.
18159 (vec_extractv2sf): Renamed to ...
18160 (vec_extractv2sfsf): ... this.
18161 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
18162 (vec_init<mode>): Renamed to ...
18163 (vec_init<mode><VEC_base_l>): ... this.
18164 (vec_extract<mode>): Renamed to ...
18165 (vec_extract<mode><VEC_base_l>): ... this.
18166 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
18167 (vec_initv2sfsf): ... this.
18168 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
18169 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18170 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
18171 gen_vec_init* calls.
18173 2017-08-01 Richard Biener <rguenther@suse.de>
18175 PR tree-optimization/81297
18176 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
18177 TREE_OVERFLOW from INTEGER_CSTs.
18179 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
18181 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
18183 2017-07-31 Carl Love <cel@us.ibm.com>
18185 * config/rs6000/rs6000-c: Add support for built-in functions
18186 vector signed char vec_xl_be (signed long long, signed char *);
18187 vector unsigned char vec_xl_be (signed long long, unsigned char *);
18188 vector signed int vec_xl_be (signed long long, signed int *);
18189 vector unsigned int vec_xl_be (signed long long, unsigned int *);
18190 vector signed long long vec_xl_be (signed long long, signed long long *);
18191 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
18192 vector signed short vec_xl_be (signed long long, signed short *);
18193 vector unsigned short vec_xl_be (signed long long, unsigned short *);
18194 vector double vec_xl_be (signed long long, double *);
18195 vector float vec_xl_be (signed long long, float *);
18196 * config/rs6000/altivec.h (vec_xl_be): Add #define.
18197 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
18198 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
18200 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
18201 (altivec_expand_builtin): Add switch statement to call
18202 altivec_expand_xl_be for each builtin.
18203 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
18204 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
18205 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
18206 __builtin_vsx_le_be_v16qi.
18207 * doc/extend.texi: Update the built-in documentation file for the
18208 new built-in functions.
18210 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18213 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
18215 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18217 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18219 * config.gcc: Add z14.
18220 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
18221 CPU model numbers for z13s and z14.
18222 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
18224 * config/s390/s390-opts.h (enum processor_type): Rename
18225 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18226 * config/s390/s390.c (processor_table): Add field for CPU name to
18227 be passed to Binutils.
18228 (s390_asm_output_machine_for_arch): Use the new field in
18229 processor_table for Binutils.
18230 (s390_expand_builtin): Replace arch12 with z14.
18231 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18232 (s390_get_sched_attrmask): Likewise.
18233 (s390_get_unit_mask): Likewise.
18234 * config/s390/s390.opt: Add z14 to processor_type enum.
18236 2017-07-31 Martin Jambor <mjambor@suse.cz>
18239 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
18240 regardless of optimization level.
18242 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
18243 Martin Liska <mliska@suse.cz>
18245 * predict.def: Remove old comment and adjust probability.
18246 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
18247 PREDICT statements.
18249 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18252 * config/i386/i386.c (ix86_function_naked): New function.
18253 (ix86_can_use_return_insn_p): Return false for naked functions.
18254 (ix86_expand_prologue): Skip prologue for naked functions.
18255 (ix86_expand_epilogue): Skip epilogue for naked functions
18256 and emit trap instruction.
18257 (ix86_warn_func_return): New function.
18258 (ix86_attribute_table): Add "naked" attribute specification.
18259 (TARGET_WARN_FUNC_RETURN): Define.
18260 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
18262 2017-07-31 Martin Liska <mliska@suse.cz>
18264 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
18265 (dump_gimple_bb_header): Always dump BB info.
18266 (pp_cfg_jump): Do not append info about BB when dumping a jump.
18268 2017-07-31 Martin Liska <mliska@suse.cz>
18271 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
18272 also with current_function_decl non-null equality.
18274 2017-07-31 Jakub Jelinek <jakub@redhat.com>
18277 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
18278 change type to the element type, instead add eltype variable and
18279 use it where we are interested in the element type.
18281 PR tree-optimization/81603
18282 * ipa-polymorphic-call.c
18283 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
18284 offset arithmetic in offset_int, bail out if the resulting bit offset
18285 doesn't fit into shwi.
18287 2017-07-31 Martin Liska <mliska@suse.cz>
18289 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
18290 (gimplify_save_expr): Fix comment.
18292 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
18295 * config/i386/i386.c (ix86_function_arg): Update arguments for
18297 (ix86_compute_frame_layout): Set the initial stack offset to
18298 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
18299 INCOMING_FRAME_SP_OFFSET.
18300 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
18301 stack before exception handler returns.
18302 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
18303 the 'ERROR_CODE' for exception handler.
18305 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
18307 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
18308 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
18309 (ASM_OUTPUT_REG_POP): Ditto.
18310 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
18311 instead of asm_fprintf to output pure string.
18313 2017-07-29 Jakub Jelinek <jakub@redhat.com>
18315 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
18316 to imported_module_or_decl hook.
18317 (debug_nothing_tree_tree_tree_bool): Remove.
18318 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
18319 * debug.c (do_nothing_debug_hooks): Use
18320 debug_nothing_tree_tree_tree_bool_bool instead of
18321 debug_nothing_tree_tree_tree_bool.
18322 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18323 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
18324 * sdbout.c (sdb_debug_hooks): Likewise.
18325 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
18326 (gen_namespace_die): Add DW_AT_export_symbols attribute if
18328 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
18329 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
18330 attribute, don't add anything.
18332 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18334 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
18335 (fold_build2_stat_loc): Likewise.
18336 (fold_build3_stat_loc): Likewise.
18337 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
18338 (fold_build1_loc): Remove macro.
18339 (fold_build2_loc): Likewise.
18340 (fold_build3_loc): Likewise.
18342 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18344 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
18345 (gimple_build_debug_bind_source_stat): Likewise.
18346 * gimple.h (gimple_build_debug_bind): Remove macro.
18347 (gimple_build_debug_bind_source): Likewise.
18349 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18351 * bitmap.c (bitmap_alloc): Adjust.
18352 (bitmap_gc_alloc): Likewise.
18353 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
18355 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18357 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
18358 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
18359 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
18360 (bitmap_gc_alloc_stat): Likewise.
18361 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
18363 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18365 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
18366 * rtl.h (shallow_copy_rtx): Remove macro.
18368 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18370 * emit-rtl.c (gen_raw_REG): Adjust.
18371 * gengenrtl.c (gendef): Likewise.
18372 * rtl.c (rtx_alloc_stat): Remove _stat from name.
18373 * rtl.h (rtx_alloc): Remove macro.
18375 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18377 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
18378 (build_tree_list_stat): Likewise.
18379 * tree.h (build_tree_list): Remove macro.
18380 (build_tree_list_vec): Likewise.
18382 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18384 * tree.c (make_vector_stat): Remove _stat from name.
18385 (build_vector_stat): Likewise.
18386 * tree.h (make_vector_stat): Remove macro.
18387 (build_vector_stat): Likewise.
18389 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18391 * tree.h (build_var_debug_value): Remove prototype.
18393 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18395 * tree.c (tree_cons_stat): Remove _stat from name.
18396 * tree.h (tree_cons): Remove macro.
18398 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18400 * tree.c (build_vl_exp_stat): Remove _stat from name.
18401 * tree.h (build_vl_exp): Remove macro.
18403 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18405 * tree.c (build_decl_stat): Remove _stat from name.
18406 * tree.h (build_decl): Remove macro.
18408 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18410 * gimple.c (gimple_build_with_ops_stat): Adjust.
18411 (gimple_alloc_stat): Remove _stat from name.
18412 * gimple.h (gimple_alloc): Remove macro.
18414 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18416 * tree.c (make_tree_vec_stat): Remove _stat from name.
18417 (grow_tree_vec_stat): Likewise.
18418 * tree.h (make_tree_vec_stat): Adjust prototype.
18419 (grow_tree_vec_stat): Likewise.
18420 (make_tree_vec): Remove macro.
18421 (grow_tree_vec): Likewise.
18423 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18425 * fold-const.c (fold_build1_stat_loc): Adjust.
18426 (fold_build2_stat_loc): Likewise.
18427 (fold_build3_stat_loc): Likewise.
18428 * tree.c (build0_stat): Remove _stat from name.
18429 (build1_stat): Likewise.
18430 (build2_stat): Likewise.
18431 (build3_stat): Likewise.
18432 (build4_stat): Likewise.
18433 (build5_stat): Likewise.
18434 * tree.h (build1_loc): Remove macro, and rename _stat function
18436 (build2_loc): Likewise.
18437 (build3_loc): Likewise.
18438 (build4_loc): Likewise.
18439 (build5_loc): Likewise.
18441 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18443 * tree.c (make_int_cst_stat): Remove _stat from name.
18444 * tree.h (make_int_cst_stat): Adjust prototype.
18445 (make_int_cst): Remove macro.
18447 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18449 * tree.c (make_tre_binfo_stat): Remove _stat from name.
18450 * tree.h (make_tree_binfo_stat): Adjust prototype.
18451 (make_tree_binfo): Remove.
18453 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18455 * tree.c (copy_node_stat): Rename to copy_node.
18456 (build_distinct_type_copy): Adjust.
18457 * tree.h (copy_node_stat): Adjust prototype.
18458 (copy_node): Remove macro.
18460 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18462 * tree.c (make_node_stat): rename to make_node.
18463 (build_tree_list_stat): Adjust.
18464 (build0_stat): Likewise.
18465 (build2_stat): Likewise.
18466 (build3_stat): Likewise.
18467 (build4_stat): Likewise.
18468 (build5_stat): Likewise.
18469 (build_decl_stat): Likewise.
18470 * tree.h (make_node_stat): Adjust prototype.
18471 (make_node): remove macro.
18473 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
18475 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
18476 (PPC_FEATURE2_SCV): Likewise.
18477 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
18479 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18481 * config/aarch64/aarch64.c
18482 (aarch64_internal_mov_immediate): Add new special pattern.
18483 * config/aarch64/aarch64.md (*movdi_aarch64):
18484 Add reg/32bit const mov case.
18486 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18487 Richard Sandiford <richard.sandiford@linaro.org>
18489 * config/aarch64/aarch64.md (mov<mode>): Generalize.
18490 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18491 Add integer and movi cases.
18492 (movi-split-hf-df-sf split, fp16): New.
18493 (enabled): Added TARGET_FP_F16INST.
18494 * config/aarch64/iterators.md (GPF_HF): New.
18495 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
18497 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18499 * config/aarch64/aarch64.c
18500 (aarch64_simd_container_mode): Add prototype.
18501 (aarch64_expand_mov_immediate): Add HI support.
18502 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
18503 (aarch64_can_const_movi_rtx_p): New.
18504 (aarch64_preferred_reload_class):
18505 Remove restrictions of using FP registers for certain SIMD operations.
18506 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
18507 (aarch64_valid_floating_const): Add integer move validation.
18508 (aarch64_simd_imm_scalar_p): Remove.
18509 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
18510 (aarch64_legitimate_constant_p): Expand list of supported cases.
18511 * config/aarch64/aarch64-protos.h
18512 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
18513 (aarch64_reinterpret_float_as_int): New.
18514 (aarch64_simd_imm_scalar_p): Remove.
18515 * config/aarch64/constraints.md (Uvi): New.
18516 (Dd): Split into Ds and new Dd.
18517 * config/aarch64/aarch64.md (*movsi_aarch64):
18519 (*movdi_aarch64): Add SIMD mov case.
18521 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18523 * tree-predcom.c: (struct chain): Handle store-store chain in which
18524 stores for elimination only store loop invariant values.
18525 (execute_pred_commoning_chain): Ditto.
18526 (prepare_initializers_chain_store_elim): Ditto.
18527 (prepare_finalizers): Ditto.
18528 (is_inv_store_elimination_chain): New function.
18529 (initialize_root_vars_store_elim_1): New function.
18531 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18533 * tree-predcom.c: Revise general description of the pass.
18534 (enum chain_type): New enum type for store elimination.
18535 (struct chain): New field supporting store elimination.
18536 (struct component): Ditto.
18537 (dump_chain): Dump store-stores chain.
18538 (release_chain): Release resources.
18539 (split_data_refs_to_components): Compute and create component
18540 contains only stores for elimination.
18541 (get_chain_last_ref_at): New function.
18542 (make_invariant_chain): Initialization.
18543 (make_rooted_chain): Specify chain type in parameter and record it.
18544 (add_looparound_copies): Skip for store-stores chain.
18545 (determine_roots_comp): Compute type of chain and pass it to
18547 (initialize_root_vars_store_elim_2): New function.
18548 (finalize_eliminated_stores): New function.
18549 (remove_stmt): Handle store for elimination.
18550 (execute_pred_commoning_chain): Execute predictive commoning on
18551 store-store chains.
18552 (determine_unroll_factor): Skip unroll for store-stores chain.
18553 (prepare_initializers_chain_store_elim): New function.
18554 (prepare_initializers_chain): Hanlde store-store chain.
18555 (prepare_finalizers_chain, prepare_finalizers): New function.
18556 (tree_predictive_commoning_loop): Return integer value indicating
18557 if loop is unrolled or lcssa form is corrupted.
18558 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
18560 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18562 * tree-predcom.c (initialize_root): Delete.
18563 (execute_pred_commoning_chain): Initialize root vars and replace
18564 reference of non-combined chain directly, rather than call above
18567 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18569 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
18570 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
18572 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18574 * tree-predcom.c (struct chain): New field init_seq.
18575 (release_chain): Release init_seq.
18576 (prepare_initializers_chain): Record intialization stmts in above
18578 (insert_init_seqs): New function.
18579 (tree_predictive_commoning_loop): Call insert_init_seqs.
18581 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18583 * tree-predcom.c (determine_roots_comp): Skip trivial components.
18585 2017-07-28 Richard Biener <rguenther@suse.de>
18587 * match.pd: Remove superfluous :c.
18588 * genmatch.c (simplify::id): Add member.
18589 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
18591 (current_id): New global.
18592 (dt_node::parent): Move from ...
18593 (dt_operand::parent): ... here. Add for_id member.
18594 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
18595 (decision_tree::find_node): Relax order requirement when
18596 merging DT_TRUE nodes to ones inbetween the current simplify
18597 and the one we try to merge with. Add diagnostic whenever
18598 we need to enforce pattern order by not merging.
18599 (decision_tree::insert): Set current_id.
18600 (decision_tree::print_node): Dump parent node and for_id.
18601 (parser::last_id): Add member.
18602 (parser::push_simplify): Assign unique id.
18603 (parser::parser): Initialize last_id.
18605 2017-07-28 Martin Liska <mliska@suse.cz>
18608 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
18609 gimple_build_debug_bind.
18611 2017-07-28 Richard Biener <rguenther@suse.de>
18613 PR tree-optimization/81502
18614 * match.pd: Add pattern combining BIT_INSERT_EXPR with
18616 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
18618 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
18619 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
18620 for BIT_FIELD_REF args.
18621 * fold-const.c (make_bit_field_ref): Likewise.
18622 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
18624 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18627 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
18628 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
18629 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
18630 Or it into SANITIZER_UNDEFINED.
18631 * ubsan.c: Include gimple-fold.h and varasm.h.
18632 (ubsan_expand_ptr_ifn): New function.
18633 (instrument_pointer_overflow): New function.
18634 (maybe_instrument_pointer_overflow): New function.
18635 (instrument_object_size): Formatting fix.
18636 (pass_ubsan::execute): Call instrument_pointer_overflow
18637 and maybe_instrument_pointer_overflow.
18638 * internal-fn.c (expand_UBSAN_PTR): New function.
18639 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
18640 * sanitizer.def (__ubsan_handle_pointer_overflow,
18641 __ubsan_handle_pointer_overflow_abort): New builtins.
18642 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
18643 * internal-fn.def (UBSAN_PTR): New internal function.
18644 * opts.c (sanitizer_opts): Add pointer-overflow.
18645 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
18646 * fold-const.c (build_range_check): Compute pointer range check in
18647 integral type if pointer arithmetics would be needed. Formatting
18650 2017-07-28 Martin Liska <mliska@suse.cz>
18653 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
18654 parameters that are of a variable-length.
18656 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
18658 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
18660 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
18661 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
18662 (CRT_CALL_STATIC_FUNCTION): Likewise.
18663 (ASM_DEFAULT_SPEC): New define.
18664 (ASM_SPEC32): Likewise.
18665 (ASM_SPEC64): Likewise.
18666 (ASM_SPEC_COMMON): Likewise.
18667 (ASM_SPEC): Likewise.
18668 (INVALID_64BIT): Likewise.
18669 (LINK_OS_DEFAULT_SPEC): Likewise.
18670 (LINK_OS_SPEC32): Likewise.
18671 (LINK_OS_SPEC64): Likewise.
18672 (POWERPC_LINUX): Likewise.
18673 (PTRDIFF_TYPE): Likewise.
18674 (RESTORE_FP_PREFIX): Likewise.
18675 (RESTORE_FP_SUFFIX): Likewise.
18676 (SAVE_FP_PREFIX): Likewise.
18677 (SAVE_FP_SUFFIX): Likewise.
18678 (SIZE_TYPE): Likewise.
18679 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
18680 (TARGET_64BIT): Likewise.
18681 (TARGET_64BIT): Likewise.
18682 (TARGET_AIX): Likewise.
18683 (WCHAR_TYPE_SIZE): Likewise.
18684 (WCHAR_TYPE): Undefine.
18685 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
18686 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
18687 (CPP_OS_RTEMS_SPEC): Delete.
18688 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
18689 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
18691 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
18693 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18695 PR tree-optimization/81578
18696 * tree-parloops.c (build_new_reduction): Bail out if
18697 reduction_code isn't one of the standard OpenMP reductions.
18698 Move the details printing after that decision.
18700 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18702 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
18703 related to reload_in_progress.
18704 (splat_input_operand): Likewise.
18705 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
18707 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
18709 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
18710 (TARGET_INSTANTIATE_DECLS): Likewise.
18711 (legitimate_indexed_address_p): Delete reload_in_progress code.
18712 (rs6000_debug_legitimate_address_p): Likewise.
18713 (rs6000_eliminate_indexed_memrefs): Likewise.
18714 (rs6000_emit_le_vsx_store): Likewise.
18715 (rs6000_emit_move_si_sf_subreg): Likewise.
18716 (rs6000_emit_move): Likewise.
18717 (register_to_reg_type): Likewise.
18718 (rs6000_pre_atomic_barrier): Likewise.
18719 (rs6000_machopic_legitimize_pic_address): Likewise.
18720 (rs6000_allocate_stack_temp): Likewise.
18721 (rs6000_address_for_fpconvert): Likewise.
18722 (rs6000_address_for_altivec): Likewise.
18723 (rs6000_secondary_memory_needed_rtx): Delete function.
18724 (rs6000_check_sdmode): Likewise.
18725 (rs6000_alloc_sdmode_stack_slot): Likewise.
18726 (rs6000_instantiate_decls): Likewise.
18727 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
18728 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
18729 Delete reload_in_progress.
18730 (*vec_reload_and_plus_<mptrsize>): Likewise.
18731 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
18732 (vsx_div_v2di): Likewise.
18733 (vsx_udiv_v2di): Likewise.
18735 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18737 * config/rs6000/rs6000.opt (mlra): Replace with stub.
18738 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
18739 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
18740 (rs6000_debug_reg_global): Delete print of LRA status.
18741 (rs6000_option_override_internal): Delete dead LRA related code.
18742 (rs6000_lra_p): Delete function.
18743 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
18745 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
18747 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
18748 * config/riscv/rtems.h: New file.
18750 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18751 Sudakshina Das <sudi.das@arm.com>
18753 * config/aarch64/aarch64.md
18754 (define_split for and<mode>3nr_compare): Move
18755 non aarch64_logical_operand to a register.
18756 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
18757 register immediate operand to a register.
18758 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
18760 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18762 PR middle-end/81564
18763 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
18765 2017-07-27 Richard Biener <rguenther@suse.de>
18767 PR tree-optimization/81573
18768 PR tree-optimization/81494
18769 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
18770 multi defuse cycle case.
18772 2017-07-27 Richard Biener <rguenther@suse.de>
18774 PR tree-optimization/81571
18775 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
18778 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
18780 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
18781 earlier and only if MASK_FPU is set. Adjust formatting.
18783 2017-07-27 Martin Liska <mliska@suse.cz>
18785 * opt-functions.awk: Add validation of value of Init.
18786 * optc-gen.awk: Pass new argument.
18788 2017-07-27 Martin Liska <mliska@suse.cz>
18790 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
18791 Fix wrong condition.
18793 2017-07-27 Martin Liska <mliska@suse.cz>
18795 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
18796 BBs and edges seen by autoFDO.
18798 2017-07-27 Richard Biener <rguenther@suse.de>
18800 PR tree-optimization/81502
18801 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
18802 with incompatible but same sized type.
18803 (execute_update_addresses_taken): Likewise.
18805 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
18807 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
18808 flag_tree_loop_vectorize rather than flag_tree_vectorize.
18810 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18813 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
18814 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
18815 Change s_operand to memory_operand.
18817 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
18819 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
18820 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
18821 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
18822 Emit instructions rather than returning an expression. Handle TFmode
18823 and KFmode by casting to TImode.
18824 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
18825 (rs6000_emit_le_vsx_store): Likewise.
18826 * config/rs6000/vsx.md (VSX_TI): New iterator.
18827 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
18828 (*vsx_le_undo_permute_<mode>): Likewise.
18829 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
18830 emit the split sequence.
18831 (*vsx_le_perm_store_<mode>): Likewise.
18833 2017-07-27 Jakub Jelinek <jakub@redhat.com>
18835 PR tree-optimization/81555
18836 PR tree-optimization/81556
18837 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
18838 if true, force CHANGED for the recursive invocation.
18839 (reassociate_bb): Remember original length of ops array, pass
18840 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
18842 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
18843 attributes for noipa attribute. For naked attribute use
18844 lookup_attribute first before lookup_attribute_spec.
18845 * final.c (rest_of_handle_final): Disable IPA RA for functions with
18847 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
18848 for functions with noipa attribute.
18849 (cgraph_externally_visible_p): Return true for functions with noipa
18851 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
18852 for functions with noipa attribute.
18853 * doc/extend.texi: Document noipa function attribute.
18854 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
18855 also for functions with noipa attribute.
18856 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
18858 2017-07-26 Andrew Pinski <apinski@cavium.com>
18860 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
18861 vec_unalign_load_cost and vec_unalign_store_cost.
18863 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
18865 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
18866 -mvsx-small-integer option.
18867 (ISA_3_0_MASKS_IEEE): Likewise.
18868 (OTHER_VSX_VECTOR_MASKS): Likewise.
18869 (POWERPC_MASKS): Likewise.
18870 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
18871 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
18872 code, only testing for DImode being allowed in non-VSX floating
18874 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
18875 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
18877 (rs6000_option_override_internal): Delete -mvsx-small-integer.
18878 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
18879 TARGET_P8_VECTOR test.
18880 (rs6000_secondary_reload_simple_move): Likewise.
18881 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
18882 since TARGET_P9_VECTOR was already tested.
18883 (rs6000_opt_masks): Remove -mvsx-small-integer.
18884 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
18885 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18887 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
18888 test for TARGET_VEXTRACTUB was used, and that uses
18890 (p9 extract splitter): Likewise.
18891 (vsx_extract_<mode>_di_p9): Likewise.
18892 (vsx_extract_<mode>_store_p9): Likewise.
18893 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
18894 for TARGET_P9_VECTOR was used. Delete code that is now dead with
18895 the elimination of TARGET_VSX_SMALL_INTEGER.
18896 (vsx_extract_<mode>_p8): Likewise.
18897 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
18898 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
18899 (vsx_set_<mode>_p9): Likewise.
18900 (vsx_set_v4sf_p9): Likewise.
18901 (vsx_set_v4sf_p9_zero): Likewise.
18902 (vsx_insert_extract_v4sf_p9): Likewise.
18903 (vsx_insert_extract_v4sf_p9_2): Likewise.
18904 * config/rs6000/rs6000.md (sign extend splitter): Change
18905 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
18906 (floatsi<mode>2_lfiwax_mem): Likewise.
18907 (floatunssi<mode>2_lfiwzx_mem): Likewise.
18908 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
18909 since a test for TARGET_P9_VECTOR was used.
18910 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18911 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
18912 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18913 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18914 TARGET_P8_VECTOR test.
18915 (fix_trunc<mode>si2_stfiwx): Likewise.
18916 (fix_trunc<mode>si2_internal): Likewise.
18917 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
18918 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18920 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18921 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18922 TARGET_P8_VECTOR test.
18923 (fixuns_trunc<mode>si2_stfiwx): Likewise.
18924 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
18925 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18927 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18928 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
18929 since a test for TARGET_P9_VECTOR was used.
18930 (splitter for loading small constants): Likewise.
18932 2017-07-26 Andrew Pinski <apinski@cavium.com>
18934 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
18937 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
18940 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
18941 (fp_valid_at): Likewise.
18943 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18945 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
18946 (qdf24xx_addrcost_table): Likewise.
18947 (cortexa57_tunings): Update to use generic_branch_cost.
18948 (cortexa72_tunings): Likewise.
18949 (cortexa73_tunings): Likewise.
18950 (qdf24xx_tunings): Likewise.
18952 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18954 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
18955 (thunderx2t99_branch_cost): Likewise.
18956 (cortexa35_tunings): Update to use generic_branch_cost.
18957 (cortexa53_tunings): Likewise.
18958 (cortexa57_tunings): Likewise.
18959 (cortexa72_tunings): Likewise.
18960 (cortexa73_tunings): Likewise.
18961 (thunderx2t99_tunings): Likewise.
18963 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
18965 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
18966 (sparc_option_override): Honour MASK_FSMULD.
18967 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
18968 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
18969 * config/sparc/sparc.opt (mfsmuld): New option.
18970 * doc/invoke.texi (mfsmuld): Document option.
18972 2017-07-26 Marek Polacek <polacek@redhat.com>
18974 PR middle-end/70992
18975 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
18977 2017-07-26 Richard Biener <rguenther@suse.de>
18979 * gimple-match-head.c (do_valueize): Return OP if valueize
18981 (get_def): New helper to get at the def stmt of a SSA name
18982 if valueize allows.
18983 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
18984 do_valueize to get at the def stmt.
18985 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
18987 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
18989 PR middle-end/46932
18990 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
18992 2017-07-26 Martin Liska <mliska@suse.cz>
18995 * function.c (expand_function_start): Make expansion of
18996 nonlocal_goto_save_area after parm_birth_insn.
18998 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19000 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19001 from all CPU target flags enable members.
19003 2017-07-26 Richard Biener <rguenther@suse.de>
19005 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19006 (decision_tree::gen): Make 'type' const.
19007 (write_predicate): Likewise.
19009 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19011 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19012 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19013 (rs6000_option_override_internal): Likewise.
19014 (rs6000_expand_vector_set): Likewise.
19015 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19016 (TARGET_UPPER_REGS_SF): Likewise.
19017 (TARGET_UPPER_REGS_DI): Likewise.
19018 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19019 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19020 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19021 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19022 (Splitters for DI constants in Altivec registers): Likewise.
19023 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19024 (vsx_set_v4sf_p9): Likewise.
19025 (vsx_set_v4sf_p9_zero): Likewise.
19026 (vsx_insert_extract_v4sf_p9): Likewise.
19027 (vsx_insert_extract_v4sf_p9_2): Likewise.
19029 2017-07-25 Carl Love <cel@us.ibm.com>
19031 * doc/extend.texi: Update the built-in documentation file for the
19032 existing built-in functions
19033 vector signed char vec_cnttz (vector signed char);
19034 vector unsigned char vec_cnttz (vector unsigned char);
19035 vector signed short vec_cnttz (vector signed short);
19036 vector unsigned short vec_cnttz (vector unsigned short);
19037 vector signed int vec_cnttz (vector signed int);
19038 vector unsigned int vec_cnttz (vector unsigned int);
19039 vector signed long long vec_cnttz (vector signed long long);
19040 vector unsigned long long vec_cnttz (vector unsigned long long);
19042 2017-07-25 Andrew Pinski <apinski@cavium.com>
19044 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19045 accesses where the use is for the first operand of a BIT_INSERT.
19047 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19050 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19051 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19053 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19055 * config/i386/gstabs.h: Delete.
19056 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19058 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19060 * config/i386/i386.c (ix86_decompose_address): Do not check for
19061 register RTX when looking at index_reg or base_reg.
19062 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
19064 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
19066 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
19067 to update EH info here.
19069 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19071 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
19073 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19075 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
19077 2017-07-25 Torsten Duwe <duwe@suse.de>
19079 * common.opt: Introduce -fpatchable-function-entry
19080 command line option, and its variables function_entry_patch_area_size
19081 and function_entry_patch_area_start.
19082 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
19083 including a two-value parser.
19084 * target.def (print_patchable_function_entry): New target hook.
19085 * targhooks.h (default_print_patchable_function_entry): New function.
19086 * targhooks.c (default_print_patchable_function_entry): Likewise.
19087 * toplev.c (process_options): Switch off IPA-RA if
19088 patchable function entries are being generated.
19089 * varasm.c (assemble_start_function): Look at the
19090 patchable-function-entry command line switch and current
19091 function attributes and maybe generate NOP instructions by
19092 calling the print_patchable_function_entry hook.
19093 * doc/extend.texi: Document patchable_function_entry attribute.
19094 * doc/invoke.texi: Document -fpatchable_function_entry
19095 command line option.
19096 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
19098 * doc/tm.texi: Re-generate.
19100 2017-07-25 Jakub Jelinek <jakub@redhat.com>
19103 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
19104 TARGET_AVX512DQ rather than TARGET_AVX512BW.
19106 2017-07-25 Tamar Christina <tamar.christina@arm.com>
19108 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
19110 2017-07-25 Richard Biener <rguenther@suse.de>
19112 PR tree-optimization/81455
19113 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
19114 not walk in cycles when looking for guards.
19116 2017-07-25 Richard Biener <rguenther@suse.de>
19118 PR tree-optimization/81529
19119 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
19120 when optimizing backedge uses.
19122 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
19124 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
19126 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
19127 to dl_section_ref. On AIX, append an expression to subtract
19128 the size of the section length to dl_section_ref.
19130 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
19132 * configure.ac: If any of the config.* scripts fail, exit 1.
19133 * configure: Regenerate.
19135 2017-07-25 Richard Biener <rguenther@suse.de>
19137 PR middle-end/81546
19138 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
19139 of immediate uses, be more verbose on errors.
19141 2017-07-25 Richard Biener <rguenther@suse.de>
19143 PR tree-optimization/81510
19144 * tree-vect-loop.c (vect_is_simple_reduction): When the
19145 reduction stmt is not inside the loop bail out.
19147 2017-07-25 Richard Biener <rguenther@suse.de>
19149 PR tree-optimization/81303
19150 * tree-vect-loop-manip.c (vect_loop_versioning): Build
19151 profitability check against LOOP_VINFO_NITERSM1.
19153 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19155 * domwalk.c (cmp_bb_postorder): Simplify.
19156 (sort_bbs_postorder): New function. Use it...
19157 (dom_walker::walk): ...here to optimize common cases.
19159 2017-07-25 Martin Liska <mliska@suse.cz>
19162 * ipa-visibility.c (function_and_variable_visibility): Make the
19163 redirection just on target that supports aliasing.
19164 Fix GNU coding style.
19166 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
19169 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
19171 * config/arm/rtems.h: Likewise.
19172 * config/bfin/rtems.h: Likewise.
19173 * config/i386/rtemself.h: Likewise.
19174 * config/lm32/rtems.h: Likewise.
19175 * config/m32c/rtems.h: Likewise.
19176 * config/m68k/rtemself.h: Likewise.
19177 * config/microblaze/rtems.h: Likewise.
19178 * config/mips/rtems.h: Likewise.
19179 * config/moxie/rtems.h: Likewise.
19180 * config/nios2/rtems.h: Likewise.
19181 * config/powerpcspe/rtems.h: Likewise.
19182 * config/rs6000/rtems.h: Likewise.
19183 * config/rtems.h: Likewise.
19184 * config/sh/rtems.h: Likewise.
19185 * config/sh/rtemself.h: Likewise.
19186 * config/sparc/rtemself.h: Likewise.
19188 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19191 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
19192 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
19193 * tree-ssa-structalias.c (alias_get_name): Same.
19195 2017-07-25 Bin Cheng <bin.cheng@arm.com>
19198 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
19199 instructions if no du chain is found.
19201 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19203 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
19205 2017-07-25 Richard Biener <rguenther@suse.de>
19207 PR middle-end/81505
19208 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
19211 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19213 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
19214 upper-regs options.
19215 (ISA_2_7_MASKS_SERVER): Likewise.
19216 (ISA_3_0_MASKS_IEEE): Likewise.
19217 (OTHER_P8_VECTOR_MASKS): Likewise.
19218 (OTHER_VSX_VECTOR_MASKS): Likewise.
19219 (POWERPC_MASKS): Likewise.
19220 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
19221 duplicate list of options.
19222 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
19223 explicit -mupper-regs options.
19224 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
19225 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
19226 alias for -mupper-regs-df.
19227 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19228 (rs6000_init_hard_regno_mode_ok): Likewise.
19229 (rs6000_option_override_internal): Likewise.
19230 (rs6000_opt_masks): Likewise.
19231 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
19232 options in terms of whether -mvsx or -mpower8-vector was used.
19233 (TARGET_UPPER_REGS_DI): Likewise.
19234 (TARGET_UPPER_REGS_SF): Likewise.
19235 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
19236 -mupper-regs-* options.
19238 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19240 * passes.c (emergency_dump_function): Print some empty lines and a
19241 header before the RTL dump.
19243 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19245 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
19247 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
19250 * config/aarch64/aarch64.c (aarch64_classify_symbol):
19251 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
19253 2017-07-24 Carl Love <cel@us.ibm.com>
19255 * config/rs6000/rs6000-c.c: Add support for built-in functions
19256 vector float vec_extract_fp32_from_shorth (vector unsigned short);
19257 vector float vec_extract_fp32_from_shortl (vector unsigned short);
19258 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
19259 vec_extract_fp_from_shortl): Add defines for the two builtins.
19260 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
19261 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
19263 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
19264 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
19265 * doc/extend.texi: Update the built-in documentation file for the
19266 new built-in function.
19268 2017-07-24 Jakub Jelinek <jakub@redhat.com>
19271 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
19273 * doc/generic.texi: Likewise.
19274 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19275 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19277 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
19279 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
19280 (aarch64_mls_elt_merge<mode>): Likewise.
19282 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
19284 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19285 having __cxa_atexit.
19287 2017-07-23 Michael Collison <michael.collison@arm.com>
19289 * config/arm/arm.c (arm_option_override): Deprecate
19290 use of -mstructure-size-boundary.
19291 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
19292 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
19294 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19297 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19298 Reduce cost estimate for direct moves.
19300 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
19303 * config/i386/i386.c (ix86_option_override_internal): Disable
19304 BMI, BMI2 and TBM instructions for -m16.
19306 2017-07-21 Carl Love <cel@us.ibm.com>
19308 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19309 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19310 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19311 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19312 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19313 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19314 VMULOSW): New enum "unspec" values.
19315 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19316 altivec_vmulosw): New patterns.
19317 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19318 VMULOSW): Add definitions.
19320 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
19322 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
19323 (qdf24xx): Likewise.
19324 * config/aarch64/aarch64-options-extensions.def (rdma); New.
19325 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
19326 (AARCH64_FL_V8_1): Renumber.
19327 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
19328 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
19329 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
19330 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
19331 rdma to feature modifiers list.
19333 2017-07-21 Yury Gribov <tetra2005@gmail.com>
19335 PR middle-end/56727
19336 * ipa-visibility (function_and_variable_visibility): Convert
19337 recursive PLT call to direct call if appropriate.
19339 2017-07-21 Andrew Pinski <apinski@cavium.com>
19341 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
19342 operand 1 to see if the types precision matches.
19343 * fold-const.c (operand_equal_p): Likewise.
19345 2017-07-21 Richard Biener <rguenther@suse.de>
19347 PR tree-optimization/81303
19348 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
19349 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
19350 (vect_peeling_hash_get_lowest_cost): Adjust.
19351 (vect_enhance_data_refs_alignment): Likewise. Use
19352 vect_get_peeling_costs_all_drs to compute the penalty for no
19353 peeling to match up costs.
19355 2017-07-21 Richard Biener <rguenther@suse.de>
19357 PR tree-optimization/81500
19358 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
19359 we didn't identify a reduction path.
19361 2017-07-21 Tom de Vries <tom@codesourcery.com>
19362 Cesar Philippidis <cesar@codesourcery.com>
19364 PR gcov-profile/81442
19365 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
19368 2017-07-21 Tom de Vries <tom@codesourcery.com>
19371 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
19373 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
19374 nvptx_override_options_after_change.
19376 2017-07-21 Ulrich Drepper <drepper@redhat.com>
19378 * dwarf2out.c (output_file_names): Avoid double testing for
19379 dwarf_version >= 5.
19381 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
19383 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
19385 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19387 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
19389 (try_crossjump_to_edge): Do not punt on partitioned functions.
19391 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19393 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
19394 Put all BBs reachable only via paths crossing cold region to cold
19396 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
19398 2016-07-21 Richard Biener <rguenther@suse.de>
19400 PR tree-optimization/81303
19401 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
19402 into account prologue and epilogue iterations when raising
19403 min_profitable_iters to sth at least covering one vector iteration.
19405 2017-07-21 Tamar Christina <tamar.christina@arm.com>
19407 * config/arm/arm.c (arm_test_cpu_arch_dat):
19410 2017-07-20 Nathan Sidwell <nathan@acm.org>
19412 Remove TYPE_METHODS.
19413 * tree.h (TYPE_METHODS): Delete.
19414 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
19415 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
19416 (dbxout_type_methods): Scan TYPE_FIELDS.
19417 (dbxout_type): Don't check TYPE_METHODS here.
19418 * function.c (use_register_for_decl): Always ignore register for
19419 class types when not optimizing.
19420 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
19421 * tree.c (free_lang_data_in_type): Stitch out member functions and
19422 templates from TYPE_FIELDS.
19423 (build_distinct_type_copy, verify_type_variant,
19424 verify_type): Member fns are on TYPE_FIELDS.
19425 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
19426 * tree-pretty-print.c (dump_generic_node): Likewise.
19428 2017-07-20 Jakub Jelinek <jakub@redhat.com>
19431 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
19432 V2TImode and V4TImode.
19433 (ix86_expand_vector_extract): Likewise.
19434 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
19435 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
19436 (ssescalarmode): Handle V4TImode and V2TImode.
19437 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
19438 (*vec_extractv2ti, *vec_extractv4ti): New insns.
19439 (VEXTRACTI128_MODE): New mode iterator.
19440 (splitter for *vec_extractv?ti first element): New.
19441 (VEC_INIT_MODE): New mode iterator.
19442 (vec_init<mode>): Consolidate 3 expanders into one using
19443 VEC_INIT_MODE mode iterator.
19445 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19447 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
19448 non_spilled_static_chain_regno_p.
19450 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19452 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
19454 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
19456 * bb-reorder.c (connect_traces): Allow copying of blocks within
19459 2017-07-20 Richard Biener <rguenther@suse.de>
19461 * gimple.h (gimple_phi_result): Add gphi * overload.
19462 (gimple_phi_result_ptr): Likewise.
19463 (gimple_phi_arg): Likewise. Adjust index assert to only
19464 allow actual argument accesses rather than all slots available
19466 (gimple_phi_arg_def): Add gphi * overload.
19467 * tree-phinodes.c (make_phi_node): Initialize only actual
19469 (resize_phi_node): Clear memory not covered by old node,
19470 do not initialize excess argument slots.
19471 (reserve_phi_args_for_new_edge): Initialize new argument slot
19474 2017-07-20 Bin Cheng <bin.cheng@arm.com>
19476 PR tree-optimization/81388
19478 2016-07-21 Bin Cheng <bin.cheng@arm.com>
19480 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
19481 by removing computation of may_be_zero.
19483 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19484 Tom de Vries <tom@codesourcery.com>
19486 PR middle-end/81030
19487 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
19488 when gimple level profile disagrees with what RTL expander did.
19490 2017-07-20 Richard Biener <rguenther@suse.de>
19492 PR tree-optimization/61171
19493 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
19494 (vect_analyze_stmt): Add slp instance parameter.
19495 (vectorizable_reduction): Likewise.
19496 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
19497 (vect_is_simple_reduction): Deal with chains not detected
19498 as SLP reduction chain, specifically not properly associated
19499 chains containing a mix of plus/minus.
19500 (get_reduction_op): Remove.
19501 (get_initial_defs_for_reduction): Simplify, pass in whether
19502 this is a reduction chain, pass in the SLP node for the PHIs.
19503 (vect_create_epilog_for_reduction): Get the SLP instance as
19505 (vectorizable_reduction): Get the SLP instance as arg.
19506 During analysis remember the SLP node with the PHIs in the
19507 instance. Simplify getting at the vectorized reduction PHIs.
19508 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
19509 through SLP instance.
19510 (vect_slp_analyze_operations): Likewise.
19511 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
19512 (vect_transform_stmt): Likewise.
19514 2017-07-20 Tom de Vries <tom@codesourcery.com>
19516 PR tree-optimization/81489
19517 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
19518 read of phi arg location to before loop that modifies phi.
19520 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19522 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
19525 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19527 PR middle-end/81331
19528 * except.c (execute): Fix ordering issue.
19530 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19532 PR rtl-optimization/81423
19533 * combine.c (make_compound_operation_int): Don't try to optimize
19534 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
19536 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19538 PR rtl-optimization/81423
19539 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
19540 with a constant that is -1 in the truncated to mode.
19542 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19544 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
19545 (determine_unlikely_bbs): ... here.
19546 * predict.h (propagate_unlikely_bbs_forward): Declare.
19547 * cfgexpand.c (pass_expand::execute): Use it.
19548 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
19550 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
19551 propagate_unlikely_bbs_forward.
19553 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19555 PR middle-end/81331
19556 * except.c (maybe_add_nop_after_section_switch): New function.
19559 2017-07-19 Tom de Vries <tom@codesourcery.com>
19561 * gimple.h (gimple_phi_set_arg): Make assert more strict.
19563 2017-07-19 Tom de Vries <tom@codesourcery.com>
19565 * gimple.h (gimple_phi_arg): Make assert more strict.
19567 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
19569 * config.gcc (powerpc*-*-*): Add mmintrin.h.
19570 * config/rs6000/mmintrin.h: New file.
19571 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
19573 2017-07-19 Jakub Jelinek <jakub@redhat.com>
19575 PR tree-optimization/81346
19576 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
19578 2017-07-19 Tom de Vries <tom@codesourcery.com>
19580 * config/nvptx/nvptx.md (VECIM): Add V2DI.
19582 2017-07-19 Tom de Vries <tom@codesourcery.com>
19584 * config/nvptx/nvptx-modes.def: Add V2DImode.
19585 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
19586 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
19587 (nvptx_output_mov_insn): Handle lack of mov.b128.
19588 (nvptx_print_operand): Handle 'H' and 'L' codes.
19589 (nvptx_vector_mode_supported): Allow V2DImode.
19590 (nvptx_preferred_simd_mode): New function.
19591 (nvptx_data_alignment): New function.
19592 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
19593 nvptx_preferred_simd_mode.
19594 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
19596 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
19598 2017-07-19 Tom de Vries <tom@codesourcery.com>
19600 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
19601 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
19602 (nvptx_vector_mode_supported): New function. Allow V2SImode.
19603 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
19604 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
19605 (mov<VECIM>_insn): New define_insn.
19606 (define_expand "mov<VECIM>): New define_expand.
19608 2017-07-19 Tom de Vries <tom@codesourcery.com>
19610 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
19612 2017-07-19 Jakub Jelinek <jakub@redhat.com>
19614 PR tree-optimization/81346
19615 * fold-const.h (fold_div_compare, range_check_type): Declare.
19616 * fold-const.c (range_check_type): New function.
19617 (build_range_check): Use range_check_type.
19618 (fold_div_compare): No longer static, rewritten into
19619 a match.pd helper function.
19620 (fold_comparison): Don't call fold_div_compare here.
19621 * match.pd (X / C1 op C2): New optimization using fold_div_compare
19622 as helper function.
19624 2017-07-19 Nathan Sidwell <nathan@acm.org>
19626 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
19627 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
19628 * tree.c (find_decls_types_r, verify_type): Use
19629 TYPE_{MIN,MAX}_VALUE_RAW.
19630 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
19631 (hash_tree): Likewise.
19632 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19634 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19637 2017-07-18 Tom de Vries <tom@codesourcery.com>
19639 PR middle-end/81464
19640 * omp-expand.c (expand_omp_for_static_chunk): Handle
19641 equal-argument loop exit phi.
19643 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
19646 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
19647 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
19648 operand 2 predicate.
19649 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
19650 operand 2 predicate.
19651 (ror,rol -> rorx splitters): Use const_int_operand as
19652 operand 2 predicate.
19654 2017-06-18 Richard Biener <rguenther@suse.de>
19656 PR tree-optimization/81410
19657 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
19658 the gap in the ! slp_perm SLP case after each group.
19660 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19662 PR middle-end/81463
19663 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
19666 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19668 PR middle-end/81462
19669 * predict.c (set_even_probabilities): Cleanup; do not affect
19670 probabilities that are already known.
19671 (combine_predictions_for_bb): Call even when count is set.
19673 2017-07-18 Nathan Sidwell <nathan@acm.org>
19675 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
19678 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19681 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
19682 optimization for loop niter analysis.
19684 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
19687 * config/avr/avr.c (avr_optimize_casesi): Don't use
19688 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
19690 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
19692 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
19693 body_cost_vec from _vect_peel_extended_info.
19694 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
19695 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
19698 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19700 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
19702 2017-07-18 Richard Biener <rguenther@suse.de>
19704 PR tree-optimization/80620
19705 PR tree-optimization/81403
19706 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
19707 info when re-using a VN table entry.
19709 2017-07-18 Richard Biener <rguenther@suse.de>
19711 PR tree-optimization/81418
19712 * tree-vect-loop.c (vectorizable_reduction): Properly compute
19713 vectype_in. Verify that with lane-reducing reduction operations
19714 we have a single def-use cycle.
19716 2017-07-17 Carl Love <cel@us.ibm.com>
19718 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
19720 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19721 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19722 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19723 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19724 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19725 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19726 VMULOSW): New enum "unspec" values.
19727 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19728 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19729 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19730 altivec_vmulosw): New patterns.
19731 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19732 VMULOSW): Add definitions.
19733 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
19735 * config/alpha/alpha.c: Include predict.h.
19737 2017-07-17 Yury Gribov <tetra2005@gmail.com>
19739 * tree-vrp.c (compare_assert_loc): Fix comparison function
19740 to return predictable results.
19742 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19744 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
19746 (subdi3): Likewise.
19747 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
19748 * doc/invoke.texi (mexpand-adddi): Update text.
19750 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19752 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
19753 that also clobbers the CC register. The old expand code is moved
19755 (*arc_clzsi2): ... here.
19756 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
19757 the CC register. The old expand code is moved to ...
19758 (arc_ctzsi2): ... here.
19760 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19762 * config/arc/arc.opt (mindexed-loads): Use initial value
19763 TARGET_INDEXED_LOADS_DEFAULT.
19764 (mauto-modify-reg): Use initial value
19765 TARGET_AUTO_MODIFY_REG_DEFAULT.
19766 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19767 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19768 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19769 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19771 2017-07-17 Martin Liska <mliska@suse.cz>
19774 * opts.c (finish_options): Do not allow -fgnu-tm
19775 w/ -fsanitize={kernel-,}address. Say sorry.
19777 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19780 * tree-loop-distribution.c (classify_partition): Only assert on
19781 numer of iterations.
19782 (merge_dep_scc_partitions): Delete prameter. Update function call.
19783 (distribute_loop): Remove code handling loop with unknown niters.
19784 (pass_loop_distribution::execute): Skip loop with unknown niters.
19786 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19789 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
19790 function sort_partitions_by_post_order.
19792 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19794 PR tree-optimization/81374
19795 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
19796 the max index of basic blocks, rather than number of basic blocks.
19798 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19800 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
19802 (arc_legitimate_pic_operand_p): Likewise.
19803 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
19805 (arc_needs_pcl_p): Likewise.
19806 (arc_legitimate_pc_offset_p): Likewise.
19807 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
19808 function is also used in constrains.md.
19809 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
19810 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
19811 PLUS. Only return true/false in known cases, otherwise assert.
19812 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
19813 is already called in arc_legitimate_constant_p.
19814 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
19816 (LEGITIMATE_PIC_OPERAND_P): Use
19817 arc_raw_symbolic_reference_mentioned_p function.
19818 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
19823 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19824 Andrew Burgess <andrew.burgess@embecosm.com>
19826 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
19827 (arc_return_address_register): New function.
19828 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
19829 (arc_handle_fndecl_attribute): Add naked attribute.
19830 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19831 (TARGET_WARN_FUNC_RETURN): Likewise.
19832 (arc_allocate_stack_slots_for_args): New function.
19833 (arc_warn_func_return): Likewise.
19834 (machine_function): Change type fn_type.
19835 (arc_compute_function_type): Consider new naked function type,
19836 change function return type.
19837 (arc_must_save_register): Adapt to handle new
19838 arc_compute_function_type's return type.
19839 (arc_expand_prologue): Likewise.
19840 (arc_expand_epilogue): Likewise.
19841 (arc_return_address_regs): Delete.
19842 (arc_return_address_register): New function.
19843 (arc_epilogue_uses): Use above function.
19844 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
19845 (arc_function_type): Change encoding, add naked type.
19846 (ARC_INTERRUPT_P): Change to handle the new encoding.
19847 (ARC_FAST_INTERRUPT_P): Likewise.
19848 (ARC_NORMAL_P): Define.
19849 (ARC_NAKED_P): Likewise.
19850 (arc_compute_function_type): Delete prototype.
19851 * config/arc/arc.md (in_ret_delay_slot): Use
19852 arc_return_address_register function.
19853 (simple_return): Likewise.
19854 (p_return_i): Likewise.
19856 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19858 PR tree-optimization/81428
19859 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
19860 can't be built for those types.
19862 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19864 Remove stuff dead since r239246.
19866 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
19867 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
19868 (avr_inform_devices): Remove dead stuff.
19870 2017-07-17 Tamar Christina <tamar.christina@arm.com>
19872 * config/arm/arm_neon.h: Fix softp typo.
19874 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19876 PR tree-optimization/81365
19877 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
19878 aggregate moves onto bb predecessor edges, make sure there are no
19879 loads that could alias the lhs in between the start of bb and the
19882 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19885 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
19886 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
19887 [LSHIFTRT, outer_code = TRUNCATE]: Same.
19889 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19891 PR tree-optimization/81396
19892 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
19893 (init_symbolic_number): Initialize it to 1.
19894 (perform_symbolic_merge): Add n_ops from both operands into the new
19896 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
19897 without base_addr as useless if they need more than one operation.
19898 (bswap_replace): Handle !bswap case for NULL base_addr.
19900 2017-07-17 Tom de Vries <tom@codesourcery.com>
19903 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
19906 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
19908 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
19909 conditional builtin define __FIX_LEON3FT_B2BST.
19911 2017-07-17 Daniel Cederman <cederman@gaisler.com>
19913 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
19914 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
19917 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19919 PR rtl-optimization/81424
19920 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
19921 to remove potential trapping from operands if -fnon-call-exceptions.
19923 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19925 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
19926 profile_proability for scalling.
19927 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
19929 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19931 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
19933 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19935 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
19936 fixpoint arithmetics.
19938 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19940 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
19941 fixpoint arithmetics.
19943 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19945 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
19946 fixpoint arithmetics.
19948 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19950 * profile-count.h (profile_probability::from_reg_br_prob_note,
19951 profile_probability::to_reg_br_prob_note): New functions.
19952 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
19953 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
19954 * predict.c (probability_reliable_p): Update.
19955 (edge_probability_reliable_p): Update.
19956 (br_prob_note_reliable_p): Update.
19957 (invert_br_probabilities): Update.
19958 (add_reg_br_prob_note): New function.
19959 (combine_predictions_for_insn): Update.
19960 * asan.c (asan_clear_shadow): Update.
19961 * cfgbuild.c (compute_outgoing_frequencies): Update.
19962 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
19963 (update_br_prob_note): Update.
19964 (rtl_verify_edges): Update.
19965 (purge_dead_edges): Update.
19966 (fixup_reorder_chain): Update.
19967 * emit-rtl.c (try_split): Update.
19968 * ifcvt.c (cond_exec_process_insns): Update.
19969 (cond_exec_process_if_block): Update.
19970 (dead_or_predicable): Update.
19971 * internal-fn.c (expand_addsub_overflow): Update.
19972 (expand_neg_overflow): Update.
19973 (expand_mul_overflow): Update.
19974 * loop-doloop.c (doloop_modify): Update.
19975 * loop-unroll.c (compare_and_jump_seq): Update.
19976 * optabs.c (emit_cmp_and_jump_insn_1): Update.
19977 * predict.h: Update.
19978 * reorg.c (mostly_true_jump): Update.
19980 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
19981 * config/alpha/alpha.c (emit_unlikely_jump): Update.
19982 * config/arc/arc.c: (emit_unlikely_jump): Update.
19983 * config/arm/arm.c: (emit_unlikely_jump): Update.
19984 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
19985 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
19986 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
19987 (ix86_print_operand): Update.
19988 (ix86_split_fp_branch): Update.
19989 (predict_jump): Update.
19990 * config/ia64/ia64.c (ia64_print_operand): Update.
19991 * config/mmix/mmix.c (mmix_print_operand): Update.
19992 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
19993 (rs6000_expand_split_stack_prologue): Update.
19994 * config/rs6000/rs6000.c: Update.
19995 * config/s390/s390.c (s390_expand_vec_strlen): Update.
19996 (s390_expand_vec_movstr): Update.
19997 (s390_expand_cs_tdsi): Update.
19998 (s390_expand_split_stack_prologue): Update.
19999 * config/sh/sh.c (sh_print_operand): Update.
20000 (expand_cbranchsi4): Update.
20001 (expand_cbranchdi4): Update.
20002 * config/sparc/sparc.c (output_v9branch): Update.
20003 * config/spu/spu.c (get_branch_target): Update.
20004 (ea_load_store_inline): Update.
20005 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20006 * config/tilepro/tilepro.c: Update.
20008 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20010 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20011 (gimplify_save_expr): Likewise.
20013 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20015 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20017 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20019 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20021 * timevar.def (TV_IPA_FNSUMMARY): Define.
20023 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20025 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20026 to back store errata sensitive sequence from being generated.
20027 (sqrtdf2_fix): Likewise.
20029 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20031 * tree-ssa-threadupdate.c (compute_path_counts,
20032 update_joiner_offpath_counts): Use profile_probability.
20034 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20037 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20039 * config/arm/arm-c.c (arm_cpu_builtins): Define
20040 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20042 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20044 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20045 array entries to represent __ieee128 versions of the
20046 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20047 scalar_extract_sig, and scalar_insert_exp built-in functions.
20048 (altivec_resolve_overloaded_builtin): Add special case handling
20049 for the __builtin_scalar_insert_exp function, as represented by
20050 the P9V_BUILTIN_VEC_VSIEDP constant.
20051 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20052 exponent support for __ieee128 argument.
20053 (VSESQP): Add scalar extract signature support for __ieee128
20055 (VSTDCNQP): Add scalar test negative support for __ieee128
20057 (VSIEQP): Add scalar insert exponent support for __int128 argument
20058 with __ieee128 result.
20059 (VSIEQPF): Add scalar insert exponent support for __ieee128
20060 argument with __ieee128 result.
20061 (VSTDCQP): Add scalar test data class support for __ieee128
20063 (VSTDCNQP): Add overload support for scalar test negative with
20064 __ieee128 argument.
20065 (VSTDCQP): Add overload support for scalar test data class
20066 __ieee128 argument.
20067 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
20068 UNSPEC_VSX_SXSIGDP.
20069 (UNSPEC_VSX_SIEXPQP): New constant.
20070 (xsxexpqp): New insn for VSX scalar extract exponent quad
20072 (xsxsigqp): New insn for VSX scalar extract significand quad
20074 (xsiexpqpf): New insn for VSX scalar insert exponent quad
20075 precision with floating point argument.
20076 (xststdcqp): New expand for VSX scalar test data class quad
20078 (xststdcnegqp): New expand for VSX scalar test negative quad
20080 (xststdcqp): New insn to match expansions for VSX scalar test data
20081 class quad precision and VSX scalar test negative quad precision.
20082 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
20083 special case operand checking to enforce that second operand of
20084 VSX scalar test data class with quad precision argument is a 7-bit
20086 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20087 prototypes and descriptions of __ieee128 versions of
20088 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
20089 scalar_test_data_class, and scalar_test_neg built-in functions.
20091 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20093 PR tree-optimization/81162
20094 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
20095 replace a negate with an add.
20097 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
20099 * doc/invoke.texi (arm/-mcpu): Document +crypto.
20101 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20103 * config/arm/arm-c.c (arm_cpu_builtins): Define
20104 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20106 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20108 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
20109 (armv8-r): Set ARM Cortex-R52 as default CPU.
20110 * config/arm/arm-tables.opt: Regenerate.
20111 * config/arm/arm-tune.md: Regenerate.
20112 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20114 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
20115 extension for -mcpu=cortex-r52.
20117 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20119 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
20120 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
20121 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
20122 (fp-armv8): Define it as FP_ARMv8 only.
20123 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
20124 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
20126 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
20127 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
20128 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
20129 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
20130 than TARGET_FPU_ARMV8.
20131 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
20132 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
20133 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
20135 * config/arm/neon.md (neon_vrint): Likewise.
20136 (neon_vcvt): Likewise.
20137 (neon_<fmaxmin_op><mode>): Likewise.
20138 (<fmaxmin><mode>3): Likewise.
20139 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
20140 * config/arm/predicates.md (arm_cond_move_operator): Check against
20141 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
20143 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
20145 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
20146 to top of function.
20148 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20150 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
20151 loop in comment with memset.
20153 2017-07-14 Martin Liska <mliska@suse.cz>
20155 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
20156 * dwarf2out.c (is_java): Remove the function.
20157 (output_pubname): Remove usage of the function.
20158 (lower_bound_default): Remove usage of DW_LANG_Java.
20159 (gen_compile_unit_die): Likewise.
20160 * gcc.c: Remove compiler defaults for .java and .zip files.
20161 * gimple-expr.c (remove_suffix): Change as there's no longer
20162 extension than 4-letter one.
20163 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
20164 (gimplify_save_expr): Likewise.
20165 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
20166 as it's possible even for other languages than Java.
20167 * langhooks.h (struct lang_hooks): Remove Java from a comment.
20168 * lto-opts.c (lto_write_options): Remove reference to Java.
20169 * opts.c (strip_off_ending): Update file extension handling.
20170 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
20171 * tree-eh.c (lower_resx): Likewise.
20172 * tree.c (free_lang_data_in_type): Remove dead code.
20173 (find_decls_types_r): Likewise.
20174 (build_common_builtin_nodes): Remove Java from a comment.
20175 (verify_type): Remove dead code.
20176 * varasm.c (assemble_external): Remove Java from a comment.
20178 2017-07-14 Martin Liska <mliska@suse.cz>
20180 * opts.c (finish_options): Add quotes.
20181 (common_handle_option): Likewise.
20183 2017-07-14 Martin Liska <mliska@suse.cz>
20185 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
20186 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
20187 Remove N_SO_PASCAL.
20188 * dwarf2out.c (lower_bound_default): Do not handle
20190 (gen_compile_unit_die): Likewise.
20191 * gcc.c: Remove default extension binding for GNU Pascal.
20192 * stmt.c: Remove Pascal language from a comment.
20193 * xcoffout.c: Likewise.
20195 2017-07-13 David Malcolm <dmalcolm@redhat.com>
20198 * diagnostic-show-locus.c (fixit_cmp): New function.
20199 (layout::layout): Sort m_fixit_hints.
20200 (column_range::column_range): Assert that the values are valid.
20201 (struct char_span): New struct.
20202 (correction::overwrite): New method.
20203 (struct source_line): New struct.
20204 (line_corrections::add_hint): Add assertions. Reimplement memcpy
20205 calls in terms of classes source_line and char_span, and
20206 correction::overwrite.
20207 (selftest::test_overlapped_fixit_printing_2): New function.
20208 (selftest::diagnostic_show_locus_c_tests): Call it.
20210 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
20212 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
20213 early if there is no lhs.
20215 2017-07-13 Martin Liska <mliska@suse.cz>
20217 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
20218 (gen_reference_type_die): Likewise.
20219 * stor-layout.c: Remove Pascal-related comment.
20221 2017-07-13 Martin Liska <mliska@suse.cz>
20223 * opts.c (finish_options): Add quotes to error messages.
20224 (parse_sanitizer_options): Likewise.
20226 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
20228 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
20230 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
20232 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
20234 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
20236 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
20238 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
20240 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
20243 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
20244 provides the hardware capability bits, define the macro
20245 __BUILTIN_CPU_SUPPORTS__.
20246 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
20247 if GLIBC does not provide the hardware capability bits. Add a
20248 gcc_unreachable call if the built-in cpu function is neither
20249 __builtin_cpu_is nor __builtin_cpu_supports.
20250 (rs6000_get_function_versions_dispatcher): Change the warning
20251 that an old GLIBC is used which does not export the capability
20252 bits to be an error.
20253 * doc/extend.texi (target_clones attribute): Document the
20254 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
20255 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
20256 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
20257 the macros defined by GCC if the newer GLIBC is available.
20259 2017-07-12 Jeff Law <law@redhat.com>
20261 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
20262 remaining includes slightly.
20263 * config/riscv/riscv-builtins.c: Include profile-count.h.
20265 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20268 * config/avr/avr.c (avr_set_current_function): In diagnostic
20269 messages: Quote keywords and (parts of) identifiers.
20270 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
20273 2017-07-12 Carl Love <cel@us.ibm.com>
20275 * config/rs6000/rs6000-c.c: Add support for built-in functions
20276 vector bool char vec_revb (vector bool char);
20277 vector bool short vec_revb (vector short char);
20278 vector bool int vec_revb (vector bool int);
20279 vector bool long long vec_revb (vector bool long long);
20280 * doc/extend.texi: Update the built-in documentation file for the
20281 new built-in functions.
20283 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20285 * config/s390/s390.md: Remove movcc splitter.
20287 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20289 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
20290 load/store on condition.
20292 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20295 * config/avr/avr.c (avr_encode_section_info)
20296 [progmem && !TREE_READONLY]: Error if progmem object needs
20299 2017-07-11 Michael Collison <michael.collison@arm.com>
20301 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
20304 2017-07-11 Carl Love <cel@us.ibm.com>
20306 * config/rs6000/rs6000-c.c: Add support for builtins
20307 vector unsigned int vec_parity_lsbb (vector signed int);
20308 vector unsigned int vec_parity_lsbb (vector unsigned int);
20309 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
20310 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
20311 vector unsigned long long vec_parity_lsbb (vector signed long long);
20312 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
20313 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
20314 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
20315 * doc/extend.texi: Update the built-in documentation file for the
20316 new built-in functions.
20318 2017-07-11 David Malcolm <dmalcolm@redhat.com>
20320 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
20321 (layout::m_primary_loc): New field.
20322 (layout::layout): Initialize new field. Move location filtering
20323 logic from here to...
20324 (layout::maybe_add_location_range): ...this new method. Add
20325 support for filtering to just the lines already specified by other
20327 (layout::will_show_line_p): New method.
20328 (gcc_rich_location::add_location_if_nearby): New method.
20329 (selftest::test_add_location_if_nearby): New test function.
20330 (selftest::diagnostic_show_locus_c_tests): Call it.
20331 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
20334 2017-07-11 Tom de Vries <tom@codesourcery.com>
20336 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
20337 (bb_first_real_insn): New function.
20338 (nvptx_single): Add extra initialization of broadcasted condition
20341 2017-07-11 Nathan Sidwell <nathan@acm.org>
20343 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
20345 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
20347 * doc/extend.texi (AVR Function Attributes): Remove weblink to
20348 Binutils doc as TEXI will mess them up.
20349 * doc/invoke.texi (AVR Options): Same here.
20351 2017-07-11 Daniel Cederman <cederman@gaisler.com>
20353 * config/sparc/sparc.opt (mfix-ut700): New option.
20354 (mfix-gr712rc): Likewise.
20355 (sparc_fix_b2bst): New variable.
20356 * doc/invoke.texi (SPARC options): Document them.
20357 (ARM options): Fix warnings.
20358 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
20359 instructions to prevent sequences that can trigger the store-store
20360 errata for certain LEON3FT processors.
20361 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
20362 (sparc_option_override): Set sparc_fix_b2bst appropriately.
20363 * config/sparc/sparc.md (fix_b2bst): New attribute.
20364 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
20366 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
20369 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
20371 (*rsqrtsf2_sse): Ditto.
20373 (div<mode>3): Macroize insn from divdf3 and divsf3
20374 using MODEF mode iterator.
20376 2017-07-10 Martin Sebor <msebor@redhat.com>
20378 PR tree-optimization/80397
20379 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
20380 instead of testing for equality to INTEGER_TYPE.
20382 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
20384 * config.gcc: Remove uclibc from arc target spec.
20386 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
20388 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
20390 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20393 * lto-wrapper.c (remove_option): New function.
20394 (merge_and_complain): Merge PIC/PIE options more realistically.
20396 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20398 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
20402 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
20403 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
20404 * config.in: Regenerate.
20405 * configure: Regenerate.
20406 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
20407 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
20408 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
20409 (TARGET_GASISR_PROLOGUES): ...target mask.
20410 * common/config/avr/avr-common.c
20411 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
20412 Set -mgas-isr-prologues.
20413 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
20414 INSERT_PASS_BEFORE for it.
20415 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
20416 * config/avr/avr.c (avr_option_override)
20417 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
20418 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
20419 (avr_attribute_table) <no_gccisr>: Add new function attribute.
20420 (avr_set_current_function) <is_no_gccisr>: Init machine field.
20421 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
20423 (make_avr_pass_pre_proep): New function.
20424 (emit_push_sfr) <treg>: Add argument to function and use it
20425 instead of TMP_REG.
20426 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
20427 and set machine->gasisr.yes.
20428 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
20429 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
20430 __gcc_isr.n_pushed to .L__stack_usage.
20431 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
20432 (avr_asm_final_postscan_insn): ...this new static function.
20433 * config/avr/avr.h (machine_function)
20434 <is_no_gccisr, use_L__stack_usage>: New fields.
20435 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
20436 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
20437 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
20438 (gasisr, *gasisr): New expander and insn.
20439 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20440 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
20441 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
20443 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
20445 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
20448 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20450 Move jump-tables out of .text again.
20453 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
20454 (ASM_OUTPUT_ADDR_VEC): New function.
20455 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
20456 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
20457 INSN_ADDRESSes as asm comment.
20458 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
20459 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
20460 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
20461 * config/avr/avr.md (*tablejump): Adjust comment.
20462 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
20463 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
20465 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
20466 (avr_output_addr_vec): New proto.
20467 (avr_log_t) <insn_addresses>: New field.
20469 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
20472 * config/i386/i386.c (ix86_function_arg_advance): Set
20473 outgoing_args_on_stack to true if there are outgoing arguments
20475 (ix86_function_arg): Likewise.
20476 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
20477 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
20478 * config/i386/i386.h (machine_function): Add
20479 outgoing_args_on_stack.
20481 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
20483 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20484 supporting pthreds.
20485 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
20487 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
20489 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
20490 (REAL_H): Remove $(MACHMODE_H).
20491 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
20493 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
20494 $(MACHMODE_H) and double-int.h.
20495 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
20497 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
20500 2017-07-07 Andrew Pinski <apinski@cavium.com>
20502 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
20503 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
20505 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20507 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
20508 Add warning if GCC was not configured to link against a GLIBC that
20509 exports the hardware capability bits.
20510 (make_resolver_func): Make resolver function private and not a
20511 COMDAT function. Create the name with clone_function_name instead
20512 of make_unique_name.
20515 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
20516 correct operand in doing the split.
20518 2017-07-07 Carl Love <cel@us.ibm.com>
20520 * config/rs6000/rs6000-c: Add support for built-in function
20521 vector unsigned short vec_pack_to_short_fp32 (vector float,
20523 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
20524 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
20525 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
20526 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
20527 (convert_4f32_8i16): Add define_expand.
20528 * doc/extend.texi: Update the built-in documentation file for the
20529 new built-in function.
20531 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20533 * config/sparc/m8.md: New file.
20534 * config/sparc/sparc.md: Include m8.md.
20536 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20538 * config/sparc/sparc.opt: New option -mvis4b.
20539 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
20540 (sparc_option_override): Handle VIS4B.
20541 (enum sparc_builtins): Define
20542 SPARC_BUILTIN_DICTUNPACK{8,16,32},
20543 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
20544 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
20545 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
20546 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
20547 (check_constant_argument): New function.
20548 (sparc_vis_init_builtins): Define builtins
20549 __builtin_vis_dictunpack{8,16,32},
20550 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
20551 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
20552 __builtin_vis_fpcmpde{8,16,32}shl and
20553 __builtin_vis_fpcmpur{8,16,32}shl.
20554 (sparc_expand_builtin): Check that the constant operands to
20555 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
20556 constant and in range.
20557 * config/sparc/sparc-c.c (sparc_target_macros): Handle
20559 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
20560 (SPARC_IMM5_P): Likewise.
20561 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
20562 (enabled): Handle vis4b.
20563 (UNSPEC_DICTUNPACK): New unspec.
20564 (UNSPEC_FPCMPSHL): Likewise.
20565 (UNSPEC_FPUCMPSHL): Likewise.
20566 (UNSPEC_FPCMPDESHL): Likewise.
20567 (UNSPEC_FPCMPURSHL): Likewise.
20568 (cpu_feature): New CPU feature `vis4b'.
20569 (dictunpack{8,16,32}): New insns.
20570 (FPCSMODE): New mode iterator.
20571 (fpcscond): New code iterator.
20572 (fpcsucond): Likewise.
20573 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
20574 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
20575 (fpcmpde{8,16,32}{si,di}shl): Likewise.
20576 (fpcmpur{8,16,32}{si,di}shl): Likewise.
20577 * config/sparc/constraints.md: Define constraints `q' for unsigned
20578 2-bit integer constants and `t' for unsigned 5-bit integer
20580 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
20582 (imm5_operand_dictunpack16): Likewise.
20583 (imm5_operand_dictunpack32): Likewise.
20584 (imm2_operand): Likewise.
20585 * doc/invoke.texi (SPARC Options): Document -mvis4b.
20586 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
20587 ditunpack* and fpcmp*shl builtins.
20589 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20591 * config.gcc: Handle m8 in --with-{cpu,tune} options.
20592 * config.in: Add HAVE_AS_SPARC6 define.
20593 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
20595 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
20597 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
20598 (CPP_CPU_SPEC): Handle m8.
20599 (ASM_CPU_SPEC): Likewise.
20600 * config/sparc/sparc-opts.h (enum processor_type): Add
20602 * config/sparc/sparc.c (m8_costs): New struct.
20603 (sparc_option_override): Handle TARGET_CPU_m8.
20604 (sparc32_initialize_trampoline): Likewise.
20605 (sparc64_initialize_trampoline): Likewise.
20606 (sparc_issue_rate): Likewise.
20607 (sparc_register_move_cost): Likewise.
20608 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
20609 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
20610 (ASM_CPU64_DEFAULT_SPEC): Likewise.
20611 (CPP_CPU_SPEC): Handle M8.
20612 (ASM_CPU_SPEC): Likewise.
20613 (AS_M8_FLAG): Define.
20614 * config/sparc/sparc.md: Add m8 to the cpu attribute.
20615 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
20616 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
20618 * configure: Regenerate.
20619 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
20622 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20624 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
20626 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
20627 ("*movdi_insn_sp32"): Do not set v3pipe.
20628 ("*movsi_insn"): Likewise.
20629 ("*movdi_insn_sp64"): Likewise.
20630 ("*movsf_insn"): Likewise.
20631 ("*movdf_insn_sp32"): Likewise.
20632 ("*movdf_insn_sp64"): Likewise.
20633 ("*zero_extendsidi2_insn_sp64"): Likewise.
20634 ("*sign_extendsidi2_insn"): Likewise.
20635 ("*mov<VM32:mode>_insn"): Likewise.
20636 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20637 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20638 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20639 ("<vlop:code><VL:mode>3"): Likewise.
20640 ("*not_<vlop:code><VL:mode>3"): Likewise.
20641 ("*nand<VL:mode>_vis"): Likewise.
20642 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
20643 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
20644 ("one_cmpl<VL:mode>2"): Likewise.
20645 ("faligndata<VM64:mode>_vis"): Likewise.
20646 ("alignaddrsi_vis"): Likewise.
20647 ("alignaddrdi_vis"): Likweise.
20648 ("alignaddrlsi_vis"): Likewise.
20649 ("alignaddrldi_vis"): Likewise.
20650 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20651 ("bmaskdi_vis"): Likewise.
20652 ("bmasksi_vis"): Likewise.
20653 ("bshuffle<VM64:mode>_vis"): Likewise.
20654 ("cmask8<P:mode>_vis"): Likewise.
20655 ("cmask16<P:mode>_vis"): Likewise.
20656 ("cmask32<P:mode>_vis"): Likewise.
20657 ("pdistn<P:mode>_vis"): Likewise.
20658 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20660 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20662 * config/sparc/sparc.md ("subtype"): New insn attribute.
20663 ("*wrgsr_sp64"): Set insn subtype.
20664 ("*rdgsr_sp64"): Likewise.
20665 ("alignaddrsi_vis"): Likewise.
20666 ("alignaddrdi_vis"): Likewise.
20667 ("alignaddrlsi_vis"): Likewise.
20668 ("alignaddrldi_vis"): Likewise.
20669 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20670 ("fexpand_vis"): Likewise.
20671 ("fpmerge_vis"): Likewise.
20672 ("faligndata<VM64:mode>_vis"): Likewise.
20673 ("bshuffle<VM64:mode>_vis"): Likewise.
20674 ("cmask8<P:mode>_vis"): Likewise.
20675 ("cmask16<P:mode>_vis"): Likewise.
20676 ("cmask32<P:mode>_vis"): Likewise.
20677 ("fchksm16_vis"): Likewise.
20678 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
20679 ("fmean16_vis"): Likewise.
20680 ("fp<plusminus_insn>64_vis"): Likewise.
20681 ("<plusminus_insn>v8qi3"): Likewise.
20682 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20683 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
20684 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
20685 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
20686 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
20687 ("*movqi_insn"): Likewise.
20688 ("*movhi_insn"): Likewise.
20689 ("*movsi_insn"): Likewise.
20690 ("movsi_pic_gotdata_op"): Likewise.
20691 ("*movdi_insn_sp32"): Likewise.
20692 ("*movdi_insn_sp64"): Likewise.
20693 ("movdi_pic_gotdata_op"): Likewise.
20694 ("*movsf_insn"): Likewise.
20695 ("*movdf_insn_sp32"): Likewise.
20696 ("*movdf_insn_sp64"): Likewise.
20697 ("*zero_extendhisi2_insn"): Likewise.
20698 ("*zero_extendqihi2_insn"): Likewise.
20699 ("*zero_extendqisi2_insn"): Likewise.
20700 ("*zero_extendqidi2_insn"): Likewise.
20701 ("*zero_extendhidi2_insn"): Likewise.
20702 ("*zero_extendsidi2_insn_sp64"): Likewise.
20703 ("ldfsr"): Likewise.
20704 ("prefetch_64"): Likewise.
20705 ("prefetch_32"): Likewise.
20706 ("tie_ld32"): Likewise.
20707 ("tie_ld64"): Likewise.
20708 ("*tldo_ldub_sp32"): Likewise.
20709 ("*tldo_ldub1_sp32"): Likewise.
20710 ("*tldo_ldub2_sp32"): Likewise.
20711 ("*tldo_ldub_sp64"): Likewise.
20712 ("*tldo_ldub1_sp64"): Likewise.
20713 ("*tldo_ldub2_sp64"): Likewise.
20714 ("*tldo_ldub3_sp64"): Likewise.
20715 ("*tldo_lduh_sp32"): Likewise.
20716 ("*tldo_lduh1_sp32"): Likewise.
20717 ("*tldo_lduh_sp64"): Likewise.
20718 ("*tldo_lduh1_sp64"): Likewise.
20719 ("*tldo_lduh2_sp64"): Likewise.
20720 ("*tldo_lduw_sp32"): Likewise.
20721 ("*tldo_lduw_sp64"): Likewise.
20722 ("*tldo_lduw1_sp64"): Likewise.
20723 ("*tldo_ldx_sp64"): Likewise.
20724 ("*mov<VM32:mode>_insn"): Likewise.
20725 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20726 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20728 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20730 * config/sparc/sparc.md ("type"): New insn type viscmp.
20731 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
20733 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
20734 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
20735 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20736 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
20738 ("n7_vis_logical_11cycle"): Likewise.
20739 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
20740 * config/sparc/niagara2.md ("niag3_vis": Likewise.
20741 * config/sparc/niagara.md ("niag_vis"): Likewise.
20742 * config/sparc/ultra3.md ("us3_fga"): Likewise.
20743 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
20745 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20747 * config/sparc/sparc.md: New instruction type `bmask'.
20748 (bmaskdi_vis): Use the `bmask' type.
20749 (bmasksi_vis): Likewise.
20750 * config/sparc/ultra3.md (us3_array): Likewise.
20751 * config/sparc/niagara7.md (n7_array): Likewise.
20752 * config/sparc/niagara4.md (n4_array): Likewise.
20753 * config/sparc/niagara2.md (niag2_vis): Likewise.
20754 (niag3_vis): Likewise.
20755 * config/sparc/niagara.md (niag_vis): Likewise.
20757 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20759 * ipa-comdats.c: Remove optimize check from gate.
20760 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
20761 for functions not optimized.
20762 (ipa_fn_summary_read): Skip optimize check.
20763 (ipa_fn_summary_write): Likewise.
20764 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
20766 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
20768 (can_inline_edge_p): Check flag_pcc_struct_return for match.
20769 (check_callers): Give up on caller which is not optimized.
20770 (inline_small_functions): Likewise.
20771 (ipa_inline): Do not give up when not optimizing.
20772 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
20773 away unoptimizes cdtors.
20774 (whole_program_function_and_variable_visibility): Do
20775 ipa_discover_readonly_nonaddressable_vars in LTO mode.
20776 * ipa.c (process_references): Do not check optimize.
20777 (symbol_table::remove_unreachable_nodes): Update optimize check.
20778 (set_writeonly_bit): Update optimize check.
20779 (pass_ipa_cdtor_merge::gate): Do not check optimize.
20780 (pass_ipa_single_use::gate): Remove.
20782 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20784 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
20785 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
20786 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
20787 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
20788 permute_load, permute_store, adjust_extract, adjust_splat,
20789 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
20790 replace_swap_with_copy, dump_swap_insn_table,
20791 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
20792 recombine_lvx_pattern, recombine_stvx_pattern,
20793 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
20794 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
20795 to file rs6000-p8swap.c.
20796 * config/rs6000/rs6000-p8swap.c: New file.
20797 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
20798 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
20799 and rs6000*-*-* targets.
20801 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20803 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
20805 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20807 * lto-wrapper.c (merge_and_complain): Do not merge
20808 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
20809 fsigned_zeros, ftrapping_math, fwrapv.
20810 (append_compiler_options): Do not track these options.
20811 (append_linker_options): Likewie
20813 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20815 * cgraphunit.c (cgraph_node::finalize_function): When
20816 !flag_toplevel_reorde set no_reorder flag.
20817 (varpool_node::finalize_decl): Likewise.
20818 (symbol_table::compile): Drop no toplevel reorder path.
20820 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20822 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
20823 edges; zero probability is not better than uninitialized.
20825 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20827 * asan.h (asan_sanitize_allocas_p): Declare.
20828 * asan.c (asan_sanitize_allocas_p): New function.
20829 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
20830 (handle_builtin_alloca): Likewise.
20831 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
20832 if !asan_sanitize_allocas_p.
20833 * params.def (asan-instrument-allocas): Add new option.
20834 * params.h (ASAN_PROTECT_ALLOCAS): Define.
20835 * opts.c (common_handle_option): Disable allocas sanitization for
20838 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20840 * asan.c: Include gimple-fold.h.
20841 (get_last_alloca_addr): New function.
20842 (handle_builtin_stackrestore): Likewise.
20843 (handle_builtin_alloca): Likewise.
20844 (asan_emit_allocas_unpoison): Likewise.
20845 (get_mem_refs_of_builtin_call): Add new parameter, remove const
20846 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
20847 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
20848 (instrument_builtin_call): Pass gimple iterator to
20849 get_mem_refs_of_builtin_call.
20850 (last_alloca_addr): New global.
20851 * asan.h (asan_emit_allocas_unpoison): Declare.
20852 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
20853 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
20854 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
20855 if function calls alloca.
20856 * gimple-fold.c (replace_call_with_value): Remove static keyword.
20857 * gimple-fold.h (replace_call_with_value): Declare.
20858 * internal-fn.c: Include asan.h.
20859 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
20860 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
20862 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20864 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
20865 (C_SELFTEST_FLAGS): New.
20866 (CPP_SELFTEST_FLAGS): New.
20867 (SELFTEST_DEPS): New, from deps of s-selftest.
20868 (C_SELFTEST_DEPS): New, from deps of s-selftest.
20869 (CPP_SELFTEST_DEPS): New.
20870 (selftest): Add dependency on s-selftest-c++.
20871 (s-selftest): Rename to...
20872 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
20873 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
20874 than SELFTEST_FLAGS.
20875 (selftest-gdb): Rename to...
20876 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
20878 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
20879 (selftest-valgrind): Rename to...
20880 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
20882 (selftest-valgrind): Reintroduce as an alias for
20883 selftest-c-valgrind.
20884 (s-selftest-c++): New.
20885 (selftest-c++-gdb): New.
20886 (selftest-c++-valgrind): New.
20888 2017-07-06 Olivier Hainque <hainque@adacore.com>
20890 * gcc.c (process_command): When deciding if undefined variables
20891 should be ignored when processing specs, accept "gcc -v" as well.
20893 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20895 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
20896 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
20898 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20900 * config/arm/arm-cpus.in (armv8-r): Add new entry.
20901 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
20902 * config/arm/arm-tables.opt: Regenerate.
20903 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
20905 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
20907 2017-07-06 Carl Love <cel@us.ibm.com>
20909 * ChangeLog: Clean up from mid air collision
20911 2017-07-06 Carl Love <cel@us.ibm.com>
20913 * config/rs6000/rs6000-c.c: Add support for built-in functions
20914 vector signed int vec_subc (vector signed int, vector signed int);
20915 vector signed __int128 vec_subc (vector signed __int128,
20916 vector signed __int128);
20917 vector unsigned __int128 vec_subc (vector unsigned __int128,
20918 vector unsigned __int128);
20919 vector signed int vec_sube (vector signed int, vector signed int,
20920 vector signed int);
20921 vector unsigned int vec_sube (vector unsigned int,
20922 vector unsigned int,
20923 vector unsigned int);
20924 vector signed __int128 vec_sube (vector signed __int128,
20925 vector signed __int128,
20926 vector signed__int128);
20927 vector unsigned __int128 vec_sube (vector unsigned __int128,
20928 vector unsigned __int128,
20929 vector unsigned __int128);
20930 vector signed int vec_subec (vector signed int, vector signed int,
20931 vector signed int);
20932 vector unsigned int vec_subec (vector unsigned int,
20933 vector unsigned int,
20934 vector unsigned int);
20935 vector signed __int128 vec_subec (vector signed __int128,
20936 vector signed __int128,
20937 vector signed__int128);
20938 vector unsigned __int128 vec_subec (vector unsigned __int128,
20939 vector unsigned __int128,
20940 vector unsigned __int128);
20941 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
20942 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
20943 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
20944 BU_ALTIVEC_OVERLOAD_X definitions.
20945 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
20946 * doc/extend.texi: Update the built-in documentation file for the new
20947 built-in functions.
20949 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20952 * diagnostic-show-locus.c (layout::layout): Use start and finish
20953 spelling location for the start and finish of each range.
20954 * genmatch.c (linemap_client_expand_location_to_spelling_point):
20955 Add unused aspect param.
20956 * input.c (expand_location_1): Add "aspect" param, and use it
20957 to access the correct part of the location.
20958 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
20960 (expand_location_to_spelling_point): Likewise.
20961 (linemap_client_expand_location_to_spelling_point): Add "aspect"
20962 param, and pass it to expand_location_1.
20964 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
20966 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
20967 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
20968 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
20969 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
20970 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
20971 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
20972 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
20973 _mm_maskz_getmant_ss): New intrinsics.
20974 (__builtin_ia32_getexpss128_mask): Changed to ...
20975 __builtin_ia32_getexpss128_round ... this.
20976 (__builtin_ia32_getexpsd128_mask): Changed to ...
20977 __builtin_ia32_getexpsd128_round ... this.
20978 * config/i386/i386-builtin-types.def
20979 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
20980 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
20981 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
20982 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
20983 __builtin_ia32_getmantss_mask_round): New builtins.
20984 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
20985 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
20986 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
20987 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
20988 * config/i386/sse.md
20989 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
20990 avx512f_sgetexp<mode><mask_scalar_name>
20991 <round_saeonly_scalar_name> ... this.
20992 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
20993 %0, %1, %2<round_saeonly_op3>}): Changed to ...
20994 vgetexp<ssescalarmodesuffix>
20995 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
20996 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
20997 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
20998 avx512f_vgetmant<mode><mask_scalar_name>
20999 <round_saeonly_scalar_name> ... this.
21000 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21001 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21002 vgetmant<ssescalarmodesuffix>
21003 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21004 %0<mask_scalar_operand4>, %1, %2
21005 <round_saeonly_scalar_mask_op4>, %3} ... this.
21006 * config/i386/subst.md (mask_scalar_operand4,
21007 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21008 round_saeonly_scalar_nimm_predicate): New subst attributes.
21010 2017-07-06 Julia Koval <julia.koval@intel.com>
21012 * config/i386/i386.c (ix86_erase_embedded_rounding):
21013 Remove code for old rounding pattern.
21015 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21017 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21019 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21021 * doc/sourcebuild.texi (Test Directives, Variants of
21022 dg-require-support): Add documentation for dg-require-stack-check.
21024 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21026 * config/i386/subst.md (mask_scalar, round_scalar,
21027 round_saeonly_scalar): New meta-templates.
21028 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21029 round_scalar_mask_operand3, round_scalar_mask_op3,
21030 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21031 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21032 round_saeonly_scalar_constraint,
21033 round_saeonly_scalar_prefix): New subst attribute.
21034 * config/i386/sse.md
21035 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21036 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21037 <round_scalar_name> ... this.
21038 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21039 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21040 <round_scalar_name> ... this.
21041 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21042 <sse>_vm<code><mode>3<mask_scalar_name>
21043 <round_saeonly_scalar_name> ... this.
21044 (v<plusminus_mnemonic><ssescalarmodesuffix>
21045 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21046 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21047 v<plusminus_mnemonic><ssescalarmodesuffix>
21048 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21049 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21050 (v<multdiv_mnemonic><ssescalarmodesuffix>
21051 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21052 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21053 v<multdiv_mnemonic><ssescalarmodesuffix>
21054 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21055 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21056 (v<maxmin_float><ssescalarmodesuffix>
21057 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21058 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
21059 v<maxmin_float><ssescalarmodesuffix>
21060 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21061 %0<mask_scalar_operand3>, %1, %<iptr>2
21062 <round_saeonly_scalar_mask_op3>} ... this.
21064 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
21066 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
21067 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
21069 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
21070 Alan Hayward <alan.hayward@arm.com>
21071 David Sherwood <david.sherwood@arm.com>
21073 * combine.c (simplify_if_then_else): Remove "enum" before
21075 * compare-elim.c (can_eliminate_compare): Likewise.
21076 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
21078 (aarch64_lookup_simd_builtin_type): Likewise.
21079 (aarch64_simd_builtin_type): Likewise.
21080 (aarch64_init_simd_builtin_types): Likewise.
21081 (aarch64_simd_expand_args): Likewise.
21082 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
21084 (aarch64_reverse_mask): Likewise.
21085 (aarch64_simd_emit_reg_reg_move): Likewise.
21086 (aarch64_gen_adjusted_ldpstp): Likewise.
21087 (aarch64_ccmp_mode_to_code): Likewise.
21088 (aarch64_operands_ok_for_ldpstp): Likewise.
21089 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21090 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
21092 (aarch64_min_divisions_for_recip_mul): Likewise.
21093 (aarch64_reassociation_width): Likewise.
21094 (aarch64_get_condition_code_1): Likewise.
21095 (aarch64_simd_emit_reg_reg_move): Likewise.
21096 (aarch64_simd_attr_length_rglist): Likewise.
21097 (aarch64_reverse_mask): Likewise.
21098 (aarch64_operands_ok_for_ldpstp): Likewise.
21099 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21100 (aarch64_gen_adjusted_ldpstp): Likewise.
21101 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
21103 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
21104 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
21105 (arm_lookup_simd_builtin_type): Likewise.
21106 (arm_simd_builtin_type): Likewise.
21107 (arm_init_simd_builtin_types): Likewise.
21108 (arm_expand_builtin_args): Likewise.
21109 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
21110 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
21111 (ft32_setup_incoming_varargs): Likewise.
21112 (ft32_function_arg): Likewise.
21113 (ft32_function_arg_advance): Likewise.
21114 (ft32_pass_by_reference): Likewise.
21115 (ft32_arg_partial_bytes): Likewise.
21116 (ft32_valid_pointer_mode): Likewise.
21117 (ft32_addr_space_pointer_mode): Likewise.
21118 (ft32_addr_space_legitimate_address_p): Likewise.
21119 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
21121 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
21122 (ix86_emit_outlined_ms2sysv_restore): Likewise.
21123 (iamcu_alignment): Likewise.
21124 (canonicalize_vector_int_perm): Likewise.
21125 (ix86_noce_conversion_profitable_p): Likewise.
21126 (ix86_mpx_bound_mode): Likewise.
21127 (ix86_operands_ok_for_move_multiple): Likewise.
21128 * config/microblaze/microblaze-protos.h
21129 (microblaze_expand_conditional_branch_reg): Likewise.
21130 * config/microblaze/microblaze.c
21131 (microblaze_expand_conditional_branch_reg): Likewise.
21132 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
21134 (rs6000_reassociation_width): Likewise.
21135 (rs6000_invalid_binary_op): Likewise.
21136 (fusion_p9_p): Likewise.
21137 (emit_fusion_p9_load): Likewise.
21138 (emit_fusion_p9_store): Likewise.
21139 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
21141 (riscv_hard_regno_mode_ok_p): Likewise.
21142 (riscv_address_insns): Likewise.
21143 (riscv_split_symbol): Likewise.
21144 (riscv_legitimize_move): Likewise.
21145 (riscv_function_value): Likewise.
21146 (riscv_hard_regno_nregs): Likewise.
21147 (riscv_expand_builtin): Likewise.
21148 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
21149 (riscv_build_integer): Likewise.
21150 (riscv_split_integer): Likewise.
21151 (riscv_legitimate_constant_p): Likewise.
21152 (riscv_cannot_force_const_mem): Likewise.
21153 (riscv_regno_mode_ok_for_base_p): Likewise.
21154 (riscv_valid_base_register_p): Likewise.
21155 (riscv_valid_offset_p): Likewise.
21156 (riscv_valid_lo_sum_p): Likewise.
21157 (riscv_classify_address): Likewise.
21158 (riscv_legitimate_address_p): Likewise.
21159 (riscv_address_insns): Likewise.
21160 (riscv_load_store_insns): Likewise.
21161 (riscv_force_binary): Likewise.
21162 (riscv_split_symbol): Likewise.
21163 (riscv_force_address): Likewise.
21164 (riscv_legitimize_address): Likewise.
21165 (riscv_move_integer): Likewise.
21166 (riscv_legitimize_const_move): Likewise.
21167 (riscv_legitimize_move): Likewise.
21168 (riscv_address_cost): Likewise.
21169 (riscv_subword): Likewise.
21170 (riscv_output_move): Likewise.
21171 (riscv_canonicalize_int_order_test): Likewise.
21172 (riscv_emit_int_order_test): Likewise.
21173 (riscv_function_arg_boundary): Likewise.
21174 (riscv_pass_mode_in_fpr_p): Likewise.
21175 (riscv_pass_fpr_single): Likewise.
21176 (riscv_pass_fpr_pair): Likewise.
21177 (riscv_get_arg_info): Likewise.
21178 (riscv_function_arg): Likewise.
21179 (riscv_function_arg_advance): Likewise.
21180 (riscv_arg_partial_bytes): Likewise.
21181 (riscv_function_value): Likewise.
21182 (riscv_pass_by_reference): Likewise.
21183 (riscv_setup_incoming_varargs): Likewise.
21184 (riscv_print_operand): Likewise.
21185 (riscv_elf_select_rtx_section): Likewise.
21186 (riscv_save_restore_reg): Likewise.
21187 (riscv_for_each_saved_reg): Likewise.
21188 (riscv_register_move_cost): Likewise.
21189 (riscv_hard_regno_mode_ok_p): Likewise.
21190 (riscv_hard_regno_nregs): Likewise.
21191 (riscv_class_max_nregs): Likewise.
21192 (riscv_memory_move_cost): Likewise.
21193 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
21194 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
21195 (rl78_addr_space_address_mode): Likewise.
21196 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21198 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
21199 (rs6000_reassociation_width): Likewise.
21200 (rs6000_invalid_binary_op): Likewise.
21201 (fusion_p9_p): Likewise.
21202 (emit_fusion_p9_load): Likewise.
21203 (emit_fusion_p9_store): Likewise.
21204 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
21205 (ok_for_simple_move_operands): Likewise.
21206 (ok_for_simple_move_strict_operands): Likewise.
21207 (ok_for_simple_arith_logic_operands): Likewise.
21208 (visium_legitimize_reload_address): Likewise.
21209 (visium_select_cc_mode): Likewise.
21210 (output_cbranch): Likewise.
21211 (visium_split_double_move): Likewise.
21212 (visium_expand_copysign): Likewise.
21213 (visium_expand_int_cstore): Likewise.
21214 (visium_expand_fp_cstore): Likewise.
21215 * config/visium/visium.c (visium_pass_by_reference): Likewise.
21216 (visium_function_arg): Likewise.
21217 (visium_function_arg_advance): Likewise.
21218 (visium_libcall_value): Likewise.
21219 (visium_setup_incoming_varargs): Likewise.
21220 (visium_legitimate_constant_p): Likewise.
21221 (visium_legitimate_address_p): Likewise.
21222 (visium_legitimize_address): Likewise.
21223 (visium_secondary_reload): Likewise.
21224 (visium_register_move_cost): Likewise.
21225 (visium_memory_move_cost): Likewise.
21226 (prepare_move_operands): Likewise.
21227 (ok_for_simple_move_operands): Likewise.
21228 (ok_for_simple_move_strict_operands): Likewise.
21229 (ok_for_simple_arith_logic_operands): Likewise.
21230 (visium_function_value_1): Likewise.
21231 (rtx_ok_for_offset_p): Likewise.
21232 (visium_legitimize_reload_address): Likewise.
21233 (visium_split_double_move): Likewise.
21234 (visium_expand_copysign): Likewise.
21235 (visium_expand_int_cstore): Likewise.
21236 (visium_expand_fp_cstore): Likewise.
21237 (visium_split_cstore): Likewise.
21238 (visium_select_cc_mode): Likewise.
21239 (visium_split_cbranch): Likewise.
21240 (output_cbranch): Likewise.
21241 (visium_print_operand_address): Likewise.
21242 * expmed.c (flip_storage_order): Likewise.
21243 * expmed.h (emit_cstore): Likewise.
21244 (flip_storage_order): Likewise.
21245 * genrecog.c (validate_pattern): Likewise.
21246 * hsa-gen.c (gen_hsa_addr): Likewise.
21247 * internal-fn.c (expand_arith_overflow): Likewise.
21248 * ira-color.c (allocno_copy_cost_saving): Likewise.
21249 * lra-assigns.c (find_hard_regno_for_1): Likewise.
21250 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
21251 (process_invariant_for_inheritance): Likewise.
21252 * lra-eliminations.c (move_plus_up): Likewise.
21253 * omp-low.c (lower_oacc_reductions): Likewise.
21254 * simplify-rtx.c (simplify_subreg): Likewise.
21255 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
21256 (TARGET_CHKP_BOUND_MODE): Likewise..
21257 * targhooks.c (default_chkp_bound_mode): Likewise.
21258 (default_setup_incoming_vararg_bounds): Likewise.
21259 * targhooks.h (default_chkp_bound_mode): Likewise.
21260 (default_setup_incoming_vararg_bounds): Likewise.
21261 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
21262 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
21263 (have_whole_vector_shift): Likewise.
21264 * tree-vect-stmts.c (vectorizable_load): Likewise.
21265 * doc/tm.texi: Regenerate.
21267 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21269 Graceful degrade if Binutils PR21472 is not available.
21272 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
21273 .rodata in flash test fails.
21274 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
21275 * confgure: Regenerate.
21276 * config.in: Regenerate.
21277 * config/avr/avr.c (avr_asm_named_section)
21278 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
21279 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
21280 (avr_asm_init_sections): Same.
21282 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21284 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
21285 (fma<VH:mode>4_intrinsic): Likewise.
21286 (*fmsub<VCVTF:mode>4): Likewise.
21287 (*fmsub<VH:mode>4_intrinsic): Likewise.
21289 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21292 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
21293 Don't depend on "optimize > 0".
21294 (out_movhi_r_mr, out_movqi_mr_r): Same.
21295 (out_movhi_mr_r, out_movqi_r_mr): Same.
21296 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
21297 io_address_operand on "optimize > 0".
21299 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21301 * tree-loop-distribution.c: Add general explanantion on the pass.
21302 (generate_loops_for_partition): Mark distributed loop.
21303 (pg_add_dependence_edges): New parameter. Handle alias data
21304 dependence specially and record it in the parameter if asked.
21305 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
21306 (init_partition_graph_vertices, add_partition_graph_edge): New.
21307 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
21308 (free_partition_graph_vdata, build_partition_graph): New.
21309 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
21310 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
21311 (data_ref_segment_size, latch_dominated_by_data_ref): New.
21312 (compute_alias_check_pairs, version_loop_by_alias_check): New.
21313 (version_for_distribution_p, finalize_partitions): New.
21314 (distribute_loop): Handle alias data dependence specially. Factor
21315 out loop fusion code as functions and call these functions.
21317 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21319 * tree-loop-distribution.c (classify_partition): New parameter and
21320 better handle reduction statement.
21321 (rdg_build_partitions): Revise comment.
21322 (distribute_loop): Compute statements in all partitions and pass it
21323 to classify_partition.
21325 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21327 * tree-loop-distribution.c (enum partition_type): New.
21328 (struct partition): New field type.
21329 (partition_merge_into): Add parameter. Update partition type.
21330 (data_dep_in_cycle_p, update_type_for_merge): New functions.
21331 (build_rdg_partition_for_vertex): Compute partition type.
21332 (rdg_build_partitions): Dump partition type.
21333 (distribute_loop): Update calls to partition_merge_into.
21335 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21337 * tree-loop-distribution.c (struct ddr_hasher): New.
21338 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
21340 (classify_partition): Call get_data_dependence.
21341 (pg_add_dependence_edges): Ditto.
21342 (distribute_loop): Release data dependence hash table.
21344 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21346 * tree-loop-distribution.c (ref_base_address): Delete.
21347 (similar_memory_accesses): Rename ...
21348 (share_memory_accesses): ... to this. Check if partitions access
21349 the same memory reference.
21350 (distribute_loop): Call share_memory_accesses.
21352 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21354 * tree-loop-distribution.c (struct partition): New field recording
21355 its data reference.
21356 (partition_alloc, partition_free): Init and release data refs.
21357 (partition_merge_into): Merge data refs.
21358 (build_rdg_partition_for_vertex): Collect data refs for partition.
21359 (pg_add_dependence_edges): Change parameters from vector to bitmap.
21361 (distribute_loop): Remve data refs from vertice data of partition
21364 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21366 * tree-loop-distribution.c (params.h): Include header file.
21367 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
21368 (datarefs_vec): New global var.
21369 (create_rdg_vertices): Use datarefs_vec directly.
21370 (free_rdg): Don't free data references.
21371 (build_rdg): Update use. Don't free data references.
21372 (distribute_loop): Compute global variable for data references.
21373 Bail out if there are too many data references.
21375 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21377 * tree-loop-distribution.c (loop_nest): New global var.
21378 (build_rdg): Use loop directly, rather than loop nest.
21379 (pg_add_dependence_edges): Remove loop nest parameter. Use global
21381 (distribute_loop): Compute global variable loop nest. Update use.
21383 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21385 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
21386 (partition_merge_into): New parameter. Dump reason for fusion.
21387 (distribute_loop): Update use of partition_merge_into.
21389 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21391 * tree-loop-distribution.c (bb_top_order_index): New.
21392 (bb_top_order_index_size, bb_top_order_cmp): New.
21393 (stmts_from_loop): Use topological order.
21394 (pass_loop_distribution::execute): Compute and release topological
21395 order for basic blocks.
21397 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21399 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
21402 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21404 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
21405 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
21406 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
21407 * internal-fn.def (LOOP_DIST_ALIAS): New.
21408 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
21409 (fold_loop_internal_call): ... this.
21410 (vect_loop_dist_alias_call): New function.
21411 (set_uid_loop_bbs): Call fold_loop_internal_call.
21412 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
21415 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21418 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
21419 Require dead FLAGS_REG at the beginning of a peephole.
21421 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21424 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
21425 arguments in the call to __builtin_ia32_sbb_u32.
21426 (_subborrow_u64): Swap _X and _Y arguments in the call to
21427 __builtin_ia32_sbb_u64.
21429 2017-07-04 Jakub Jelinek <jakub@redhat.com>
21432 * tree-vrp.c (compare_assert_loc): Turn into a function template
21433 with stable template parameter. Only test if a->e is NULL,
21434 !a->e == !b->e has been verified already. Use e == NULL or
21435 e != NULL instead of e or ! e tests. If stable is true, don't use
21436 iterative_hash_expr, on the other side allow a or b or both NULL
21437 and sort the NULLs last.
21438 (process_assert_insertions): Sort using compare_assert_loc<false>
21439 instead of compare_assert_loc, later sort using
21440 compare_assert_loc<true> before calling process_assert_insertions_for
21441 in a loop. Use break instead of continue once seen NULL pointer.
21443 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21445 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21446 Cortex-R7 and Cortex-R8 processors.
21448 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21450 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
21451 uninitialized while src is not.
21453 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
21455 * common/config/arm/arm-common.c: Adjust include path for
21457 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
21458 (arm-cpu.h): Create in build directory. Adjust dependency rules.
21459 (arm-cpu-data.h): Likewise.
21460 (arm-cpu-cdata.h): Likewise.
21461 * config/arm/arm-cpu.h: Delete.
21462 * config/arm/arm-cpu-cdata.h: Delete.
21463 * config/arm/arm-cpu-data.h: Delete.
21465 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
21467 * config/arm/arm-cpus.in (cortex-a55): New.
21468 (cortex-a75): Likewise.
21469 (cortex-a75.cortex-a55): Likewise.
21470 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
21472 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
21473 * config/arm/arm-cpu-cdata.h: Regenerate.
21474 * config/arm/arm-cpu-data.h: Regenerate.
21475 * config/arm/arm-cpu.h: Regenerate.
21476 * config/arm/arm-tables.opt: Regenerate.
21477 * config/arm/arm-tune.md: Regenerate.
21479 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21481 * haifa-sched.c (sched_create_recovery_edges): Update profile.
21483 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21485 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
21488 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
21490 PR tree-optimization/81292
21491 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
21492 full_string_p, also call adjust_related_strinfos if the adjustment
21493 is simple, otherwise invalidate related strinfos.
21495 2017-07-04 Martin Liska <mliska@suse.cz>
21498 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
21499 newly created variable as DECL_IGNORED_P.
21501 2017-07-04 Martin Liska <mliska@suse.cz>
21504 * ipa-inline.c (inline_small_functions):
21505 Use xstrdup_for_dump.
21507 2017-07-04 Tom de Vries <tom@codesourcery.com>
21509 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
21511 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21514 * config/darwin.c (darwin_function_switched_text_sections):
21517 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
21519 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
21521 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21523 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
21525 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21527 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
21528 min_profitable_iters, and th as inclusive lower bounds.
21529 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
21530 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
21531 for min_profitable_iters and min_profitable_estimate.
21532 (vect_transform_loop): Treat th as an inclusive lower bound.
21533 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
21535 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21538 * config/darwin.c (darwin_function_switched_text_sections):
21539 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
21540 in two pieces, and suppress the use of buf.
21542 2017-07-03 Nathan Sidwell <nathan@acm.org>
21544 * hash-table.h (hash_table_mod1): Fix indentation.
21546 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21548 PR middle-end/81290
21549 * predict.c (force_edge_cold): Be more careful about propagation
21551 * profile-count.h (profile_probability::guessed,
21552 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
21554 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
21556 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
21558 * doc/invoke.texi (rcpc architecture extension): Document it.
21560 2017-07-03 Richard Biener <rguenther@suse.de>
21562 PR tree-optimization/60510
21563 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
21564 the scalar reduction PHI and use it.
21565 (vectorizable_reduction): Properly guard the single_defuse_cycle
21566 path for non-SLP reduction chains where we cannot use it.
21567 Rework reduc_def/index and vector type deduction. Rework
21568 vector operand gathering during reduction op code-gen.
21569 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
21570 chains dissolve the chain and leave it to non-SLP reduction
21573 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21575 * tree-data-ref.h (dr_alignment): Declare.
21576 * tree-data-ref.c (dr_alignment): New function.
21577 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
21578 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
21580 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
21582 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21584 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
21585 and base_misalignment fields.
21586 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
21587 * tree-data-ref.c: Include builtins.h.
21588 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
21589 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
21590 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
21591 * tree-vect-data-refs.c: Include tree-cfg.h.
21592 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
21593 fields instead of calculating an alignment here.
21594 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
21595 innermost_loop_behavior fields.
21597 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21599 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
21601 (DR_STEP_ALIGNMENT): New macro.
21602 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
21603 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
21604 (create_data_ref): Print it.
21605 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
21606 to tell whether the step preserves vector (mis)alignment.
21607 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
21608 Move the check for an integer step and generalise to all INTEGER_CST.
21609 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
21610 Print the outer step alignment.
21612 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21614 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
21615 with offset_alignment.
21616 (DR_ALIGNED_TO): Delete.
21617 (DR_OFFSET_ALIGNMENT): New macro.
21618 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
21619 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
21620 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
21621 (create_data_ref): Likewise.
21622 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
21623 (vect_analyze_data_refs): Likewise.
21624 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
21625 creating dummy innermost behavior.
21627 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21629 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
21630 with a "innermost_loop_behavior *" and refeence tree.
21631 * tree-data-ref.c (dr_analyze_innermost): Likewise.
21632 (create_data_ref): Update call accordingly.
21633 * tree-predcom.c (find_looparound_phi): Likewise.
21635 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21637 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
21638 fields with dr_wrt_vec_loop.
21639 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
21640 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
21641 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
21642 (vect_dr_behavior): New function.
21643 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21644 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
21645 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
21646 track whether the step preserves the misalignment.
21647 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21648 Use vect_dr_behavior.
21649 (vect_setup_realignment): Update call accordingly.
21650 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
21651 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
21652 call to vect_create_addr_base_for_vector_ref.
21653 (vect_create_cond_for_align_checks): Likewise.
21654 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
21655 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
21656 (vect_recog_mask_conversion_pattern): Likewise.
21657 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
21658 (new_stmt_vec_info): Remove redundant zeroing.
21660 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21662 * common/config/arm/arm-common.c (arm_be8_option): New function.
21663 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
21664 (ISA_ARMv6): Add isa_bit_be8.
21665 * config/arm/arm.h (arm_be8_option): Add prototype.
21666 (BE8_SPEC_FUNCTION): New define.
21667 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
21668 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
21669 (mlittle-endian): Similarly.
21670 (mbe8, mbe32): New options.
21671 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
21672 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
21674 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21676 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
21678 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21680 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
21681 (cleanup_tree_cfg_bb): Use it.
21682 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
21684 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
21686 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21689 * loop-doloop.c (add_test): Update profile.
21691 2017-07-03 Martin Liska <mliska@suse.cz>
21694 * sanopt.c (rewrite_usage_of_param): New function.
21695 (sanitize_rewrite_addressable_params): Likewise.
21696 (pass_sanopt::execute): Call rewrite_usage_of_param.
21698 2017-07-03 Richard Biener <rguenther@suse.de>
21700 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
21701 back to using VIEW_CONVERT_EXPR.
21703 2017-07-03 Martin Liska <mliska@suse.cz>
21706 * doc/extend.texi: Document when a resolver function is
21707 generated for target_clones.
21709 2017-07-03 Martin Liska <mliska@suse.cz>
21711 * asan.c (asan_emit_stack_protection): Unpoison just red zones
21712 and shadow memory of auto variables which are subject of
21713 use-after-scope sanitization.
21714 (asan_expand_mark_ifn): Add do set only when is_poison.
21716 2016-07-03 Richard Biener <rguenther@suse.de>
21718 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
21720 (vect_force_simple_reduction): Record reduction def -> phi mapping.
21721 (vectorizable_reduction): Perform reduction PHI creation when
21722 visiting a reduction PHI and adjust and simplify code generation
21723 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
21724 (vect_transform_loop): Visit reduction PHIs.
21725 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
21726 defs into the SLP tree.
21727 (vect_build_slp_tree): Reduction defs terminate the recursion.
21728 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
21730 (vect_get_vec_defs_for_stmt_copy): Export.
21731 (vect_get_vec_defs): Likewise.
21732 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
21734 (vect_get_vec_defs_for_stmt_copy): Declare.
21735 (vect_get_vec_defs): Likewise.
21737 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21739 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
21740 parameter with a "loop" parameter and use it instead of the
21741 loop containing DR_STMT. Don't check simple_iv when doing
21742 BB analysis. Describe the two analysis modes in the comment.
21744 2017-07-03 Tom de Vries <tom@codesourcery.com>
21746 PR tree-optimization/69468
21747 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
21748 (find_same_succ_bb): Handle ignore_edge_flags.
21750 2017-07-03 Tom de Vries <tom@codesourcery.com>
21752 PR tree-optimization/81192
21753 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
21755 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
21757 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
21759 2017-07-03 Tom de Vries <tom@codesourcery.com>
21761 PR tree-optimization/81192
21762 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
21763 BB_SAME_SUCC (bb) == NULL.
21765 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21767 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
21770 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21772 * dumpfile.c: Include profile-count.h
21773 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
21775 (insert_cond_bb): Update profile.
21776 * tree-cfg.h (insert_cond_bb): Update prototype.
21777 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
21778 * tree-dump.c: Do not include tree-cfg.
21780 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21782 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
21784 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21786 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
21789 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21791 * tree-complex.c (expand_complex_div_wide): update profile.
21793 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21794 Alan Hayward <alan.hayward@arm.com>
21795 David Sherwood <david.sherwood@arm.com>
21797 * Makefile.in (MACHMODE_H): Remove insn-modes.h
21798 (CORETYPES_H): New define.
21799 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
21800 (insn-modes-inline.h, s-modes-inline-h): New rules.
21801 (generated_files): Add insn-modes-inline.h.
21802 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
21803 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
21804 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
21805 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
21806 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
21807 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
21808 (build/gencodes.o, build/genconditions.o): Likewise.
21809 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
21810 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
21811 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
21812 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
21813 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
21814 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
21815 * coretypes.h: Include everything up to real.h for generators.
21816 Include insn-modes.h first. Include wide-int-print.h after
21817 wide-int.h. Include insn-modes-inline.h and then machmode.h.
21818 * machmode.h: Don't include insn-modes.h here.
21819 * function-tests.c: Remove includes of signop.h, machmode.h,
21820 double-int.h and wide-int.h.
21822 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
21824 * optc-save-gen.awk: Likewise.
21825 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
21826 * godump.c: Remove include of wide-int-print.h.
21827 * pretty-print.h: Likewise.
21828 * wide-int-print.cc: Likewise.
21829 * wide-int.cc: Likewise.
21830 * hash-map-tests.c: Remove include of signop.h.
21831 * hash-set-tests.c: Likewise.
21832 * rtl-tests.c: Likewise.
21833 * mkconfig.sh: Remove include of machmode.h.
21834 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
21836 (emit_insn_modes_inline_h): ...this new function. Emit the code
21837 into an insn-modes-inline.h header file, adding appropriate
21838 include guards and end comments.
21839 (emit_insn_modes_c_header): Remove include of machmode.h.
21840 (emit_min_insn_modes_c_header): Include coretypes.h rather than
21842 (main): Handle -i flag and call emit_insn_modes_inline_h when
21845 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21847 * tree-ssa-strlen.c (strinfo): Rename the length field to
21848 nonzero_chars. Add a full_string_p field.
21849 (compare_nonzero_chars, zero_length_string_p): New functions.
21850 (get_addr_stridx): Add an offset_out parameter.
21851 Use compare_nonzero_chars.
21852 (get_stridx): Update accordingly. Use compare_nonzero_chars.
21853 (new_strinfo): Update after above changes to strinfo.
21854 (set_endptr_and_length): Set full_string_p.
21855 (get_string_length): Update after above changes to strinfo.
21856 (unshare_strinfo): Update call to new_strinfo.
21857 (maybe_invalidate): Likewise.
21858 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
21859 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
21860 as a uhwi instead of an shwi. Update after above changes to
21861 strinfo and new_strinfo.
21862 (zero_length_string): Assert that chainsi contains full strings.
21863 Use zero_length_string_p. Update call to new_strinfo.
21864 (adjust_related_strinfos): Update after above changes to strinfo.
21865 Copy full_string_p from origsi.
21866 (adjust_last_stmt): Use zero_length_string_p.
21867 (handle_builtin_strlen): Update after above changes to strinfo and
21868 new_strinfo. Install the lhs as the string length if the previous
21869 entry didn't describe a full string.
21870 (handle_builtin_strchr): Update after above changes to strinfo
21872 (handle_builtin_strcpy): Likewise.
21873 (handle_builtin_strcat): Likewise.
21874 (handle_builtin_malloc): Likewise.
21875 (handle_pointer_plus): Likewise.
21876 (handle_builtin_memcpy): Likewise. Track nonzero characters
21877 that aren't necessarily followed by a nul terminator.
21878 (handle_char_store): Likewise.
21880 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21882 PR tree-optimization/80769
21883 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
21884 for malloc and calloc. Document the new invariant that all related
21885 strinfos have delayed lengths or none do.
21886 (verify_related_strinfos): Move earlier in file.
21887 (set_endptr_and_length): New function, split out from...
21888 (get_string_length): ...here. Also set the lengths of related
21890 (zero_length_string): Assert that chainsi has known (rather than
21892 (adjust_related_strinfos): Likewise.
21894 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21896 PR tree-optimization/81136
21897 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
21898 assert that two references with the same misalignment have the same
21899 compile-time misalignment if those compile-time misalignments
21902 2017-07-01 Andi Kleen <ak@linux.intel.com>
21904 * print-tree.c (print_node): Print all attributes.
21906 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21908 * cfg.c (scale_bbs_frequencies): New function.
21909 * cfg.h (scale_bbs_frequencies): Declare it.
21910 * cfgloopanal.c (single_likely_exit): Cleanup.
21911 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
21913 (scale_loop_profile): Likewise.
21914 (loop_version): Likewise.
21915 (create_empty_loop_on_edge): Update.
21916 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
21917 scale_loop_frequencies, scale_loop_profile, loopify,
21918 loop_version): Update prototypes.
21919 * modulo-sched.c (sms_schedule): Update.
21920 * predict.c (unlikely_executed_edge_p): Also check probability.
21921 (probably_never_executed_edge_p): Fix typo.
21922 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21923 * tree-parloops.c (gen_parallel_loop): Update.
21924 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
21925 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21926 * tree-ssa-loop-split.c (split_loop): Update.
21927 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21928 * tree-vect-loop-manip.c (vect_do_peeling): Update.
21929 (vect_loop_versioning): Update.
21930 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21932 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21934 * trans-mem.c (split_bb_make_tm_edge): Update profile.
21936 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21938 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
21939 to keep profile consistent.
21941 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21943 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
21944 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
21945 * profile-count.h (max_safe_multiplier): Make unsigned.
21946 (profile_count::guessed_zero): New.
21948 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21950 * bb-reorder.c (fix_up_crossing_landing_pad,
21951 fix_crossing_conditional_branches): Use make_single_succ_edge
21952 to keep profile consistent.
21954 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21956 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
21959 2017-07-01 Jakub Jelinek <jakub@redhat.com>
21962 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
21963 the right scopes, make sure cond_jump isn't preserved between multiple
21964 iterations. Search for fallthru edge whenever there are 3+ edges and
21965 use find_fallthru_edge for it.
21967 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21969 Patch by Alexander Monakov <amonakov@ispras.ru>
21970 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
21971 probabilities consistently.
21973 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21975 * pa.c (pa_expand_compare_and_swap_loop): Update call of
21976 emit_cmp_and_jump_insns.
21978 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21981 * tree-inline.c (expand_call_inline): Combine profile statuses.
21983 2017-06-30 Andrew Pinski <apinski@cavium.com>
21985 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
21986 fold_stmt returned true.
21988 2017-06-30 Nathan Sidwell <nathan@acm.org>
21990 * ggc.h (empty_string): Delete.
21991 * cfgexpand.c (expand_asm_stmt): Use plain "".
21992 * optabs.c (expand_asm_memory_barrier): Likewise.
21993 * stringpool.c (empty_string): Delete.
21994 (digit_vector, digit_string): Delete.
21995 (ggc_alloc_string): Use plain "", don't optimize single digit
21996 strings. Use ggc_alloc_atomic.
21998 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22000 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22001 comparison set and one other set, use the cost of the non-comparison
22004 2017-06-30 Nathan Sidwell <nathan@acm.org>
22006 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22009 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22011 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22012 loops. Remove now unneeded calls to gimple_switch_set_label() that
22013 just set removed labels to NULL_TREE.
22015 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22017 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22018 (set_range_info): ...here. Only call set_range_info_raw if domain
22020 (set_nonzero_bits): Call set_range_info_raw.
22021 * tree-ssanames.h (set_range_info_raw): New.
22023 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22026 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22027 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22028 of nonimmediate_operand and <store_mask_constraint> instead of m
22029 for the input operand. For V8FI iterator, always split if input
22030 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22031 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22032 <store_mask_predicate> instead of register_operand and
22033 <store_mask_constraint> instead of v for the input operand. Make
22034 sure both operands aren't MEMs for if not <mask_applied>.
22036 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22038 * lto-wrapper.c (copy_file) Close both file descriptors before
22041 2017-06-30 Martin Liska <mliska@suse.cz>
22044 * multiple_target.c (create_dispatcher_calls): Make ifunc
22045 also for function that don't have calls or are not referenced.
22047 2017-06-30 Richard Biener <rguenther@suse.de>
22049 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22050 analyze the first scalar stmt. Move vector type computation
22051 for the BB case here from ...
22052 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22053 live operation processing in the SLP case properly.
22055 2017-06-30 Richard Biener <rguenther@suse.de>
22057 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
22059 2017-06-30 Martin Liska <mliska@suse.cz>
22062 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
22063 before BUILT_IN_UNWIND_RESUME when ASAN is used.
22065 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
22067 * doc/invoke.texi (AArch64): Add missing options and remove redundant
22070 2017-06-30 Richard Biener <rguenther@suse.de>
22072 PR tree-optimization/81249
22073 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
22074 condition reduction result to original scalar type.
22076 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22078 * profile-count.h (enum profile_quality): Fix typos and whitespace
22081 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22083 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
22084 type for branch probabilities.
22086 2017-06-29 Julian Brown <julian@codesourcery.com>
22087 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22089 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
22090 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
22091 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
22092 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
22094 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22096 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
22097 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
22098 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
22099 CC usage from generic code to here.
22100 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
22101 CC usage into the target macros.
22103 2017-06-29 Maya Rashish <coypu@sdf.org>
22105 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
22108 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22110 * arm/arm-builtins.c: Include profile-count.h
22111 * except.c (sjlj_emit_function_enter): Use
22112 profile_probability::unlikely.
22114 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22116 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
22117 and tocrel_offset be pointer args rather than implicitly using
22119 (legitimate_constant_pool_address_p, rs6000_emit_move,
22120 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
22121 tocrel_offset and use in toc_relative_expr_p call.
22122 (print_operand, print_operand_address): Use static tocrel_base_oac
22123 and tocrel_offset_oac.
22124 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
22127 2017-06-29 Maya Rashish <coypu@sdf.org>
22129 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
22131 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
22133 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
22134 objects, take into account only the alignment of 'op0' and 'mode1' if
22137 2017-06-29 Steve Ellcey <sellcey@cavium.com>
22139 * ccmp.c (ccmp_tree_comparison_p): New function.
22140 (ccmp_candidate_p): Update to use above function.
22141 (get_compare_parts): New function.
22142 (expand_ccmp_next): Update to use new functions.
22143 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
22145 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
22146 take mode as argument.
22147 * ccmp.h (expand_ccmp_expr): Add mode as argument.
22148 * expr.c (expand_expr_real_1): Pass mode as argument.
22150 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
22152 * combine.c (combine_instructions): Print insns to dump_file, together
22155 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22157 * asan.c (asan_emit_stack_protection): Update.
22158 (create_cond_insert_point): Update.
22159 * auto-profile.c (afdo_propagate_circuit): Update.
22160 * basic-block.h (struct edge_def): Turn probability to
22161 profile_probability.
22162 (EDGE_FREQUENCY): Update.
22163 * bb-reorder.c (find_traces_1_round): Update.
22164 (better_edge_p): Update.
22165 (sanitize_hot_paths): Update.
22166 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
22167 (make_single_succ_edge): Update.
22168 (check_bb_profile): Update.
22169 (dump_edge_info): Update.
22170 (update_bb_profile_for_threading): Update.
22171 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
22172 probabilitycount to 0.
22173 * cfgbuild.c (compute_outgoing_frequencies): Update.
22174 * cfgcleanup.c (try_forward_edges): Update.
22175 (outgoing_edges_match): Update.
22176 (try_crossjump_to_edge): Update.
22177 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
22178 (expand_gimple_tailcall): Update.
22179 (construct_init_block): Use make_single_succ_edge.
22180 (construct_exit_block): Use make_single_succ_edge.
22181 * cfghooks.c (verify_flow_info): Update.
22182 (redirect_edge_succ_nodup): Update.
22183 (split_edge): Update.
22184 (account_profile_record): Update.
22185 * cfgloopanal.c (single_likely_exit): Update.
22186 * cfgloopmanip.c (scale_loop_profile): Update.
22187 (set_zero_probability): Remove.
22188 (duplicate_loop_to_header_edge): Update.
22189 * cfgloopmanip.h (loop_version): Update prototype.
22190 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
22191 (force_nonfallthru_and_redirect): Update.
22192 (update_br_prob_note): Update.
22193 (rtl_verify_edges): Update.
22194 (purge_dead_edges): Update.
22195 (rtl_lv_add_condition_to_bb): Update.
22196 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
22197 * cgraphunit.c (init_lowered_empty_function): Update.
22198 (cgraph_node::expand_thunk): Update.
22199 * cilk-common.c: Include profile-count.h
22200 * dojump.c (inv): Remove.
22201 (jumpifnot): Update.
22202 (jumpifnot_1): Update.
22203 (do_jump_1): Update.
22205 (do_jump_by_parts_greater_rtx): Update.
22206 (do_compare_rtx_and_jump): Update.
22207 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
22208 do_jump_1. do_compare_rtx_and_jump): Update prototype.
22209 * dwarf2cfi.c: Include profile-count.h
22210 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
22211 (sjlj_emit_dispatch_table): Likewise.
22212 * explow.c: Include profile-count.h
22213 * expmed.c (emit_store_flag_force): Update.
22214 (do_cmp_and_jump): Update.
22215 * expr.c (compare_by_pieces_d::generate): Update.
22216 (compare_by_pieces_d::finish_mode): Update.
22217 (emit_block_move_via_loop): Update.
22218 (store_expr_with_bounds): Update.
22219 (store_constructor): Update.
22220 (expand_expr_real_2): Update.
22221 (expand_expr_real_1): Update.
22222 * expr.h (try_casesi, try_tablejump): Update prototypes.
22223 * gimple-pretty-print.c (dump_probability): Update.
22224 (dump_profile): New.
22225 (dump_gimple_label): Update.
22226 (dump_gimple_bb_header): Update.
22227 * graph.c (draw_cfg_node_succ_edges): Update.
22228 * hsa-gen.c (convert_switch_statements): Update.
22229 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
22230 (find_if_case_1): Update.
22231 (find_if_case_2): Update.
22232 * internal-fn.c (expand_arith_overflow_result_store): Update.
22233 (expand_addsub_overflow): Update.
22234 (expand_neg_overflow): Update.
22235 (expand_mul_overflow): Update.
22236 (expand_vector_ubsan_overflow): Update.
22237 * ipa-cp.c (good_cloning_opportunity_p): Update.
22238 * ipa-split.c (split_function): Use make_single_succ_edge.
22239 * ipa-utils.c (ipa_merge_profiles): Update.
22240 * loop-doloop.c (add_test): Update.
22241 (doloop_modify): Update.
22242 * loop-unroll.c (compare_and_jump_seq): Update.
22243 (unroll_loop_runtime_iterations): Update.
22244 * lra-constraints.c (lra_inheritance): Update.
22245 * lto-streamer-in.c (input_cfg): Update.
22246 * lto-streamer-out.c (output_cfg): Update.
22247 * mcf.c (adjust_cfg_counts): Update.
22248 * modulo-sched.c (sms_schedule): Update.
22249 * omp-expand.c (expand_omp_for_init_counts): Update.
22250 (extract_omp_for_update_vars): Update.
22251 (expand_omp_ordered_sink): Update.
22252 (expand_omp_for_ordered_loops): Update.
22253 (expand_omp_for_generic): Update.
22254 (expand_omp_for_static_nochunk): Update.
22255 (expand_omp_for_static_chunk): Update.
22256 (expand_cilk_for): Update.
22257 (expand_omp_simd): Update.
22258 (expand_omp_taskloop_for_outer): Update.
22259 (expand_omp_taskloop_for_inner): Update.
22260 * omp-simd-clone.c (simd_clone_adjust): Update.
22261 * optabs.c (expand_doubleword_shift): Update.
22262 (expand_abs): Update.
22263 (emit_cmp_and_jump_insn_1): Update.
22264 (expand_compare_and_swap_loop): Update.
22265 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
22266 * predict.c (predictable_edge_p): Update.
22267 (edge_probability_reliable_p): Update.
22268 (set_even_probabilities): Update.
22269 (combine_predictions_for_insn): Update.
22270 (combine_predictions_for_bb): Update.
22271 (propagate_freq): Update.
22272 (estimate_bb_frequencies): Update.
22273 (force_edge_cold): Update.
22274 * profile-count.c (profile_count::dump): Add missing space into dump.
22275 (profile_count::debug): Add newline.
22276 (profile_count::differs_from_p): Explicitly convert to unsigned.
22277 (profile_count::stream_in): Update.
22278 (profile_probability::dump): New member function.
22279 (profile_probability::debug): New member function.
22280 (profile_probability::differs_from_p): New member function.
22281 (profile_probability::differs_lot_from_p): New member function.
22282 (profile_probability::stream_in): New member function.
22283 (profile_probability::stream_out): New member function.
22284 * profile-count.h (profile_count_quality): Rename to ...
22285 (profile_quality): ... this one.
22286 (profile_probability): New.
22287 (profile_count): Update.
22288 * profile.c (compute_branch_probabilities): Update.
22289 * recog.c (peep2_attempt): Update.
22290 * sched-ebb.c (schedule_ebbs): Update.
22291 * sched-rgn.c (find_single_block_region): Update.
22292 (compute_dom_prob_ps): Update.
22293 (schedule_region): Update.
22294 * sel-sched-ir.c (compute_succs_info): Update.
22295 * stmt.c (struct case_node): Update.
22296 (do_jump_if_equal): Update.
22297 (get_outgoing_edge_probs): Update.
22298 (conditional_probability): Update.
22299 (emit_case_dispatch_table): Update.
22300 (expand_case): Update.
22301 (expand_sjlj_dispatch_table): Update.
22302 (emit_case_nodes): Update.
22303 * targhooks.c: Update.
22304 * tracer.c (better_p): Update.
22305 (find_best_successor): Update.
22306 * trans-mem.c (expand_transaction): Update.
22307 * tree-call-cdce.c: Update.
22308 * tree-cfg.c (gimple_split_edge): Upate.
22309 (move_sese_region_to_fn): Upate.
22310 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
22311 * tree-eh.c (lower_resx): Upate.
22312 (cleanup_empty_eh_move_lp): Upate.
22313 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22314 * tree-inline.c (copy_edges_for_bb): Update.
22315 (copy_cfg_body): Update.
22316 * tree-parloops.c (gen_parallel_loop): Update.
22317 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
22318 (gimple_gen_time_profiler): Update.
22319 * tree-ssa-dce.c (remove_dead_stmt): Update.
22320 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22321 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
22322 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
22323 (unloop_loops): Update.
22324 (try_peel_loop): Update.
22325 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22326 * tree-ssa-loop-split.c (connect_loops): Update.
22327 (split_loop): Update.
22328 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22329 (hoist_guard): Update.
22330 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
22331 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
22332 (value_replacement): Update.
22333 * tree-ssa-reassoc.c (branch_fixup): Update.
22334 * tree-ssa-tail-merge.c (replace_block_by): Update.
22335 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
22336 (create_edge_and_update_destination_phis): Update.
22337 (compute_path_counts): Update.
22338 (recompute_probabilities): Update.
22339 (update_joiner_offpath_counts): Update.
22340 (freqs_to_counts_path): Update.
22341 (duplicate_thread_path): Update.
22342 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
22343 (struct switch_conv_info): Update.
22344 (gen_inbound_check): Update.
22345 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22346 (vect_do_peeling): Update.
22347 (vect_loop_versioning): Update.
22348 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22349 (optimize_mask_stores): Update.
22350 * ubsan.c (ubsan_expand_null_ifn): Update.
22351 * value-prof.c (gimple_divmod_fixed_value): Update.
22352 (gimple_divmod_fixed_value_transform): Update.
22353 (gimple_mod_pow2): Update.
22354 (gimple_mod_pow2_value_transform): Update.
22355 (gimple_mod_subtract): Update.
22356 (gimple_mod_subtract_transform): Update.
22357 (gimple_ic): Update.
22358 (gimple_stringop_fixed_value): Update.
22359 (gimple_stringops_transform): Update.
22360 * value-prof.h: Update.
22362 2017-06-29 Carl Love <cel@us.ibm.com>
22364 * config/rs6000/rs6000-c.c: Add support for built-in functions
22365 vector signed int vec_signed (vector float);
22366 vector signed long long vec_signed (vector double);
22367 vector signed int vec_signed2 (vector double, vector double);
22368 vector signed int vec_signede (vector double);
22369 vector signed int vec_signedo (vector double);
22370 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
22371 instruction generator.
22372 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22373 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
22374 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
22376 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
22377 vunsignede_v2df): Add define_expands.
22378 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
22379 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
22380 VEC_UNSIGNEDO): Add definitions.
22381 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22382 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
22383 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
22384 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
22385 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
22386 * config/rs6000/altivec.h (vec_signed, vec_signed2,
22387 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
22388 vec_unsignede, vec_unsignedo): Add builtin defines.
22389 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
22391 * doc/extend.texi: Update the built-in documentation file for the
22392 new built-in functions.
22394 2017-06-29 Richard Biener <rguenther@suse.de>
22396 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
22397 reduction chains to LOOP_VINFO_REDUCTIONS.
22398 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
22399 SLP reductions after processing reduction chains.
22401 2017-06-29 Nathan Sidwell <nathan@acm.org>
22403 * builtins.c (fold_builtin_FUNCTION): Use
22404 lang_hooks.decl_printable_name.
22406 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
22408 PR middle-end/81194
22409 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
22410 with only one label.
22411 * stmt.c (expand_case): Assert NCASES is greater than one.
22413 2017-06-29 Richard Biener <rguenther@suse.de>
22415 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
22417 (group_case_labels): Likewise.
22418 (find_taken_edge): Push sanity checking on val to workers...
22419 (find_taken_edge_cond_expr): ... here
22420 (find_taken_edge_switch_expr): ... and here, handle cases
22421 with just a default label.
22422 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
22423 (group_case_labels): Likewise.
22424 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
22425 group_case_labels does anything cleanup the CFG again.
22427 2017-06-29 Bin Cheng <bin.cheng@arm.com>
22429 PR tree-optimization/81196
22430 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
22431 exit condition comparing two IVs.
22433 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
22435 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
22436 profile to the dummy entry at the end of the list of architectures.
22437 * config/arm/arm-cpu-cdata.h: Regenerated.
22439 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22440 Michael Collison <michael.collison@arm.com>
22443 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
22445 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
22446 (*aarch64_reg_<mode>3_minus_mask): New pattern.
22447 (*aarch64_<optab>_reg_di3_mask2): New pattern.
22448 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
22449 of shift when the shift amount is masked with constant equal to
22450 the size of the mode.
22451 * config/aarch64/predicates.md (subreg_lowpart_operator): New
22454 2017-06-29 Martin Liska <mliska@suse.cz>
22456 * config/i386/i386.opt: Change range from [1,5] to [0,5].
22458 2017-06-29 Yury Gribov <tetra2005@gmail.com>
22461 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
22464 (edge_growth_cache_entry::edge_growth_cache_entry): New
22466 (reset_edge_growth_cache): Update to use constructor.
22468 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22470 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
22471 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22472 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22474 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
22476 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
22477 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
22479 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
22481 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
22482 (*-linux-uclibc*): Add t-uclibc tmake_file.
22483 * config/t-musl: New.
22484 * config/t-uclibc: New.
22486 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
22488 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
22490 (gen_comm_data): Emit architectural setting of arch_prof.
22491 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
22493 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
22494 (armv8-m.base, armv8-m.main): Likewise.
22495 * arm-protos.h (arm_build_target): Add profile field.
22496 (arch_option): Likewise.
22497 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
22499 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
22500 arm_active_target.profile.
22502 2017-06-28 Richard Biener <rguenther@suse.de>
22504 PR middle-end/81227
22505 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
22506 TYPE_OVERFLOW_WRAPS.
22507 * match.pd (negate_expr_p): Likewise.
22508 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
22509 fold_build2, not fold_binary.
22511 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22513 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
22514 Convert memory address to Pmode.
22515 (aarch64_print_operand): Assert MEM operands are always Pmode.
22517 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22520 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
22521 Remove redundant if.
22522 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
22523 * config/arm/aarch-common-protos.h
22524 (aarch_forward_to_shift_is_not_shifted_re): Remove.
22525 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
22527 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
22530 * multiple_target.c (create_dispatcher_calls): Set the default
22531 clone to be static, not public.
22533 2017-06-28 Richard Biener <rguenther@suse.de>
22535 * tree-vect-loop.c (vectorizable_reduction): Move special
22536 cond reduction IV var creation ...
22537 (vect_create_epilog_for_reduction): ... here. Remove induction_index
22538 parameter. Use STMT_VINFO_VECTYPE.
22539 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
22542 2017-06-28 Martin Liska <mliska@suse.cz>
22545 * ipa-visibility.c (non_local_p): Handle visibility.
22547 2017-06-28 Martin Liska <mliska@suse.cz>
22550 * common.opt: Add IntegerRange to various options.
22551 * opt-functions.awk (integer_range_info): New function.
22552 * optc-gen.awk: Add integer_range_info to cl_options struct.
22553 * opts-common.c (decode_cmdline_option): Handle
22554 CL_ERR_INT_RANGE_ARG.
22555 (cmdline_handle_error): Likewise.
22556 * opts.c (print_filtered_help): Show valid interval in
22557 when --help is provided.
22558 * opts.h (struct cl_option): Add range_min and range_max fields.
22559 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
22561 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
22563 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
22564 (x * C EQ/NE y * C): New transformation.
22566 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
22568 * genmultilib (combination_space): Accept '+' in option names.
22570 2017-06-28 Martin Liska <mliska@suse.cz>
22573 * asan.c (instrument_derefs): Bail out inner references
22574 that are hard register variables.
22576 2017-06-28 Jakub Jelinek <jakub@redhat.com>
22579 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
22580 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
22582 2017-06-28 Richard Biener <rguenther@suse.de>
22584 * tree-vectorizer.h (vect_get_vec_defs): Remove.
22585 (vect_get_slp_defs): Adjust.
22586 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
22588 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
22590 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
22591 get_initial_defs_for_reduction instead of vect_get_vec_defs.
22592 (vectorizable_reduction): Adjust.
22593 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
22595 (vect_get_slp_defs): Likewise.
22596 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
22597 (vectorizable_bswap): Adjust.
22598 (vectorizable_call): Likewise.
22599 (vectorizable_conversion): Likewise.
22600 (vectorizable_assignment): Likewise.
22601 (vectorizable_shift): Likewise.
22602 (vectorizable_operation): Likewise.
22603 (vectorizable_store): Likewise.
22604 (vectorizable_condition): Likewise.
22605 (vectorizable_comparison): Likewise.
22607 2017-06-28 Michael Collison <michael.collison@arm.com>
22610 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
22612 (arm_gen_movmemqi): Removed unused variable 'i'.
22613 Convert 'for' loop into 'while' loop.
22614 (arm_expand_prologue): Remove last unnecessary set of insn.
22615 (thumb_pop): Remove unused variable 'pushed_words'.
22616 (thumb_exit): Remove last unnecessary set of regs_to_pop.
22618 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22620 * config/s390/predicates.md: Use s390_rel_address_ok_p.
22621 * config/s390/s390-protos.h: Add prototype of
22622 s390_rel_address_ok_p.
22623 * config/s390/s390.c (s390_got_symbol): New function.
22624 (s390_rel_address_ok_p): New function.
22625 (legitimize_pic_address): Use s390_rel_address_ok_p.
22626 (s390_load_got): Use s390_got_symbol.
22627 (s390_option_override): Issue error if
22628 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
22629 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
22631 * config/s390/s390.opt: New option mpic-data-is-text-relative.
22633 2017-06-27 Andrew Pinski <apinski@cavium.com>
22635 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
22636 (X * copysign (1.0, X)): New pattern.
22637 (X * copysign (1.0, -X)): New pattern.
22638 (copysign (-1.0, CST)): New pattern.
22640 2017-06-27 Joseph Myers <joseph@codesourcery.com>
22642 * genmultilib (combination_space): Remove variable.
22643 Validate reuse rules against regular expression for any sequence
22644 of multilib options in any order.
22646 2017-06-27 Michael Collison <michael.collison@arm.com>
22648 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
22649 call aarch64_split_simd_combine.
22650 * (aarch64_combine_internal<mode>): Delete pattern.
22651 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
22652 Allow register and subreg operands.
22654 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22656 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
22657 specific need, just fallback on defaults.
22658 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
22660 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22661 Olivier Hainque <hainque@adacore.com>
22663 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
22665 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
22666 targets. Pick a default if no particular attempt applied.
22667 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
22670 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22672 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
22673 (x86_64-wrs-vxworks7): Likewise.
22675 2017-06-27 Marek Polacek <polacek@redhat.com>
22678 * ubsan.c (instrument_null): Check get_base_address's result for null.
22680 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22682 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
22684 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22686 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
22687 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
22688 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
22689 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
22690 New function types.
22691 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
22692 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
22693 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
22694 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
22695 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
22696 BUILT_IN_FEUPDATEENV): New builtins.
22697 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
22698 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
22699 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
22700 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
22702 (builtin_structptr_types): Adjust size.
22703 * tree.c (builtin_structptr_types): Add four entries.
22705 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22706 Olivier Hainque <hainque@adacore.com>
22708 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
22709 (TLS_SYM): New local macro, forcing reference to __tls__ on
22710 link command lines for VxWorks 7 RTPs, triggering initialization
22712 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
22713 OS features TLS support, true for RTPs on VxWorks 7.
22714 * config/vxworks.c (vxworks_override_options): Setup emutls
22717 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22719 * predict.c (test_prediction_value_range): Use -1U instead of -1
22720 to avoid narrowing conversion warning.
22721 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
22722 to avoid narrowing conversion warning.
22723 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
22725 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
22727 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22729 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
22730 64bit configurations.
22731 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
22732 (SIZE_TYPE): Likewise.
22733 * config/vxworks.c (vxworks_emutls_var_fields): Use
22734 long_unsigned_type_node instead of unsigned_type_node as the offset
22735 field type, which is "pointer" mode in emutls.c.
22737 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22740 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
22742 PR middle-end/81207
22743 * gimple-fold.c (replace_call_with_call_and_fold): Handle
22744 gimple_vuse copying separately from gimple_vdef copying.
22746 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22748 * value-prof.c (free_hist): Remove call to memset and the enclosing if
22751 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22752 Olivier Hainque <hainque@adacore.com>
22754 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
22755 for all vxworks7 targets.
22756 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
22757 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
22758 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
22759 variations for VX6/VX7 and 32/64bits later on in ...
22760 (VXWORKS_LIB_SPEC): Leverage new macros.
22761 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
22762 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
22764 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22766 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
22767 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
22769 2017-06-26 Carl Love <cel@us.ibm.com>
22771 * config/rs6000/rs6000-c.c: Add support for built-in functions
22772 vector bool char vec_reve (vector bool char);
22773 vector signed char vec_reve (vector signed char);
22774 vector unsigned char vec_reve (vector unsigned char);
22775 vector bool int vec_reve (vector bool int);
22776 vector signed int vec_reve (vector signed int);
22777 vector unsigned int vec_reve (vector unsigned int);
22778 vector bool long long vec_reve (vector bool long long);
22779 vector signed long long vec_reve (vector signed long long);
22780 vector unsigned long long vec_reve (vector unsigned long long);
22781 vector bool short vec_reve (vector bool short);
22782 vector signed short vec_reve (vector signed short);
22783 vector double vec_reve (vector double);
22784 vector float vec_reve (vector float);
22785 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
22786 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
22787 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
22788 (altivec_vreve): New pattern.
22789 * config/rs6000/altivec.h (vec_reve): New define.
22790 * doc/extend.texi (vec_rev): Update the built-in documentation file
22791 for the new built-in functions.
22793 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22795 PR tree-optimization/71815
22796 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
22798 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
22800 (slsr_process_phi): Likewise.
22801 (replace_uncond_cands_and_profitable_phis): Don't replace a
22802 multiply candidate with a stride of 1 (copy or cast).
22803 (phi_incr_cost): Call uses_consumed_by_stmt rather than
22805 (lowest_cost_path): Likewise.
22806 (total_savings): Likewise.
22808 2017-06-26 Richard Biener <rguenther@suse.de>
22811 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
22812 Use def_builtin_pure for all gather builtins.
22814 2017-06-26 Richard Biener <rguenther@suse.de>
22816 PR tree-optimization/81203
22817 * tree-tailcall.c (find_tail_calls): Do not move stmts into
22818 non-dominating BBs.
22820 2017-06-26 Marek Polacek <polacek@redhat.com>
22823 * doc/invoke.texi: Document -Wmultistatement-macros.
22825 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
22827 * doc/sourcebuild.texi (ARM-specific attributes): Document new
22828 arm_neon_ok_no_float_abi effective target.
22830 2017-06-26 Richard Biener <rguenther@suse.de>
22832 PR tree-optimization/80928
22833 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
22834 (copy_bbs): Set BB_DUPLICATED flag early.
22835 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
22837 (execute_on_shrinking_pred): Likewise.
22838 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
22839 BB_DUPLICATED blocks.
22840 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
22841 iterate over all PHIs considering removal of *gsi.
22843 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
22845 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
22848 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22850 * config/rs6000/rs6000-string.c: (expand_block_clear,
22851 do_load_for_compare, select_block_compare_mode,
22852 compute_current_alignment, expand_block_compare,
22853 expand_strncmp_align_check, expand_strn_compare,
22854 expand_block_move, rs6000_output_load_multiple)
22855 Move functions related to string/block move/compare
22856 to a separate file.
22857 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
22858 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
22859 for this function which is now used in two files.
22860 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
22861 * config.gcc: Add rs6000-string.o to extra_objs for
22862 targets powerpc*-*-* and rs6000*-*-*.
22864 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
22867 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
22868 32-bit, since indexed is not valid for DImode.
22869 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
22870 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
22871 (define_peephole2 for Altivec d-form load): Add 32-bit support.
22872 (define_peephole2 for Altivec d-form store): Likewise.
22875 * multiple_target.c (create_dispatcher_calls): Only create the
22876 dispatcher call if the function is the default clone of a
22877 versioned function.
22879 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
22881 PR middle-end/80902
22882 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
22883 a call, force the call to not be a tail call.
22885 2017-06-23 Jeff Law <law@redhat.com>
22887 * doc/contrib.texi: Add entry for Steven Pemberton's work on
22890 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
22892 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
22893 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
22894 handling for early expansion of vector shifts (sl,sr,sra,rl).
22895 (builtin_function_type): Add vector shift right instructions
22896 to the unsigned argument list.
22898 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
22900 rtl-optimizatoin/79286
22901 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
22902 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
22903 trap. PIC register plus a const unspec without offset can never trap.
22905 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
22907 * tree.h (builtin_structptr_type): New type.
22908 (builtin_structptr_types): Declare new array.
22909 * tree.c (builtin_structptr_types): New array.
22910 (free_lang_data, build_common_tree_nodes): Use it.
22912 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
22915 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
22918 2017-06-22 Matt Turner <mattst88@gmail.com>
22920 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
22921 Lake models to skylake case. Assume skylake for unknown
22922 models with clflushopt.
22924 2017-06-22 Jeff Law <law@redhat.com>
22926 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
22927 frame sizes that do not satisfy aarch64_uimm12_shift.
22929 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
22931 * profile-count.h (apply_probability,
22932 apply_scale, probability_in): Fix checks for zero.
22934 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22936 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
22937 * doc/cppdiropts.texi (-I @var{dir}): Document it.
22939 2016-06-22 Richard Biener <rguenther@suse.de>
22941 * tree-vect-loop.c (vect_model_reduction_cost): Handle
22942 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
22943 REDUC_MAX_EXPR support.
22944 (vectorizable_reduction): Likewise.
22945 (vect_create_epilog_for_reduction): Likewise.
22947 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
22949 * match.pd (A / (1 << B) -> A >> B): New.
22950 * generic-match-head.c: Include optabs-tree.h.
22951 * gimple-match-head.c: Likewise.
22952 * optabs-tree.h (target_supports_op_p): New.
22953 * optabs-tree.c (target_supports_op_p): New.
22955 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22957 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
22958 $gcc_cv_ld --help output.
22959 (gcc_cv_ld_demangle): Likewise.
22960 (gcc_cv_ld_eh_frame_hdr): Likewise.
22961 (gcc_cv_ld_pie): Likewise.
22962 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
22963 (gcc_cv_ld_buildid): Likewise.
22964 (gcc_cv_ld_sysroot): Likewise.
22965 (ld_bndplt_support): Likewise.
22966 (ld_pushpopstate_support): Likewise.
22967 * configure: Regenerate.
22968 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
22970 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22973 * config/i386/sse.md (round<mode>2): Renumber match_dup and
22974 operands indexes to avoid gap between operands and match_dups.
22976 2017-06-21 Andrew Pinski <apinski@cavium.com>
22978 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
22979 Increment Arith_shift and Arith_shift_reg by 1.
22980 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
22982 * config/aarch64/aarch64.c (thunderx_tunings): Enable
22983 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
22984 (aarch64_strip_extend): Add new argument and test for it.
22985 (aarch64_cheap_mult_shift_p): New function.
22986 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
22987 add a cost if it is true.
22988 Update calls to aarch64_strip_extend.
22989 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
22991 2017-06-21 Andrew Pinski <apinski@cavium.com>
22993 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
22995 (thunderxt88): Likewise.
22996 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
22997 (thunderx_prefetch_tune): New variable.
22998 (thunderx2t99_prefetch_tune): Update for the correct values.
22999 (thunderxt88_tunings): New variable.
23000 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23001 generic_prefetch_tune.
23002 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23004 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23006 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23007 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23008 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23009 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23010 (aarch64_atomic_cas<mode>, GPI): Likewise.
23012 2017-06-21 Martin Liska <mliska@suse.cz>
23014 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23015 statements on cold and hot labels.
23016 * predict.c (tree_estimate_probability_bb): Remove the
23017 prediction from this place.
23019 2017-06-21 Martin Liska <mliska@suse.cz>
23021 PR tree-optimization/79489
23022 * gimplify.c (maybe_add_early_return_predict_stmt): New
23024 (gimplify_return_expr): Call the function.
23025 * predict.c (tree_estimate_probability_bb): Remove handling
23027 * predict.def: Update comment about early return predictor.
23028 * gimple-predict.h (is_gimple_predict): New function.
23029 * predict.def: Change default value of early return to 66.
23030 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23032 * passes.def: Put pass_strip_predict_hints to the beginning of
23035 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23037 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23038 FUNCTION_DECL declarations.
23039 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23041 (dwaf2out_decl): Likewise.
23042 * godump.c (go_early_global_decl): Skip call to the real debug hook
23043 for FUNCTION_DECL declarations.
23044 * passes.c (rest_of_decl_compilation): Skip call to the
23045 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23046 -fdump-go-spec is passed.
23048 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23050 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23051 Reorder for compactness.
23052 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23053 (def_builtin_pure, def_builtin_pure2): New functions.
23054 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23056 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23058 * match.pd (nop_convert): New predicate.
23059 ((A +- CST1) +- CST2): Allow some NOP conversions.
23061 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23064 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
23065 with ctors/dtors if GOVD_SHARED is set.
23067 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23069 * config/aarch64/aarch64.md (movti_aarch64):
23070 Emit mov rather than orr.
23071 (movtf_aarch64): Likewise.
23072 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
23073 Emit mov rather than orr.
23075 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23077 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
23078 Swap alternatives, make integer dup more expensive.
23080 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23082 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23083 Return true for non-tls symbols.
23085 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
23087 * config/aarch64/aarch64-cores.def (cortex-a55): New.
23088 (cortex-a75): Likewise.
23089 (cortex-a75.cortex-a55): Likewise.
23090 * config/aarch64/aarch64-tune.md: Regenerate.
23091 * doc/invoke.texi (-mtune): Document new values for -mtune.
23093 2017-06-21 Tom de Vries <tom@codesourcery.com>
23095 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
23096 stack_size feature.
23097 (Effective-Target Keywords, Other attributes): Suggest using
23098 dg-add-options stack_size feature to get stack limit in stack_size
23099 effective target documentation.
23101 2017-06-21 Julian Brown <julian@codesourcery.com>
23102 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23104 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
23105 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
23106 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
23108 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
23109 attribute type list for neon_multiply.
23110 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
23111 attribute type list for neon_multiply.
23112 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
23113 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
23114 attribute type list for neon_multiply.
23115 * config/arm/types.md (crypto_pmull): Add.
23116 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
23117 attribute type list.
23119 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
23121 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
23124 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23126 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
23127 to make sure not to dereference a NULL cost_classes_ptr pointer.
23129 2017-06-20 Carl Love <cel@us.ibm.com>
23131 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23132 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23133 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23134 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23135 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23136 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23137 VMULOSW): New enum "unspec" values.
23138 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23139 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23140 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23141 altivec_vmulosw): New patterns.
23142 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23143 VMULOSW): Add definitions.
23145 2017-06-20 Julia Koval <julia.koval@intel.com>
23147 * config/i386/i386.c: Fix rounding expand for new pattern.
23148 * config/i386/subst.md: Fix pattern (parallel -> unspec).
23150 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23152 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
23153 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
23155 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23157 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
23160 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23162 * config/aarch64/aarch64-cores.def: Rearrange to sort by
23163 architecture, then by implementer ID.
23164 * config/aarch64/aarch64-tune.md: Regenerate.
23166 2017-06-20 Richard Biener <rguenther@suse.de>
23168 PR middle-end/81097
23169 * fold-const.c (split_tree): Fold to type before negating.
23171 2017-06-20 David Malcolm <dmalcolm@redhat.com>
23173 * diagnostic-show-locus.c
23174 (selftest::test_fixit_deletion_affecting_newline): New function.
23175 (selftest::diagnostic_show_locus_c_tests): Call it.
23177 2017-06-20 Andreas Schwab <schwab@suse.de>
23180 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
23183 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23185 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
23186 __ARM_FEATURE_COPROC according to support.
23188 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23190 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23191 Rewritten to avoid overflow for > 32-bit pointers.
23194 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
23195 by removing enum keyword.
23196 (ubsan_type_descriptor): Likewise. Formatting fix.
23199 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
23200 splitter): Require TARGET_SSE2 in the condition.
23202 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
23205 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
23206 for doing vector set of SFmode on ISA 3.0.
23207 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
23208 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
23210 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
23211 SFmode value into a V4SF variable that was extracted from another
23212 V4SF variable without converting the element to double precision
23213 and back to single precision vector format.
23214 (vsx_insert_extract_v4sf_p9_2): Likewise.
23216 2017-06-19 Jakub Jelinek <jakub@redhat.com>
23218 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
23219 in UWHI to avoid undefined overflow.
23222 * ubsan.h (enum ubsan_encode_value_phase): New.
23223 (ubsan_encode_value): Change second argument to
23224 enum ubsan_encode_value_phase with default value of
23225 UBSAN_ENCODE_VALUE_GENERIC.
23226 * ubsan.c (ubsan_encode_value): Change second argument to
23227 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
23228 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
23229 create_tmp_var_raw instead of create_tmp_var and use a
23231 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
23232 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
23233 ubsan_encode_value callers.
23236 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
23237 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
23238 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
23240 2017-06-19 Richard Biener <rguenther@suse.de>
23242 PR middle-end/81118
23243 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
23244 estimates if we changed anything.
23246 2017-06-19 Richard Biener <rguenther@suse.de>
23248 PR tree-optimization/80887
23249 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
23250 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
23251 simplified lookups, then reset mprts_hook.
23252 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
23254 (try_to_simplify): Likewise.
23256 2017-06-19 Martin Liska <mliska@suse.cz>
23259 * gimplify.c (gimplify_switch_expr):
23260 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
23262 2017-06-19 Martin Liska <mliska@suse.cz>
23264 * doc/install.texi: Document that PGO runs in 4 stages.
23266 2017-06-19 Martin Liska <mliska@suse.cz>
23269 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
23270 to dispatcher function name.
23271 * multiple_target.c (replace_function_decl): New function.
23272 (create_dispatcher_calls): Redirect both edges and references.
23274 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
23276 * profile-count.c (profile_count::dump): Dump quality.
23277 (profile_count::differs_from_p): Update for unsigned val.
23278 * profile-count.h (profile_count_quality): New enum.
23279 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
23281 2017-06-19 Richard Biener <rguenther@suse.de>
23283 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
23284 struct function as arg.
23285 (estimate_numbers_of_iterations): Export overload with loop arg.
23286 (free_numbers_of_iterations_estimates_loop): Use an overload of
23287 free_numbers_of_iterations_estimates instead.
23288 * tree-cfg.c (remove_bb): Adjust.
23289 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
23290 * tree-parloops.c (gen_parallel_loop): Likewise.
23291 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
23293 (tree_unroll_loops_completely): Likewise.
23294 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
23295 Use an overload instead and export.
23296 (estimated_loop_iterations): Adjust.
23297 (max_loop_iterations): Likewise.
23298 (likely_max_loop_iterations): Likewise.
23299 (estimate_numbers_of_iterations): Take struct function as arg
23301 (loop_exits_before_overflow): Adjust.
23302 (free_numbers_of_iterations_estimates_loop): Use an overload.
23303 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
23304 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
23306 2017-06-19 Richard Biener <rguenther@suse.de>
23309 * ipa-prop.c (find_constructor_constant_at_offset): Handle
23310 RANGE_EXPR conservatively.
23312 2017-06-16 Carl Love <cel@us.ibm.com>
23314 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23315 definitions for vec_float, vec_float2, vec_floato,
23316 vec_floate built-ins.
23317 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
23318 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
23320 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
23321 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
23322 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
23323 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
23324 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
23325 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
23326 vec_floato): Add builtin defines.
23327 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
23328 Update the built-in documentation file for the new built-in
23331 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23333 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
23334 (mthumb): Mark as the negative of -marm.
23336 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23338 * doc/invoke.texi (ARM Options, -mcpu): Document supported
23340 (ARM Options, -mtune): Document that this accepts the same
23341 extension options as -mcpu.
23342 (ARM Options, -mfpu): Document addition of -mfpu=auto.
23344 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23346 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
23347 permitted extensions.
23349 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23351 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
23352 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
23353 (armv8-m.main): Add option +nodsp.
23354 * config/arm/arm-cpu-cdata.h: Regenerated.
23356 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23358 * config/arm/t-fuchsia: New file.
23359 * config.gcc (arm*-*-fuchsia*): Use it.
23361 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23363 * config/arm/t-symbian: Rewrite for new option infrastructure.
23365 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23367 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
23368 (MULTILIB_REQUIRED): Likewise.
23370 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23372 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
23373 (MULTILIB_RESUE): Likewise.
23374 (MULTILIB_MATCHES): Likewise.
23375 (MULTLIB_REQUIRED): Likewise.
23377 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23379 * config/arm/t-rtems: Rewrite for new option framework.
23381 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23383 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
23384 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
23385 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
23386 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
23387 * config/arm/t-multilib: ... here.
23388 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
23389 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
23390 armv7-a and armv8*-a when A-profile libraries have not been built.
23391 * config/arm/t-rmprofile: Rewrite.
23393 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23395 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
23396 with a backslash. Remove the backslash after substituting unescaped
23398 * doc/fragments.texi (MULTILIB_REUSE): Document it.
23400 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23402 * config.gcc: (arm*-*-*): When building a-profile libraries, force
23403 the driver to pass through the default setting of -mfloat-abi.
23404 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
23406 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
23407 (all_feat_combs): New rule.
23408 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
23410 * config/arm/t-aprofile: Rewrite.
23412 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23414 * config/arm/arm.h (FPUTYPE_AUTO): Define.
23415 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
23416 fpu is not specified by the user/command-line.
23417 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
23418 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
23419 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
23420 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
23421 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
23422 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
23424 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23426 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
23427 * config/arm/t-arm-elf: Rewritten.
23429 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23431 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
23432 have some floating-point instructions.
23433 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
23434 (TARGET_MAYBE_HARD_FLOAT): New macro.
23435 * config/arm/arm-builtins.c (arm_init_builtins): Use
23436 TARGET_MAYBE_HARD_FLOAT.
23437 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
23439 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23441 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
23442 (configargs.h): Include it.
23443 (arm_print_hint_for_fpu_option): New function.
23444 (arm_parse_fpu_option): New function.
23445 (candidate_extension): New class.
23446 (arm_canon_for_multilib): New function.
23447 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
23448 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
23449 (ARCH_CANONICAL_SPECS): New macro.
23450 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
23452 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23454 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
23455 are set after handling multilib fragments. Set target_cpu_default2
23458 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23460 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
23462 (arm*-*-*): Set target_cpu_default2 to a quoted string.
23463 * config/arm/parsecpu.awk (check_cpu): Validate any extension
23465 (check_arch): Likewise.
23466 * config/arm/arm.c (arm_configure_build_target): Handle
23467 TARGET_CPU_DEFAULT being a string constant. Scan any feature
23468 options in the default.
23470 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23472 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
23473 when an option is an alias of another.
23474 * config/arm/parsecpu.awk (optalias): New parser token.
23475 (gen_comm_data): Mark non-alias options as such. Emit entries
23476 for extension aliases.
23477 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
23478 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
23479 (armv6kz, armv6zk, armv6t2): Likewise.
23480 (armv7): Make vfpv3-d16 an alias.
23481 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
23483 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
23484 Sort in canonical order.
23485 (armv8-a): Sort in canonical order.
23486 (armv8.1-a, armv8.2-a): Likewise.
23487 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
23489 (cortex-a9): Sort in canonical order.
23490 * config/arm/arm.c (selftests.h): Include it.
23491 (arm_test_cpu_arch_data): New function.
23492 (arm_run_self_tests): New function.
23493 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
23494 (targetm): Move declaration to the end of the file.
23495 * arm-cpu-cdata.h: Regenerated.
23497 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23499 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
23500 call to target_mode_check describing the type of option passed.
23501 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
23502 (arm_target_thumb_only): Use arm_parse_arch_option_name or
23503 arm_parse_cpu_option_name to match parameters against list of
23505 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
23506 arm_arch_core_flags data structure.
23507 * config/arm/arm-cpu_cdata.h: Regenerated.
23509 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23511 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
23513 (arm_print_hint_for_cpu_option): Likewise.
23514 (arm_print_hint_for_arch_option): Likewise.
23515 (arm_parse_cpu_option_name): Likewise.
23516 (arm_parse_arch_option_name): Likewise.
23517 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
23518 of entries in the all_fpus list.
23519 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
23520 (arm_parse_cpu_option_name): Declare.
23521 (arm_parse_arch_option_name): Declare.
23522 (arm_parse_option_features): Declare.
23523 (arm_intialize_isa): Declare.
23524 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
23526 (gen_comm_data): ... here. Make definitions non-static.
23527 * config/arm/arm-cpu-data.h: Regenerated.
23528 * config/arm/arm-cpu-cdata.h: Regenerated.
23530 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23532 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
23533 (cpu_arch_extension): New structure.
23534 (cpu_arch_option, arch_option, cpu_option): New structures.
23535 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
23536 architecture types.
23537 (gen_data): Generate new format data tables.
23538 * config/arm/arm.c (cpu_tune): New structure.
23539 (cpu_option, processors): Delete.
23540 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
23541 (arm_print_hint_for_cpu_option): ... this and ...
23542 (arm_print_hint_for_arch_option): ... this.
23543 (arm_parse_arch_cpu_name): Delete. Replace with ...
23544 (arm_parse_cpu_option_name): ... this and ...
23545 (arm_parse_arch_option_name): ... this.
23546 (arm_unrecognized_feature): Change type of target parameter to
23548 (arm_parse_arch_cpu_features): Delete. Replace with ...
23549 (arm_parse_option_features): ... this.
23550 (arm_configure_build_target): Rework to use new configuration data
23552 (arm_print_tune_info): Rework for new configuration data tables.
23553 * config/arm/arm-cpu-data.h: Regenerated.
23554 * config/arm/arm-cpu.h: Regenerated.
23556 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23558 * Makefile.in (OBJS): Move sbitmap.o from here ...
23559 (OBJS-libcommon): ... to here.
23561 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23563 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
23564 (ISA_ALL_CRYPTO): New macro.
23565 (ISA_ALL_SIMD): New macro
23566 (ISA_ALL_FP): New macro.
23567 * config/arm/arm.c (fpu_bitlist): Update initializer.
23568 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
23570 (arm9e): Add fpu. Add option for nofp
23571 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
23572 (arm926ej-s, arm1026ej-s): Likewise.
23573 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
23574 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
23575 neon-fp16, neon-vfpv4, nofp and nosimd.
23576 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
23577 (cortex-a8): Add fpu. Add option for nofp.
23578 (cortex-a9): Add fpu. Add options for nosimd and nofp.
23579 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
23580 (cortex-r4f): Add fpu.
23581 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
23582 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
23584 (cortex-r8): Likewise.
23585 (cortex-m4): Add fpu. Add option for nofp.
23586 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
23587 (cortex-a17.cortex-a7): Likewise.
23588 (cortex-a32): Add fpu. Add options for crypto and nofp.
23589 (cortex-a35, cortex-a53): Likewise.
23590 (cortex-a57): Add fpu. Add option for crypto.
23591 (cortex-a72, cortex-a73): Likewise.
23592 (exynos-m1): Likewise.
23593 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
23594 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
23595 (cortex-m33): Add fpu. Add option for nofp.
23596 * config/arm/arm-cpu-cdata.h: Regenerated
23597 * config/arm/arm-cpu-data.h: Regenerated.
23599 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23601 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
23602 (armv5te, armv5tej): Likewise.
23603 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
23604 (armv7): Add options fp and vfpv3-d16.
23605 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
23606 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
23608 (armv7ve): Likewise.
23609 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
23610 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
23611 (armv8-a): Add nocrypto option.
23612 (armv8.1-a, armv8.2-a): Likewise.
23613 (armv8-m.main): add options fp, fp.dp and nofp.
23615 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23617 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
23619 (armv8-a+crc): Delete.
23620 (armv8.1-a): Add options simd, crypto and nofp.
23621 (armv8.2-a): Add options fp16, simd, crypto and nofp.
23622 (armv8.2-a+fp16): Delete.
23623 (armv8-m.main): Add option dsp.
23624 (armv8-m.main+dsp): Delete.
23625 (cortex-a8): Add fpu. Add nofp option.
23626 (cortex-a9): Add fpu. Add nofp and nosimd options.
23627 * config/arm/parsecpu.awk (gen_data): Generate option tables and
23628 link to main cpu and architecture data structures.
23629 (gen_comm_data): Only put isa attributes from the main architecture
23631 (option): New statement for architecture and CPU entries.
23632 * arm.c (struct cpu_option): New structure.
23633 (struct processors): Add entry for options.
23634 (arm_unrecognized_feature): New function.
23635 (arm_parse_arch_cpu_name): Ignore any characters after the first
23637 (arm_parse_arch_cpu_feature): New function.
23638 (arm_configure_build_target): Separate out any CPU and architecture
23639 features and parse separately. Don't error out if -mfpu=auto is
23640 used with only an architecture string.
23641 (arm_print_asm_arch_directives): New function.
23642 (arm_file_start): Call it.
23643 * config/arm/arm-cpu-cdata.h: Regenerated.
23644 * config/arm/arm-cpu-data.h: Likewise.
23645 * config/arm/arm-tables.opt: Likewise.
23647 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23649 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
23650 assembler when it is not -mfpu=auto.
23652 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23654 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
23655 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
23656 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
23657 (ASM_CPU_SPEC): Rewrite.
23658 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
23659 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
23660 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
23661 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
23662 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
23663 copied string is NUL-terminated. Also strip any characters prefixed
23665 (arm_rewrite_selected_arch): New function.
23666 (arm_rewrite_march): New function.
23668 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23670 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
23671 (x_arm_cpu_string, x_arm_tune_string): Likewise.
23672 (march, mcpu, mtune): Convert to string-based options.
23673 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
23674 (arm_parse_arch_cpu_name): New function.
23675 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
23676 identify selected architecture or CPU.
23677 (arm_option_save): New function.
23678 (TARGET_OPTION_SAVE): Redefine.
23679 (arm_option_restore): Restore string options.
23680 (arm_option_print): Print string options.
23682 2017-06-16 Martin Sebor <msebor@redhat.com>
23684 PR tree-optimization/80933
23685 PR tree-optimization/80934
23686 * builtins.c (fold_builtin_3): Do not handle bcmp here.
23687 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
23688 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
23689 (gimple_fold_builtin): Call them.
23691 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23693 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
23694 as unlikely; update profile.
23696 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23698 * predict.c (force_edge_cold): Handle declaring edges impossible
23701 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23703 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
23705 (try_unroll_loop_completely): Fix reporting.
23707 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23709 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
23711 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
23714 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
23715 if given a non-constant argument for an intrinsic which requires a
23718 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23720 * profile.c (compare_freqs): New function.
23721 (branch_prob): Sort edge list.
23722 (find_spanning_tree): Assume that the list is priority sorted.
23724 2017-06-16 Richard Biener <rguenther@suse.de>
23726 PR tree-optimization/81090
23727 * passes.def (pass_record_bounds): Remove.
23728 * tree-pass.h (make_pass_record_bounds): Likewise.
23729 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
23730 make_pass_record_bounds): Likewise.
23731 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
23732 not free niter estimates at the beginning but at the end.
23733 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
23735 2017-06-16 Richard Biener <rguenther@suse.de>
23737 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
23738 initializer to workaround ICE in host GCC 4.8.
23740 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23742 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
23744 (clone_inlined_nodes): Update.
23746 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23748 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
23749 prefetch settings, and enable prefetching by default at -O3.
23751 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23753 * config/aarch64/aarch64.c (aarch64_override_options_internal):
23754 Set flag_prefetch_loop_arrays according to tuning data.
23756 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23758 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
23759 New tune structure.
23760 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
23761 [Unrelated to main purpose of the patch] Place the pointer field last
23762 to enable type checking errors when tune structure are wrongly merged.
23763 * config/aarch64/aarch64.c (generic_prefetch_tune,)
23764 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
23765 (thunderx2t99_prefetch_tune): New tune constants.
23766 (tune_params *_tunings): Update all tunings (no functional change).
23767 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
23768 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
23769 from tunings structures.
23771 2017-06-16 Jakub Jelinek <jakub@redhat.com>
23774 * ubsan.c (instrument_null): Add T argument, use it instead
23775 of computing it based on IS_LHS.
23776 (instrument_object_size): Likewise.
23777 (pass_ubsan::execute): Adjust instrument_null and
23778 instrument_object_size callers to pass gimple_get_lhs or
23779 gimple_assign_rhs1 result to it. Use instrument_null instead of
23780 calling get_base_address and instrument_mem_ref. Handle
23781 aggregate call arguments for object-size sanitization.
23783 2017-06-16 Yury Gribov <tetra2005@gmail.com>
23785 PR tree-optimization/81089
23786 * tree-vrp.c (is_masked_range_test): Validate operands of
23789 2017-06-15 Martin Sebor <msebor@redhat.com>
23792 * dumpfile.c (dump_register): Avoid calling memset to initialize
23793 a class with a default ctor.
23794 * gcc.c (struct compiler): Remove const qualification.
23795 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
23796 * hash-table.h: Ditto.
23797 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
23799 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
23800 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
23802 * params.h (struct param_info): Make struct members non-const.
23803 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
23804 with default initialization.
23805 * vec.h (vec_copy_construct, vec_default_construct): New helper
23807 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
23808 with vec_copy_construct.
23809 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
23810 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
23811 * doc/invoke.texi (-Wclass-memaccess): Document.
23813 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23815 * emit-rtl.h (is_leaf): Update comment about local
23816 register allocator.
23818 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23821 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
23822 for a variable to have a section before checking if the section has a
23824 Set section to.persistent if persistent attribute is set.
23825 Warn if .persistent attribute is used on an automatic variable.
23827 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
23829 PR rtl-optimization/80474
23830 * reorg.c (update_block): Do not ignore instructions in a delay slot.
23832 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
23834 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
23837 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
23839 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
23840 (casesi): Emit bounds checking as RTL.
23841 (casesi_internal_mips16_<mode>): Remove bounds checking.
23843 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
23845 * config/xtensa/xtensa.c (xtensa_option_override): Append
23846 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
23847 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
23848 xtensa_doloop_hooks): Define unconditionally.
23849 (xtensa_reorg_loops): Only call reorg_loops in the presence of
23851 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
23852 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
23853 for it in xtensa_option_override.
23854 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
23855 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
23857 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
23859 * doc/cppopts.texi: Document '-' special value to -MF.
23861 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
23863 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
23864 (cortex_a53_fconst): Likewise.
23865 (cortex_a53_fpmul): Likewise.
23866 (cortex_a53_f_load_64): Likewise.
23867 (cortex_a53_f_load_many): Likewise.
23868 (cortex_a53_advsimd_alu): Likewise.
23869 (cortex_a53_advsimd_alu_q): Likewise.
23870 (cortex_a53_advsimd_mul): Likewise.
23871 (cortex_a53_advsimd_mul_q): Likewise.
23872 (fpmac bypass): Add new bypass for fpmac-fpmac case.
23873 Add missing fmul, r2f_cvt and fconst cases.
23875 2017-06-14 Richard Biener <rguenther@suse.de>
23877 PR middle-end/81088
23878 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
23880 (fold_binary_loc): When associating do not treat pre-existing
23881 TREE_OVERFLOW on literal constants as a reason to allow
23882 TREE_OVERFLOW on associated literal constants.
23884 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23886 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
23887 (MASK_FEATURES): New macro.
23888 * config/sparc/sparc.c (sparc_option_override): Remove the special
23889 handling of -mfpu and generalize it to all MASK_FEATURES switches.
23891 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23893 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
23894 a division of 0 if non-call exceptions are enabled.
23896 2017-06-14 Andrew Pinski <apinski@cavium.com>
23897 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23900 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
23901 Improve vector initialization code gen for only variable case.
23903 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23905 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
23907 2017-06-14 Richard Biener <rguenther@suse.de>
23909 PR tree-optimization/81083
23910 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
23913 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23915 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
23916 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
23917 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
23918 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
23919 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
23920 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
23922 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23924 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
23925 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
23927 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23929 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
23931 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23933 * config/rs6000/t-rtems: Don't handle SPE.
23935 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23937 * config/rs6000/t-linux: Don't handle SPE.
23939 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23941 * config/rs6000/eabispe.h: Delete file.
23943 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23945 * config/rs6000/t-spe: Delete file.
23947 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23949 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
23950 (rs6000_legitimate_offset_address_p): Return false for anything in
23951 V2SImode or V2SFmode.
23953 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23955 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
23956 except V2SF and V2SI. Rearrange the vector modes, and add comments.
23957 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
23959 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
23960 (rs6000_legitimate_offset_address_p): Ditto.
23961 (rs6000_emit_move): Ditto.
23962 (rs6000_init_builtins): Remove V4HI_type_node.
23964 2017-06-13 Martin Liska <mliska@suse.cz>
23967 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
23968 (gate_asan): Likewise.
23969 * asan.h (asan_no_sanitize_address_p): Remove the function.
23970 (sanitize_flags_p): New function.
23971 * builtins.def: Fix coding style.
23972 * common.opt: Use renamed enum value.
23973 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
23974 * doc/extend.texi: Document no_sanitize attribute.
23975 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
23976 to SANITIZE_UNDEFINED_NONDEFAULT.
23977 * gcc.c (sanitize_spec_function): Use the renamed enum value.
23978 * gimple-fold.c (optimize_atomic_compare_exchange_p):
23979 Use sanitize_flags_p.
23980 * gimplify.c (gimplify_function_tree): Likewise.
23981 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
23982 * opts.c (parse_no_sanitize_attribute): New function.
23983 (common_handle_option): Use renamed enum value.
23984 * opts.h (parse_no_sanitize_attribute): Declare.
23985 * tree.c (sanitize_flags_p): New function.
23986 * tree.h: Declared here.
23987 * tsan.c: Use sanitize_flags_p.
23988 * ubsan.c (ubsan_expand_null_ifn): Likewise.
23989 (instrument_mem_ref): Likewise.
23990 (instrument_bool_enum_load): Likewise.
23991 (do_ubsan_in_current_function): Remove the function.
23992 (pass_ubsan::execute): Use sanitize_flags_p.
23993 * ubsan.h: Remove do_ubsan_in_current_function
23994 * tree-cfg.c (print_no_sanitize_attr_value): New function.
23995 (dump_function_to_file): Use it here.
23997 2017-06-13 Martin Jambor <mjambor@suse.cz>
23999 PR tree-optimization/80803
24000 PR tree-optimization/81063
24001 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24002 (propagate_subaccesses_across_link): Enqueue subtree whenever
24003 necessary instead of relying on the caller.
24005 2017-06-13 Martin Jambor <mjambor@suse.cz>
24007 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24008 that have a first_link.
24009 (sort_and_splice_var_accesses): Do not check first_link before
24011 (subtree_mark_written_and_enqueue): Likewise.
24012 (propagate_all_subaccesses): Likewise and do not stop at first
24013 parent with a first_link.
24015 2017-06-13 Martin Jambor <mjambor@suse.cz>
24017 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24020 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24022 * match.pd: New pattern.
24024 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24026 * tree-vrp.c (is_masked_range_test): New function.
24027 (register_edge_assert_for): Determine ranges for
24030 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24032 PR tree-optimization/67328
24033 * fold-const.c (maskable_range_p): New function.
24034 (build_range_check): Generate bittests if possible.
24036 2017-06-13 Martin Liska <mliska@suse.cz>
24038 * gimple-pretty-print.c (dump_probability): Add new argument.
24039 (dump_edge_probability): Dump both probability and count.
24040 (dump_gimple_label): Likewise.
24041 (dump_gimple_bb_header): Likewise.
24043 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24046 * config/avr/avr-devices.c: Fix indentation.
24047 * config/avr/gen-avr-mmcu-specs.c: Dito.
24049 2017-06-13 Richard Biener <rguenther@suse.de>
24051 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24052 instead get vector type from stmt_info.
24053 (vectorizable_reduction): Adjust. Remove dead code.
24055 2017-06-13 Richard Biener <rguenther@suse.de>
24057 PR middle-end/81065
24058 * fold-const.c (extract_muldiv_1): Remove bogus distribution
24059 case of C * (x * C2 + C3).
24060 (fold_addr_of_array_ref_difference): Properly fold index difference.
24062 2017-06-12 David S. Miller <davem@davemloft.net>
24065 * config/sparc/sparc.md (return expander): Emit frame blockage if
24066 function uses alloca.
24068 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
24070 * combine.c (make_field_assignment): Check len rather than the mode
24071 precision when calling force_to_mode.
24073 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
24075 Support multilibs and devices that see flash in RAM address range.
24078 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
24079 (avr_mcu_t) <flash_pm_offset>: New field.
24080 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
24081 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
24082 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
24083 (AVR_TINY_PM_OFFSET): Remove macro.
24084 * config/avr/avr.opt (-mshort-calls): New option.
24085 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24086 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
24087 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
24088 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
24089 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
24090 instead of avr_arch->have_jmp_call.
24091 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
24092 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
24093 avr_arch->flash_pm_offset to define.
24094 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
24095 new field flash_pm_offset. Add entry for avrxmega3.
24096 (avr_texinfo): Add entry for avrxmega3.
24097 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
24098 attiny212, attiny214,
24099 attiny412, attiny414, attiny416, attiny417,
24100 attiny814, attiny816, attiny817,
24101 attiny1614, attiny1616, attiny1617,
24102 attiny3214, attiny3216, attiny3217.
24103 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
24104 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
24105 (avr_print_operand_address) [AVR_TINY]: Same.
24106 (avr_asm_init_sections) <readonly_data_section>: Only patch
24107 callback if avr_arch->flash_pm_offset = 0.
24108 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
24109 for rodata if avr_arch->flash_pm_offset != 0.
24110 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
24111 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
24112 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
24113 (m_options): Append opt_rcall.
24114 (m_dirnames): Append dir_rcall.
24115 * config/avr/t-multilib: Regenerate.
24117 * configure.ac [target=avr]: Check whether avrxmega3 default
24118 linker description file works as needed.
24119 * configure: Regenerate.
24120 * doc/avr-mmcu.texi: Regenerate.
24121 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
24122 <__AVR_ARCH__>: Document avrxmega3 and 103.
24123 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
24124 <__AVR_SHORT_CALLS__>: Document it.
24125 <__AVR_PM_BASE_ADDRESS__>: Document it.
24126 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
24127 (AVR Variable Attributes) <progmem>: Document this is
24128 not needed for avrxmega3.
24129 (AVR Named Address Spaces) <__flash>: Dito.
24131 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
24133 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
24135 2017-06-12 Doug Rupp <rupp@adacore.com>
24137 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
24138 Append vxworks-stdint.h to the tm_file list.
24139 * config/vxworks-stdint.h: New file.
24141 2017-06-12 Martin Liska <mliska@suse.cz>
24143 PR tree-optimization/81041
24144 * tree-profile.c (gimple_gen_ic_func_profiler):
24145 Create an extra BB in profile-generate
24146 (gimple_gen_time_profiler): Likewise.
24148 2017-06-12 Jakub Jelinek <jakub@redhat.com>
24150 PR tree-optimization/81003
24151 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
24152 (update_range_test): Use it instead of force_gimple_operand_gsi.
24154 2017-06-12 Richard Biener <rguenther@suse.de>
24156 PR tree-optimization/81053
24157 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
24158 with backedge value not defined in loop. Simplify def stmt
24161 2017-06-11 Tom de Vries <tom@codesourcery.com>
24164 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
24166 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
24167 nvptx_cannot_force_const_mem.
24169 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24171 * opts.c (finish_options): Move test for flag_split_stack after
24172 it has been initialized.
24174 2017-06-11 Jason Merrill <jason@redhat.com>
24176 * tree.h (id_equal): New.
24177 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
24178 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
24179 instead of strcmp of IDENTIFIER_POINTER.
24181 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24183 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
24184 (mark_all_inlined_calls_cdtor): Fix formating.
24185 (inline_transform): Rescale profile before inlining.
24187 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24189 * cgraph.h (cgraph_edge::clone): Update prototype.
24190 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
24191 (cgraph_node::create_clone): Update.
24192 (cgraph_node::create_version_clone): Update.
24193 * tree-inline.c (copy_bb): Update.
24194 (expand_call_inline): Update.
24196 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24198 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
24199 factored out from ...
24200 (rs6000_emit_prologue): ... here.
24202 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24204 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
24205 factored out from ...
24206 (rs6000_emit_prologue): ... here.
24208 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24210 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
24212 (handle_missing_profiles): Fix computation of tp_first_run.
24213 (counts_to_freqs): Do not touch freqs when count is 0.
24215 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24217 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
24220 2017-06-10 Tom de Vries <tom@codesourcery.com>
24222 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24223 attributes): Document signal effective target.
24225 2017-06-10 Tom de Vries <tom@codesourcery.com>
24227 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24228 Document effective target stack_size.
24230 2017-06-09 David Malcolm <dmalcolm@redhat.com>
24232 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
24233 to the edit_context if they can be auto-applied.
24235 2017-06-9 Ian Lance Taylor <iant@golang.org>
24237 * opts.c (finish_options): If -fsplit-stack, disable implicit
24238 -forder-blocks-and-partition.
24239 * doc/invoke.texi (Optimize Options): Document that when using
24240 -fsplit-stack -forder-blocks-and-partition is not implicitly
24243 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24245 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
24246 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
24247 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
24248 * builtins.def (abort, trap, unreachable): Declare cold.
24249 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
24250 * tree-core.h (ECF_COLD): New.
24251 * tree.c (set_call_expr_flags): Handle ECF_COLD.
24252 (build_common_builtin_nodes): Mark unreachable and abort as cold.
24254 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24256 * predict.c (unlikely_executed_stmt_p): Cleanup.
24258 2017-06-09 Richard Biener <rguenther@suse.de>
24260 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
24261 model if the ref is always written to.
24263 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24265 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
24267 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24269 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
24272 2017-06-09 Tom de Vries <tom@codesourcery.com>
24275 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
24276 "target cannot support label values" when encountering LABEL_REF.
24278 2017-06-09 Martin Liska <mliska@suse.cz>
24280 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
24281 (gimple_gen_ic_func_profiler): Emit direct comparison
24282 of __gcov_indirect_call_callee with NULL.
24283 (gimple_gen_time_profiler): Change probability from
24284 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
24286 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24288 * profile.c (edge_gcov_counts): Turn to pointer.
24289 (compute_branch_probabilities, compute_branch_probabilities): Update.
24290 (branch_prob): Do not clear edge_gcov_count.
24291 * profile.h (edge_gcov_counts): Turn to pointer.
24292 (edge_gcov_count): Update.
24294 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24296 * gimple.h (gimple_check_failed): Mark cold.
24298 2017-06-09 Richard Biener <rguenther@suse.de>
24300 PR tree-optimization/66623
24301 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
24302 refactor check_reduction into two parts, properly computing
24303 whether we have to check reduction validity for outer loop
24306 2017-06-09 Richard Biener <rguenther@suse.de>
24308 PR tree-optimization/79483
24309 * graphite-scop-detection.c (order): New global.
24310 (get_order): Compute bb to order mapping that satisfies code
24311 generation constraints.
24312 (cmp_pbbs): New helper.
24313 (build_scops): Start domwalk at entry block, sort generated
24316 2017-06-09 Richard Biener <rguenther@suse.de>
24318 PR middle-end/81007
24319 * ipa-polymorphic-call.c
24320 (ipa_polymorphic_call_context::restrict_to_inner_class):
24321 Skip FIELD_DECLs with error_mark_node type.
24322 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
24325 2017-06-09 Martin Liska <mliska@suse.cz>
24327 * predict.c (struct branch_predictor): New struct.
24328 (test_prediction_value_range): New test.
24329 (predict_c_tests): New function.
24330 * selftest-run-tests.c (selftest::run_tests): Run the function.
24331 * selftest.h: Declare new tests.
24333 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
24336 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
24337 gen_add3_insn did not fail.
24338 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
24339 r0, construct that number in a temporary reg and add that reg to r0.
24340 If asked to put the result in r0 as well, fail.
24342 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
24344 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24345 for early expansion of vec_eqv.
24347 2017-06-08 Jakub Jelinek <jakub@redhat.com>
24349 PR middle-end/81005
24350 * ubsan.c (instrument_null): Avoid pointless code temporary.
24351 (pass_ubsan::execute): Instrument aggregate arguments of calls.
24353 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
24357 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
24360 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
24363 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24365 * predict.c (unlikely_executed_edge_p): Move ahead.
24366 (probably_never_executed_edge_p): Use it.
24368 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24370 PR middle-end/79988
24371 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
24372 gimple_call_builtin_p call.
24374 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24376 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
24377 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
24378 rtl_check_failed_type2, rtl_check_failed_code1,
24379 rtl_check_failed_code2, rtl_check_failed_code_mode,
24380 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
24381 rtvec_check_failed_bounds, rtl_check_failed_flag,
24382 _fatal_insn_not_found, _fatal_insn): Likewise.
24383 * tree.h (tree_contains_struct_check_failed,
24384 tree_check_failed, tree_not_check_failed,
24385 tree_class_check_failed, tree_range_check_failed,
24386 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
24387 tree_vec_elt_check_failed, phi_node_elt_check_failed,
24388 tree_operand_check_failed, omp_clause_check_failed,
24389 omp_clause_operand_check_failed, omp_clause_range_check_failed):
24392 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24394 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
24395 flag_branch_probabilities.
24396 * ipa-inline.c (edge_badness): Likewise.
24397 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
24398 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
24399 * predict.c (maybe_hot_frequency_p): Likewise.
24400 (probably_never_executed): Likewise.
24401 * sched-ebb.c (schedule_ebbs): Likewise.
24402 * sched-rgn.c (find_single_block_region): Likewise.
24403 * tracer.c (tail_duplicate): Likewise.
24405 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24407 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
24408 longer requires x_flag_profile_use.
24410 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24412 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
24413 instead of flag_reorder_blocks_and_partition.
24414 * dbxout.c (dbxout_function_end): Likewise.
24415 * dwarf2out.c (gen_subprogram_die): Likewise.
24416 * haifa-sched.c (sched_create_recovery_edges): Likewise.
24417 * hw-doloop.c (reorg_loops): Likewise.
24418 * varasm.c (assemble_start_function,
24419 assemble_end_function): Likewise.
24420 (decide_function_section): Do not check for
24421 flag_reorder_blocks_and_partition.
24423 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24425 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24427 (chkp_get_hard_register_fake_addr_expr): Ditto.
24428 (chkp_build_addr_expr): Add check for hard reg case.
24429 (chkp_parse_array_and_component_ref): Ditto.
24430 (chkp_find_bounds_1): Ditto.
24431 (chkp_process_stmt): Don't generate bounds store for
24434 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24436 * predict.c (maybe_hot_bb_p): Do not check profile status.
24437 (maybe_hot_edge_p): Likewise.
24438 (probably_never_executed): Check for zero counts even if profile
24440 (unlikely_executed_edge_p): New function.
24441 (unlikely_executed_stmt_p): New function.
24442 (unlikely_executed_bb_p): New function.
24443 (set_even_probabilities): Use unlikely predicates.
24444 (combine_predictions_for_bb): Likewise.
24445 (predict_paths_for_bb): Likewise.
24446 (predict_paths_leading_to_edge): Likewise.
24447 (determine_unlikely_bbs): New function.
24448 (estimate_bb_frequencies): Use it.
24449 (compute_function_frequency): Use zero counts even if profile is
24451 * profile-count.h: Fix typo.
24453 2017-08-08 Julia Koval <julia.koval@intel.com>
24455 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
24456 _mm512_mask_cvtsepi16_storeu_epi8,
24457 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
24458 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
24459 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
24460 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
24461 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
24462 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
24463 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
24464 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
24465 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
24466 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
24467 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
24468 __builtin_ia32_pmovuswb256mem_mask,
24469 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
24470 __builtin_ia32_pmovwb512mem_mask): New builtins.
24472 2017-08-08 Julia Koval <julia.koval@intel.com>
24474 PR target/73350,80862
24475 * config/i386/subst.md (round): Fix round pattern.
24476 * config/i386/i386.c (ix86_erase_embedded_rounding):
24477 Fix erasing rounding for the fixed pattern.
24479 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24481 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
24483 2017-06-08 Martin Liska <mliska@suse.cz>
24485 PR gcov-profile/80911
24486 * gcov.c (block_info::block_info): New constructor.
24488 2017-06-07 Carl Love <cel@us.ibm.com>
24490 * config/rs6000/rs6000-c: The return type of the following
24491 built-in functions was implemented as int not long long. Fix sign
24492 of return value for the unsigned version of vec_mulo and vec_mule.
24493 vector unsigned long long vec_bperm (vector unsigned long long,
24494 vector unsigned char)
24495 vector signed long long vec_mule (vector signed int,
24497 vector unsigned long long vec_mule (vector unsigned int,
24498 vector unsigned int)
24499 vector signed long long vec_mulo (vector signed int,
24501 vector unsigned long long vec_mulo (vector unsigned int,
24502 vector unsigned int)
24503 * doc/extend.texi: Fix the documentation for the built-in
24506 2017-06-07 Carl Love <cel@us.ibm.com>
24509 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
24512 2017-06-07 Carl Love <cel@us.ibm.com>
24514 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
24515 support, Generate doublehv for signed int/float for BE case only.
24517 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
24519 * doc/invoke.texi (mcx16): Rewrite.
24521 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24523 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
24524 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
24525 *mov<mode>_softfloat, and an anonymous splitter): Use
24526 nonimmediate_operand instead of rs6000_nonimmediate_operand.
24528 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24530 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
24532 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24533 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
24534 (rs6000_debug_reg_global): Adjust.
24535 (rs6000_init_hard_regno_mode_ok): Adjust.
24536 (rs6000_dbx_register_number): Adjust.
24537 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
24538 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24539 Remove SPE_ACC and SPEFSCR.
24540 (REG_ALLOC_ORDER): Ditto.
24541 (FRAME_POINTER_REGNUM): Change to 111.
24542 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
24543 (REG_CLASS_NAMES): Ditto.
24544 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
24545 (REGISTER_NAMES): Ditto.
24546 (ADDITIONAL_REG_NAMES): Ditto.
24547 (rs6000_reg_names): Ditto.
24548 * config/rs6000/rs6000.md: Renumber some register number
24551 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24553 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
24555 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24556 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
24558 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
24559 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24560 Delete the SPE high registers.
24561 (REG_ALLOC_ORDER): Ditto.
24562 (enum reg_class): Remove SPE_HIGH_REGS.
24563 (REG_CLASS_NAMES): Ditto.
24564 (REG_CLASS_CONTENTS): Delete the SPE high registers.
24565 (REGISTER_NAMES): Ditto.
24566 (rs6000_reg_names): Ditto.
24567 * doc/tm.texi.in: Remove SPE as example.
24568 * doc/tm.texi: Regenerate.
24570 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24572 * config/rs6000/8540.md (ppc8540_brinc): Delete.
24573 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
24574 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
24575 * config/rs6000/rs6000.md (type): Remove "brinc".
24577 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24579 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
24580 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
24581 * config/rs6000/linuxspe.h: Delete file.
24582 * config/rs6000/rs6000.md: Don't include spe.md.
24583 * config/rs6000/spe.h: Delete file.
24584 * config/rs6000/spe.md: Delete file.
24585 * config/rs6000/t-rs6000: Remove spe.md.
24587 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24589 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
24590 (reg_or_none500mem_operand): Delete.
24591 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
24592 instead of reg_or_none500mem_operand.
24594 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24596 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24597 handling of SPE flags.
24598 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
24600 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24602 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
24604 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
24605 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
24606 paired_divv2sf3): Similar.
24607 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
24608 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
24609 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
24611 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
24612 Rename the paired_* instruction patterns.
24613 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
24615 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
24616 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
24617 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
24618 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
24619 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
24620 PAIRED_VECTOR_MODE.
24621 (struct machine_function): Delete field spe_insn_chain_scanned_p.
24622 (spe_func_has_64bit_regs_p): Delete.
24623 (spe_expand_predicate_builtin): Delete.
24624 (spe_expand_evsel_builtin): Delete.
24625 (TARGET_DWARF_REGISTER_SPAN): Do not define.
24626 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
24627 (invalid_e500_subreg): Delete.
24628 (rs6000_legitimize_address): Always force_reg op2 as well, for
24629 paired single memory accesses.
24630 (rs6000_member_type_forces_blk): Delete.
24631 (rs6000_spe_function_arg): Delete.
24632 (rs6000_expand_unop_builtin): Delete SPE handling.
24633 (rs6000_expand_binop_builtin): Ditto.
24634 (spe_expand_stv_builtin): Delete.
24635 (bdesc_2arg_spe): Delete.
24636 (spe_expand_builtin): Delete.
24637 (spe_expand_predicate_builtin): Delete.
24638 (spe_expand_evsel_builtin): Delete.
24639 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
24640 (spe_init_builtins): Delete.
24641 (spe_func_has_64bit_regs_p): Delete.
24642 (savres_routine_name): Delete "info" parameter. Adjust callers.
24643 (rs6000_emit_stack_reset): Ditto.
24644 (rs6000_dwarf_register_span): Delete.
24645 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
24646 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
24647 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
24649 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
24651 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
24652 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
24653 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
24654 mulv2sf3, divv2sf3): Delete expanders.
24656 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24658 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
24660 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24662 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
24663 * config/rs6000/rs6000.c: Ditto.
24665 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24667 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
24668 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
24669 comparison_operator.
24671 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24673 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
24674 * config/rs6000/rs6000.opt: Ditto.
24675 * config/rs6000/t-rtems: Ditto.
24677 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24679 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
24680 TARGET_E500_SINGLE by 0, simplify.
24681 * config/rs6000/rs6000.c: Ditto.
24682 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
24683 (spe_build_register_parallel): Delete.
24684 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
24685 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
24686 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
24687 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
24688 (E500_CONVERT): Delete.
24689 * config/rs6000/spe.md: Remove many patterns and all define_constants.
24691 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24693 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
24694 * config/rs6000/dfp.md: Ditto.
24695 (negdd2, *negdd2_fpr): Merge.
24696 (absdd2, *absdd2_fpr): Merge.
24697 (negtd2, *negtd2_fpr): Merge.
24698 (abstd2, *abstd2_fpr): Merge.
24699 * config/rs6000/e500.h: Delete file.
24700 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
24701 TARGET_FPRS by 1 and simplify.
24702 * config/rs6000/rs6000-c.c: Ditto.
24703 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
24704 TARGET_DF_SPE by 0.
24705 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
24707 * config/rs6000/rs6000.md: Ditto.
24708 (floatdidf2, *floatdidf2_fpr): Merge.
24709 (move_from_CR_gt_bit): Delete.
24710 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
24711 (E500_CR_IOR_COMPARE): Delete.
24712 (All patterns that require !TARGET_FPRS): Delete.
24713 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
24715 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24717 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
24719 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24721 * graphds.c (add_edge): Intitialize edge's attached data.
24722 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
24723 pointer parameter. Call pointed function on each edge during
24724 graph traversing. Skip traversing the edge when the function
24726 (graphds_dfs, graphds_scc): Ditto.
24727 (for_each_edge): New parameter. Pass the new parameter to callback
24729 * graphds.h (skip_edge_callback): New function pointer type.
24730 (graphds_dfs, graphds_scc): New function pointer parameter.
24731 (graphds_edge_callback, for_each_edge): New parameter.
24733 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24735 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
24736 out code checking if runtime alias check is possible to below ...
24737 Call the new function.
24738 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
24739 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
24741 2017-06-07 Marek Polacek <polacek@redhat.com>
24744 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
24745 TYPE_OVERFLOW_WRAPS check.
24747 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24749 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
24750 if versioning is required.
24751 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
24752 peeling with the check for versioning.
24754 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24756 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
24757 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
24758 Set true to new parameter if new ssa variable is defined.
24759 (vect_gen_vector_loop_niters): Refactor. Set range information
24760 for the new vector loop bound variable.
24761 (vect_do_peeling): Ditto.
24763 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24765 * tree-affine.c (ssa.h): Include header file.
24766 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
24767 has wrapping overflow behavior.
24769 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24771 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
24773 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24775 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
24776 (tree_to_aff_combination): ... here.
24778 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24780 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
24781 reg_pressure model function.
24782 (ivopts_global_cost_for_size): Delete.
24783 (determine_set_costs, iv_ca_recount_cost): Call new model function
24784 ivopts_estimate_reg_pressure.
24786 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24788 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
24789 expensive than udiv. Remove floating point cases from mod.
24791 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24793 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
24794 Increase idiv cost.
24796 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24798 * config/aarch64/aarch64.md
24799 (copysignsf3): Fix mask generation.
24801 2017-06-07 Jakub Jelinek <jakub@redhat.com>
24803 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
24805 (class dump_manager): Add register_dumps method.
24806 * dumpfile.c: Include langhooks.h.
24807 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
24808 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
24809 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
24810 (dump_manager::dump_register): Start with 512 entries instead of 32.
24811 (dump_manager::register_dumps): New method.
24812 * toplev.c (general_init): Instead of invoking register_dumps
24813 langhook, invoke register_dumps method on the dump manager.
24814 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
24817 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
24819 * doc/md.texi: Clarify the restrictions on a define_insn condition.
24820 Say that # requires an associated define_split to exist, and that
24821 the define_split must be suitable for use after register allocation.
24823 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24825 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
24826 (compute_outgoing_frequencies): Also initialize zero counts.
24827 (find_many_sub_basic_blocks): Do not produce uninitialized profile
24828 around loops; preserve more of profile when nothing changes.
24830 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
24832 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
24834 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
24835 * config/arm/arm-cpu-cdata.h: Regenerate.
24836 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
24837 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
24838 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
24839 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
24840 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
24842 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
24844 * config/arm/t-rmprofile: Likewise.
24845 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
24847 2017-06-06 David S. Miller <davem@davemloft.net>
24850 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
24851 blockage if function uses alloca.
24853 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24855 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
24856 New "uid" fields to hold pretty-print IDs of group and ref.
24857 Memory references are now identified as <group_id>:<ref_id>
24858 instead of using [random] addresses.
24859 (dump_mem_details): Simplify, no functional change.
24860 (dump_mem_ref): Simplify and make output more concise.
24861 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
24862 (find_or_create_group): Initialize group uid.
24863 (record_ref): Initialize ref uid. Improve debug output.
24864 (prune_group_by_reuse, should_issue_prefetch_p,)
24865 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
24866 (mark_nontemporal_store, determine_loop_nest_reuse):
24867 Improve debug output.
24869 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24871 * dbgcnt.def (prefetch): New debug counter.
24872 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
24873 (schedule_prefetches): Stop issueing prefetches if debug counter
24876 2017-06-06 Tom de Vries <tom@codesourcery.com>
24878 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
24879 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
24881 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24883 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
24884 Use aarch64_reg_or_zero predicate for operand 4.
24885 (aarch64_compare_and_swap<mode> define_insn_and_split):
24886 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
24887 (aarch64_store_exclusive<mode>): Likewise for operand 2.
24889 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24891 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
24892 (arm_compute_save_core_reg_mask): This.
24893 (thumb1_compute_save_reg_mask): Rename into ...
24894 (thumb1_compute_save_core_reg_mask): This.
24895 (arm_compute_save_reg0_reg12_mask): Adapt comment.
24896 (arm_compute_frame_layout): Likewise.
24898 2017-06-06 Richard Biener <rguenther@suse.de>
24900 PR tree-optimization/80974
24901 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
24902 keep or clear leaders SSA info.
24904 2017-06-06 Tom de Vries <tom@codesourcery.com>
24906 * config/nvptx/nvptx.c (split_mode_p): New function.
24907 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
24909 2017-06-06 Tom de Vries <tom@codesourcery.com>
24911 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
24913 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24916 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
24919 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24921 * shrink-wrap.c (handle_simple_exit): Update profile.
24922 (try_shrink_wrapping): Upate profile.
24924 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24926 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
24927 (tree_guess_outgoing_edge_probabilities): New.
24928 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
24929 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
24931 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24933 * ipa-split.c (split_function): Initialize return bb profile.
24935 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24937 * profile.c (compute_branch_probabilities): Also initialize
24938 EXIT_BLOCK profile.
24940 2017-06-06 Richard Biener <rguenther@suse.de>
24942 PR tree-optimization/80928
24943 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
24944 (vect_analyze_loop_operations): Properly guard analysis for
24946 (vect_transform_loop): Likewise.
24947 (vect_analyze_loop_2): Also reset SLP type on PHIs.
24948 (vect_model_induction_cost): Do not cost for pure SLP.
24949 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
24950 of induction in inner loop vectorization.
24951 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
24952 (vect_get_and_check_slp_defs): Handle vect_induction_def.
24953 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
24955 (vect_analyze_slp_cost_1): Cost induction.
24956 (vect_detect_hybrid_slp_stmts): Handle PHIs.
24957 (vect_get_slp_vect_defs): Likewise.
24958 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
24959 (vect_transform_stmt): Handle SLP reductions.
24960 * tree-vectorizer.h (vectorizable_induction): Adjust.
24962 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24964 * config/rs6000/rs6000.c (make_resolver_func): Update
24965 init_lowered_empty_function call.
24967 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
24969 * doc/invoke.texi: Document the -fprofile-abs-path option.
24970 * common.opt (fprofile-abs-path): New option.
24971 * gcov-io.h (gcov_write_filename): Declare.
24972 * gcov-io.c (gcov_write_filename): New function.
24973 * coverage.c (coverage_begin_function): Use gcov_write_filename.
24974 * profile.c (output_location): Likewise.
24976 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24978 * shring-wrap.c: Revert accidental commit.
24980 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
24982 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
24984 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
24986 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
24988 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
24989 profile in callgraph edge.
24990 * profile-count.h (apply_probability): If THIS is 0, then result is 0
24991 (apply_scale): Likewise.
24992 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
24993 Also scale profile when inlining function with zero profile.
24994 (initialize_cfun): Update exit block profile even when it is zero.
24995 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
24996 when profile is read.
24998 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25000 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25001 (CLONE_*): New constants to define the processors we can generate
25002 code for with the target_clone attribute.
25003 (rs6000_clone_map): New array to identify which clone processors
25004 the current program is running on.
25005 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25006 target_clone attribute.
25007 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25008 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25009 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25010 (cpu_expand_builtin): Add support for target_clone attribute.
25011 (rs6000_valid_attribute_p): Allow "default" attribute.
25012 (get_decl_name): New debug function to simplify printing the
25013 current function name in debugging statements.
25014 (rs6000_clone_priority): New functions to support the target_clone
25015 attribute, and be able to generate code to switch between ISA 2.05
25016 through ISA 3.0 (power6 through power9).
25017 (rs6000_compare_version_priority): Likewise.
25018 (rs6000_get_function_versions_dispatcher): Likewise.
25019 (make_resolver_func): Likewise.
25020 (add_condition_to_bb): Likewise.
25021 (dispatch_function_versions): Likewise.
25022 (rs6000_generate_version_dispatcher_body): Likewise.
25023 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25024 (fusion_gpr_load_p): Fix a spacing issue.
25025 * doc/extend.texi (Common Function Attributes): Document that the
25026 PowerPC supports the target_clone attribute.
25028 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25030 * config/arm/arm.h: explain F symbol found in description of ARM
25031 register allocation in its legend.
25033 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25035 * config/mips/frame-header-opt.c: Include profile-count.h.
25036 * config/riscv/riscv.c: Include profile-count.h
25038 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25040 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25042 (sm_set_flag_if_changed): Add bbs field.
25043 (execute_sm_if_changed_flag_set): Pass BBS.
25044 (execute_sm): Update.
25046 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25048 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25051 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25053 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25054 (peephole2): New peephole2 to emit the above.
25055 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25057 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25059 * config/aarch64/aarch64.c (define_peephole2 above
25060 *sub_<shift>_<mode>): New peephole.
25062 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25064 * config/i386/i386.c (make_resolver_func): Update.
25065 * Makefile.in: Add profile-count.h and profile-count.o
25066 * auto-profile.c (afdo_indirect_call): Update to new API.
25067 (afdo_set_bb_count): Update.
25068 (afdo_propagate_edge): Update.
25069 (afdo_propagate_circuit): Update.
25070 (afdo_calculate_branch_prob): Update.
25071 (afdo_annotate_cfg): Update.
25072 * basic-block.h: Include profile-count.h
25073 (struct edge_def): Turn count to profile_count.
25074 (struct basic_block_def): Likewie.
25075 (REG_BR_PROB_BASE): Move to profile-count.h
25076 (RDIV): Move to profile-count.h
25077 * bb-reorder.c (max_entry_count): Turn to profile_count.
25078 (find_traces): Update.
25079 (rotate_loop):Update.
25080 (connect_traces):Update.
25081 (sanitize_hot_paths):Update.
25082 * bt-load.c (migrate_btr_defs): Update.
25083 * cfg.c (RDIV): Remove.
25084 (init_flow): Use alloc_block.
25085 (alloc_block): Uninitialize count.
25086 (unchecked_make_edge): Uninitialize count.
25087 (check_bb_profile): Update.
25088 (dump_edge_info): Update.
25089 (dump_bb_info): Update.
25090 (update_bb_profile_for_threading): Update.
25091 (scale_bbs_frequencies_int): Update.
25092 (scale_bbs_frequencies_gcov_type): Update.
25093 (scale_bbs_frequencies_profile_count): New.
25094 * cfg.h (update_bb_profile_for_threading): Update.
25095 (scale_bbs_frequencies_profile_count): Declare.
25096 * cfgbuild.c (compute_outgoing_frequencies): Update.
25097 (find_many_sub_basic_blocks): Update.
25098 * cfgcleanup.c (try_forward_edges): Update.
25099 (try_crossjump_to_edge): Update.
25100 * cfgexpand.c (expand_gimple_tailcall): Update.
25101 (construct_exit_block): Update.
25102 * cfghooks.c (verify_flow_info): Update.
25103 (dump_bb_for_graph): Update.
25104 (split_edge): Update.
25105 (make_forwarder_block): Update.
25106 (duplicate_block): Update.
25107 (account_profile_record): Update.
25108 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
25109 (get_estimated_loop_iterations): Update.
25110 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
25111 (single_likely_exit): Update.
25112 * cfgloopmanip.c (scale_loop_profile): Update.
25114 (set_zero_probability): Update.
25115 (lv_adjust_loop_entry_edge): Update.
25116 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
25117 (purge_dead_edges): Update.
25118 (rtl_account_profile_record): Update.
25119 * cgraph.c (cgraph_node::create): Uninitialize count.
25120 (symbol_table::create_edge): Uninitialize count.
25121 (cgraph_update_edges_for_call_stmt_node): Update.
25122 (cgraph_edge::dump_edge_flags): Update.
25123 (cgraph_node::dump): Update.
25124 (cgraph_edge::maybe_hot_p): Update.
25125 * cgraph.h: Include profile-count.h
25126 (create_clone), create_edge, create_indirect_edge): Update.
25127 (cgraph_node): Turn count to profile_count.
25128 (cgraph_edge0: Likewise.
25129 (make_speculative, clone): Update.
25130 (create_edge): Update.
25131 (init_lowered_empty_function): Update.
25132 * cgraphclones.c (cgraph_edge::clone): Update.
25133 (duplicate_thunk_for_node): Update.
25134 (cgraph_node::create_clone): Update.
25135 * cgraphunit.c (cgraph_node::analyze): Update.
25136 (cgraph_node::expand_thunk): Update.
25137 * final.c (dump_basic_block_info): Update.
25138 * gimple-streamer-in.c (input_bb): Update.
25139 * gimple-streamer-out.c (output_bb): Update.
25140 * graphite.c (print_global_statistics): Update.
25141 (print_graphite_scop_statistics): Update.
25142 * hsa-brig.c: Include basic-block.h.
25143 * hsa-dump.c: Include basic-block.h.
25144 * hsa-gen.c (T sum_slice): Update.
25145 (convert_switch_statements):Update.
25146 * hsa-regalloc.c: Include basic-block.h.
25147 * ipa-chkp.c (chkp_produce_thunks): Update.
25148 * ipa-cp.c (struct caller_statistics): Update.
25149 (init_caller_stats): Update.
25150 (gather_caller_stats): Update.
25151 (ipcp_cloning_candidate_p): Update.
25152 (good_cloning_opportunity_p): Update.
25153 (get_info_about_necessary_edges): Update.
25154 (dump_profile_updates): Update.
25155 (update_profiling_info): Update.
25156 (update_specialized_profile): Update.
25157 (perhaps_add_new_callers): Update.
25158 (decide_about_value): Update.
25159 (ipa_cp_c_finalize): Update.
25160 * ipa-devirt.c (struct odr_type_warn_count): Update.
25161 (struct decl_warn_count): Update.
25162 (struct final_warning_record): Update.
25163 (possible_polymorphic_call_targets): Update.
25164 (ipa_devirt): Update.
25165 * ipa-fnsummary.c (redirect_to_unreachable): Update.
25166 * ipa-icf.c (sem_function::merge): Update.
25167 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
25168 * ipa-inline.c (compute_uninlined_call_time): Update.
25169 (compute_inlined_call_time): Update.
25170 (want_inline_small_function_p): Update.
25171 (want_inline_self_recursive_call_p): Update.
25172 (edge_badness): Update.
25173 (lookup_recursive_calls): Update.
25174 (recursive_inlining): Update.
25175 (inline_small_functions): Update.
25176 (dump_overall_stats): Update.
25177 (dump_inline_stats): Update.
25178 * ipa-profile.c (ipa_profile_generate_summary): Update.
25179 (ipa_propagate_frequency): Update.
25180 (ipa_profile): Update.
25181 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25182 * ipa-utils.c (ipa_merge_profiles): Update.
25183 * loop-doloop.c (doloop_modify): Update.
25184 * loop-unroll.c (report_unroll): Update.
25185 (unroll_loop_runtime_iterations): Update.
25186 * lto-cgraph.c (lto_output_edge): Update.
25187 (lto_output_node): Update.
25188 (input_node): Update.
25189 (input_edge): Update.
25190 (merge_profile_summaries): Update.
25191 * lto-streamer-in.c (input_cfg): Update.
25192 * lto-streamer-out.c (output_cfg): Update.
25193 * mcf.c (create_fixup_graph): Update.
25194 (adjust_cfg_counts): Update.
25195 (sum_edge_counts): Update.
25196 * modulo-sched.c (sms_schedule): Update.
25197 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
25198 * predict.c (maybe_hot_count_p): Update.
25199 (probably_never_executed): Update.
25200 (dump_prediction): Update.
25201 (combine_predictions_for_bb): Update.
25202 (propagate_freq): Update.
25203 (handle_missing_profiles): Update.
25204 (counts_to_freqs): Update.
25205 (rebuild_frequencies): Update.
25206 (force_edge_cold): Update.
25207 * predict.h: Include profile-count.h
25208 (maybe_hot_count_p, counts_to_freqs): UPdate.
25209 * print-rtl-function.c: Do not include cfg.h
25210 * print-rtl.c: Include basic-block.h
25211 * profile-count.c: New file.
25212 * profile-count.h: New file.
25213 * profile.c (is_edge_inconsistent): Update.
25214 (correct_negative_edge_counts): Update.
25215 (is_inconsistent): Update.
25216 (set_bb_counts): Update.
25217 (read_profile_edge_counts): Update.
25218 (compute_frequency_overlap): Update.
25219 (compute_branch_probabilities): Update; Initialize and deinitialize
25221 (branch_prob): Update.
25222 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
25223 (edge_gcov_count): New.
25224 (bb_gcov_count): New.
25225 * shrink-wrap.c (try_shrink_wrapping): Update.
25226 * tracer.c (better_p): Update.
25227 * trans-mem.c (expand_transaction): Update.
25228 (ipa_tm_insert_irr_call): Update.
25229 (ipa_tm_insert_gettmclone_call): Update.
25230 * tree-call-cdce.c: Update.
25231 * tree-cfg.c (gimple_duplicate_sese_region): Update.
25232 (gimple_duplicate_sese_tail): Update.
25233 (gimple_account_profile_record): Update.
25234 (execute_fixup_cfg): Update.
25235 * tree-inline.c (copy_bb): Update.
25236 (copy_edges_for_bb): Update.
25237 (initialize_cfun): Update.
25238 (freqs_to_counts): Update.
25239 (copy_cfg_body): Update.
25240 (expand_call_inline): Update.
25241 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25242 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
25243 (try_unroll_loop_completely): Update.
25244 (try_peel_loop): Update.
25245 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25246 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
25247 * tree-ssa-loop-split.c (connect_loops): Update.
25248 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
25249 * tree-ssa-reassoc.c (branch_fixup): Update.
25250 * tree-ssa-tail-merge.c (replace_block_by): Update.
25251 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
25252 (compute_path_counts): Update.
25253 (update_profile): Update.
25254 (recompute_probabilities): Update.
25255 (update_joiner_offpath_counts): Update.
25256 (estimated_freqs_path): Update.
25257 (freqs_to_counts_path): Update.
25258 (clear_counts_path): Update.
25259 (ssa_fix_duplicate_block_edges): Update.
25260 (duplicate_thread_path): Update.
25261 * tree-switch-conversion.c (case_bit_test_cmp): Update.
25262 (struct switch_conv_info): Update.
25263 * tree-tailcall.c (decrease_profile): Update.
25264 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25265 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25266 * value-prof.c (check_counter): Update.
25267 (gimple_divmod_fixed_value): Update.
25268 (gimple_mod_pow2): Update.
25269 (gimple_mod_subtract): Update.
25270 (gimple_ic_transform): Update.
25271 (gimple_stringop_fixed_value): Update.
25272 * value-prof.h (gimple_ic): Update.
25274 2017-06-02 Carl Love <cel@us.ibm.com>
25276 * config/rs6000/rs6000-c: Add support for built-in functions
25277 vector double vec_doublee (vector signed int);
25278 vector double vec_doublee (vector unsigned int);
25279 vector double vec_doublee (vector float);
25280 vector double vec_doubleh (vector signed int);
25281 vector double vec_doubleh (vector unsigned int);
25282 vector double vec_doubleh (vector float);
25283 vector double vec_doublel (vector signed int);
25284 vector double vec_doublel (vector unsigned int);
25285 vector double vec_doublel (vector float);
25286 vector double vec_doubleo (vector signed int);
25287 vector double vec_doubleo (vector unsigned int);
25288 vector double vec_doubleo (vector float);.
25289 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
25290 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
25292 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
25293 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
25294 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
25296 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
25297 vec_doublel, vec_doubleh.
25298 * doc/extend.texi: Update the built-in documentation file for the
25299 new built-in functions.
25301 2017-06-02 David Malcolm <dmalcolm@redhat.com>
25304 * ipa-inline-analysis.c (free_growth_caches): Set
25305 edge_removal_hook_holder to NULL after removing it.
25307 2017-06-02 Sudakshina Das <sudi.das@arm.com>
25309 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
25310 comparision with zero.
25312 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25313 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25314 for early expansion of vec_min and vec_max builtins.
25315 (builtin_function_type): Add min/max unsigned variants to those
25316 identified as having unsigned arguments.
25318 2017-06-02 Olivier Hainque <hainque@adacore.com>
25320 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
25322 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25324 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
25325 Use VALL_F16 iterator rather than VALL.
25327 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25329 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
25330 Emit CBNZ inside loop when doing a strong exchange and comparing
25331 against zero. Generate the CC flags after the loop.
25333 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
25335 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
25336 (dl_section_ref): New.
25337 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
25338 On AIX, append an expression to subtract the size of the
25339 section length to dl_section_ref.
25341 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25343 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25344 for early expansion of vector absolute builtins.
25346 2017-06-02 Richard Biener <rguenther@suse.de>
25348 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
25349 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
25351 2017-06-02 Richard Biener <rguenther@suse.de>
25353 PR tree-optimization/80948
25354 * tree-tailcall.c (find_tail_calls): Track stmts to move in
25355 stmt order as well.
25357 2017-06-02 Richard Biener <rguenther@suse.de>
25359 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
25361 * tree-vect-stmts.c (process_use): Do not mark backedge defs
25362 for inductions as relevant.
25364 2017-06-02 Richard Biener <rguenther@suse.de>
25366 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
25367 (vectorizable_induction): ... this. Remove dead code.
25369 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
25371 * builtins. (expand_builtin_alloca): Remove second parameter and
25372 infer its value from the first parameter instead.
25373 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
25375 2017-06-02 Jakub Jelinek <jakub@redhat.com>
25377 PR rtl-optimization/80903
25378 * loop-doloop.c (add_test): Unshare sequence.
25380 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
25382 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
25384 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
25386 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
25388 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
25389 xlogue_layout::get_instance, logue_layout::xlogue_layout,
25390 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
25391 (xlogue_layout::get_stub_rtx): Make static.
25392 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
25393 (xlogue_layout::compute_stub_managed_regs): Rename to...
25394 (xlogue_layout::count_stub_managed_regs): ...this.
25395 (xlogue_layout::is_stub_managed_reg): New function.
25396 (xlogue_layout::m_stub_names): Rename to...
25397 (xlogue_layout::s_stub_names): ...this, make static.
25398 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
25399 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
25400 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
25401 xlogue_layout::s_stub_names): Instantiate statics.
25402 (stub_managed_regs): Remove.
25403 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
25404 (disable_call_ms2sysv_xlogues): Rename to...
25405 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
25406 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
25408 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
25409 change after reload_completed.
25410 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
25412 (ix86_expand_prologue): Likewise.
25413 (ix86_expand_epilogue): Likewise.
25414 (ix86_expand_split_stack_prologue): Likewise.
25415 (ix86_compute_frame_layout): Remove frame parameter ...
25416 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
25417 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
25419 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
25420 (ix86_frame): Move from here ...
25421 * config/i386/i386.h (ix86_frame): ... to here.
25422 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
25423 complete ix86_frame data structure instead. Remove some_ld_name.
25425 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25427 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
25428 symbols that hold a DECL_VALUE_EXPR.
25430 2017-06-01 Martin Jambor <mjambor@suse.cz>
25432 PR tree-optimization/80898
25433 * tree-sra.c (process_subtree_disqualification): Removed.
25434 (disqualify_candidate): Do not acll
25435 process_subtree_disqualification.
25436 (subtree_mark_written_and_enqueue): New function.
25437 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
25438 RHS has been disqualified and re-queue LHS if necessary. Apart
25439 from that, ignore disqualified RHS.
25441 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25443 * config/s390/s390.c (s390_emit_epilogue): Disable early return
25444 address fetch for z10 or later.
25446 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25448 * config/arc/arc.md (tst_movb): Add guard when splitting.
25450 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25452 * config/arc/arc.c (arc_can_eliminate): Test against
25453 arc_frame_pointer_needed.
25455 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25457 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
25458 to prevent store reordering.
25459 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
25460 (type): Add block type.
25461 (stack_tie): Define special instruction to be used in
25464 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25466 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
25467 constraint. It is not valid for the pattern.
25468 (noncommutative_binary_comparison): Likewise.
25470 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25472 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
25475 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25477 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
25478 be used by the reg-alloc.
25480 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25482 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
25483 reg-alloc when having mul64 or mul32x16 instructions.
25484 (mulsidi3): Likewise.
25485 (umulsidi3): Likewise.
25486 (mulsi32x16): New pattern.
25487 (mulsi64): Likewise.
25488 (mulsidi64): Likewise.
25489 (umulsidi64): Likewise.
25490 (MUL32x16_REG): Define.
25491 (mul64_600): Use MUL32x16_REG.
25492 (mac64_600): Likewise.
25493 (umul64_600): Likewise.
25494 (umac64_600): Likewise.
25496 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25498 * config/arc/arc.md (mulsi3_700): Make it commutative.
25500 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
25502 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
25504 (*sign_extendsidi2_insn): Likewise for movstosw.
25506 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25508 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
25509 the type of the input discriminant value. Convert the
25510 discriminant value of signedness vary.
25512 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
25514 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
25515 Add to -Wall section.
25517 2017-06-01 Richard Biener <rguenther@suse.de>
25519 PR middle-end/66313
25520 * fold-const.c (fold_plusminus_mult_expr): If the factored
25521 factor may be zero use a wrapping type for the inner operation.
25522 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
25523 and handle moved defs.
25524 (process_assignment): Properly guard the unary op case. Return a
25525 tri-state indicating that moving the stmt before the call may allow
25526 to continue. Pass through to_move.
25527 (find_tail_calls): Handle moving unrelated defs before
25530 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
25533 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
25534 splitter result in the canonical way.
25536 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
25538 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
25539 also for 32bit target. Update insn attributes.
25540 (zero-extendsidi2 splitter): Allow all registers for operand 1.
25542 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
25544 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
25545 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
25546 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
25547 (_mm_maskz_min_ss): New intrinsics.
25549 2017-05-31 Martin Liska <mliska@suse.cz>
25551 * tree-vect-loop.c (vect_create_epilog_for_reduction):
25552 Change comment style to one we normally use.
25553 (vectorizable_reduction): Likewise.
25554 (vectorizable_induction): Likewise.
25555 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
25556 (vectorizable_call): Likewise.
25557 (vectorizable_simd_clone_call): Likewise.
25558 (vectorizable_conversion): Likewise.
25559 (vectorizable_assignment): Likewise.
25560 (vectorizable_shift): Likewise.
25561 (vectorizable_operation): Likewise.
25562 (vectorizable_store): Likewise.
25563 (vectorizable_load): Likewise.
25564 * tree-vectorizer.h: Likewise.
25566 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
25568 * passes.c (emergency_dump_function): New.
25569 * tree-pass.h (emergency_dump_function): Declare.
25570 * plugin.c (plugins_internal_error_function): Remove.
25571 * plugin.h (plugins_internal_error_function): Remove declaration.
25572 * toplev.c (internal_error_function): New static function. Use it...
25573 (general_init): ...here.
25575 2017-05-31 Graham Markall <graham.markall@embecosm.com>
25577 * config/arc/arc.c (arc_print_operand): Handle constant operands.
25578 (arc_rtx_costs): Add costs for new patterns.
25579 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
25580 * config/arc/predicates.md: Add _1_2_3_operand predicate.
25582 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
25584 * tree-ssa-strlen.c (get_next_strinfo): New function.
25585 (get_stridx_plus_constant): Use it.
25586 (zero_length_string): Likewise.
25587 (adjust_related_strinfos): Likewise.
25588 (adjust_last_stmt): Likewise.
25590 2017-05-31 Richard Biener <rguenther@suse.de>
25593 * config/i386/i386.c (ix86_expand_builtin): Remove assert
25594 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
25596 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
25598 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
25599 loop_vinfo argument and use of dependence distance vectors.
25600 Check instead whether the two references differ only in their
25601 initial value and assume that they have the same alignment if the
25602 difference is a multiple of the vector alignment.
25603 (vect_analyze_data_refs_alignment): Update call accordingly.
25605 2017-05-31 Martin Liska <mliska@suse.cz>
25608 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
25610 2017-05-31 Bin Cheng <bin.cheng@arm.com>
25612 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
25613 (create_intersect_range_checks): Move from ...
25614 * tree-data-ref.c (create_intersect_range_checks_index)
25615 (create_intersect_range_checks): ... to here.
25616 (create_runtime_alias_checks): New function factored from ...
25617 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
25618 here. Call above function.
25619 * tree-data-ref.h (create_runtime_alias_checks): New function.
25621 2017-05-31 Bin Cheng <bin.cheng@arm.com>
25623 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
25624 segment length for dr_b and compute it in wide_int.
25626 2017-05-31 Richard Biener <rguenther@suse.de>
25628 PR tree-optimization/80906
25629 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
25630 and pass through iv_map.
25631 (copy_bb_and_scalar_dependences): Adjust.
25632 (translate_pending_phi_nodes): Likewise.
25633 (copy_loop_close_phi_args): Handle code-generating IVs instead
25636 2017-05-30 David Malcolm <dmalcolm@redhat.com>
25638 * diagnostic-color.c (color_dict): Add "type-diff".
25639 (parse_gcc_colors): Update comment.
25640 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
25641 -fdiagnostics-show-template-tree and -fno-elide-type.
25642 (GCC_COLORS): Add type-diff to example.
25644 (-fdiagnostics-show-template-tree): New.
25645 (-fno-elide-type): New.
25646 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
25647 the pp_format_decoder callback. Call any m_format_postprocessor's
25649 (pretty_printer::pretty_printer): Initialize
25650 m_format_postprocessor.
25651 (pretty_printer::~pretty_printer): Delete any
25652 m_format_postprocessor.
25653 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
25654 (class format_postprocessor): New class.
25655 (struct pretty_printer::format_decoder): Document the new parameters.
25656 (struct pretty_printer::m_format_postprocessor): New field.
25657 * tree-diagnostic.c (default_tree_printer): Update for new
25658 bool and const char ** params.
25659 * tree-diagnostic.h (default_tree_printer): Likewise.
25661 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
25663 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
25664 (lwa_operand): Delete rs6000_gen_cell_microcode test.
25665 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25666 rs6000_gen_cell_microcode code.
25667 (rs6000_final_prescan_insn): Delete.
25668 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
25669 "warn-cell-microcode" entries.
25670 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
25671 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
25672 throughout. Change cc_reg_not_micro_cr0_operand to
25673 cc_reg_not_cr0_operand throughout.
25674 (*extendhi<mode>2_noload): Delete.
25675 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
25676 (mwarn-cell-microcode): Delete.
25677 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
25678 -mgen-cell-microcode and -mwarn-cell-microcode.
25680 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
25683 * config/i386/constraints.md (Yd): New constraint.
25685 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
25686 and (?Yd, r) alternatives. Update insn attributes.
25687 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
25688 and (?*Yd, r) alternatives. Update insn attributes.
25689 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
25691 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
25693 * gimplify.c (gimplify_modify_expr): Don't create a
25694 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
25697 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
25699 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
25701 2017-05-30 Richard Biener <rguenther@suse.de>
25703 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
25704 and reduc_def fields.
25705 (STMT_VINFO_REDUC_TYPE): New define.
25706 (STMT_VINFO_REDUC_DEF): Likewise.
25707 (vect_force_simple_reduction): Adjust prototype.
25708 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
25709 (vect_is_simple_reduction): Remove check_reduction argument.
25710 (vect_force_simple_reduction): Adjust and set
25711 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25712 (vectorizable_reduction): Do not re-do reduction analysis
25713 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25714 * tree-parloops.c (gather_scalar_reductions): Adjust.
25716 2017-05-30 Richard Biener <rguenther@suse.de>
25718 PR middle-end/80901
25719 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
25722 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
25724 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
25725 Introduce unknown_misalignment parameter and remove vf.
25726 (vect_peeling_hash_get_lowest_cost):
25727 Pass unknown_misalignment parameter.
25728 (vect_enhance_data_refs_alignment):
25729 Fix unsupportable data ref treatment.
25731 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25733 * tree-vect-data-refs.c (vect_get_data_access_cost):
25734 Workaround for SLP handling.
25735 (vect_enhance_data_refs_alignment):
25736 Compute costs for doing no peeling at all, compare to the best
25737 peeling costs so far and avoid peeling if cheaper.
25739 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25741 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
25742 Return peeling info and set costs to zero for unlimited cost
25744 (vect_enhance_data_refs_alignment): Also inspect all datarefs
25745 with unknown misalignment. Compute and costs for unknown
25746 misalignment, compare them to the costs for known misalignment
25747 and choose the cheapest for peeling.
25749 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25751 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
25752 (vect_get_peeling_costs_all_drs): Create function.
25753 (vect_peeling_hash_get_lowest_cost):
25754 Use vect_get_peeling_costs_all_drs.
25755 (vect_peeling_supportable): Create function.
25756 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
25758 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25760 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
25761 DR_HAS_NEGATIVE_STEP.
25762 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
25763 (vect_enhance_data_refs_alignment): Use.
25764 (vect_duplicate_ssa_name_ptr_info): Use.
25765 * tree-vectorizer.h (dr_misalignment): Use.
25766 (known_alignment_for_access_p): Use.
25768 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25771 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
25773 (has_section_name): New function.
25775 2017-05-30 Martin Liska <mliska@suse.cz>
25778 * auto-profile.c (get_function_decl_from_block): Fix
25781 2017-05-30 Richard Biener <rguenther@suse.de>
25783 PR middle-end/80876
25784 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
25786 2017-05-30 Martin Liska <mliska@suse.cz>
25788 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
25789 * dumpfile.h (struct dump_file_info): Remove ctors.
25791 2017-05-30 Martin Liska <mliska@suse.cz>
25793 * predict.def: Fix GNU coding style.
25795 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25797 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25798 Mark 'to' argument with ATTRIBUTE_UNUSED.
25800 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25802 * config/xtensa/xtensa.c (xtensa_emit_call): Use
25803 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
25804 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
25807 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
25809 * doc/install.texi (Options specification): Restore entry of
25810 --enable-sjlj-exceptions.
25812 2017-05-27 Michael Eager <eager@eagercon.com>
25815 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
25817 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
25819 * config/microblaze/microblaze.h
25820 (FIXED_REGISTERS): Update in macro.
25821 (CALL_USED_REGISTERS): Update in macro.
25823 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
25825 * doc/install.texi: Add links to macOS binary distributions.
25827 2017-05-27 Jakub Jelinek <jakub@redhat.com>
25831 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25833 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25835 2017-05-26 Martin Liska <mliska@suse.cz>
25837 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
25839 2017-05-26 Martin Liska <mliska@suse.cz>
25841 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
25842 always leading ';; '.
25843 (dump_bb_info): Likewise.
25844 (brief_dump_cfg): Likewise.
25845 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
25846 * dumpfile.c: Remove usage of TDF_VERBOSE.
25847 * dumpfile.h (enum dump_kind): Likewise.
25848 (dump_gimple_bb_header): Do not use TDF_COMMENT.
25849 * print-tree.c (debug_verbose): Remove.
25850 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
25851 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
25852 * tree-diagnostic.c (default_tree_printer): Replace
25853 TDF_DIAGNOSTIC with TDF_SLIM.
25855 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25857 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
25858 in parameter loop, rather than loop_vinfo.
25859 (create_intersect_range_checks): Ditto.
25860 (vect_create_cond_for_alias_checks): Update call to above functions.
25862 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25864 PR tree-optimization/80815
25865 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
25866 for merging runtime alias checks. Handle negative DR_STEPs.
25868 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25870 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
25872 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
25873 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
25874 out code pruning runtime alias checks.
25875 * tree-data-ref.c (prune_runtime_alias_test_list): New function
25876 factored out from above.
25877 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25879 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25881 (prune_runtime_alias_test_list): New decalaration.
25883 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25885 * tree-vect-data-refs.c (compare_tree): Rename and move ...
25886 * tree-data-ref.c (data_ref_compare_tree): ... to here.
25887 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
25888 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
25889 (operator==, comp_dr_with_seg_len_pair): Ditto.
25890 (vect_prune_runtime_alias_test_list): Ditto.
25892 2017-05-26 Martin Liska <mliska@suse.cz>
25895 * params.def: Bound partial-inlining-entry-probability param.
25897 2017-05-26 Marek Polacek <polacek@redhat.com>
25900 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
25903 2017-05-26 Richard Biener <rguenther@suse.de>
25905 PR tree-optimization/80842
25906 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
25909 2017-05-26 Richard Biener <rguenther@suse.de>
25911 PR tree-optimization/80844
25912 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
25914 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
25916 * doc/md.texi (Machine Constraints): Update x86 family
25917 machine constraints section to match 'config/i386/constraints.md'.
25919 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
25921 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
25923 2017-05-25 Nathan Sidwell <nathan@acm.org>
25925 * doc/invoke.texi (--enable-languages): Update documentation.
25927 2017-05-25 Martin Liska <mliska@suse.cz>
25929 * dumpfile.c: Add TDF_FOLDING.
25930 * dumpfile.h (enum dump_kind): Likewise.
25931 * genmatch.c (dt_simplify::gen_1): Use it.
25933 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25935 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
25937 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25939 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25940 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
25942 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25944 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
25945 * match.pd (X == C): Rewrite it here.
25946 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
25947 with_certain_nonzero_bits2): New predicates.
25948 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
25950 2017-05-24 Nathan Sidwell <nathan@acm.org>
25952 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
25955 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
25958 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
25960 * config/powerpcspe: New port. Files are copied from the rs6000
25961 port, with "rs6000" in filenames replaced by "powerpcspe".
25963 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
25965 PR rtl-optimization/80754
25966 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
25968 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
25970 * config/sparc/sparc.md (length): Return the correct value for -mflat
25971 sibcalls to match output_sibcall.
25973 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
25977 * config/rs6000/rs6000.c (struct machine_function): Add new field
25979 (rs6000_get_separate_components): Init that field, use it.
25980 (rs6000_components_for_bb): Use the field.
25982 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
25984 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
25986 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
25988 PR middle-end/80823
25989 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
25991 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25994 * config/s390/s390.c (s390_check_qrst_address): Check incoming
25995 address against address_operand predicate.
25996 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
25998 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26000 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26001 parameters passed indirectly.
26003 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26005 * config/i386/i386.md (*movdi_internal): Remove SSE4
26006 alternative 18 (?r, *v). Update insn attributes.
26007 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26008 Update insn attributes.
26009 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26010 Update insn attributes.
26011 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26012 alternative 1 (r, v). Remove isa attribute.
26013 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26014 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26015 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26017 2017-05-23 Tom de Vries <tom@codesourcery.com>
26019 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26022 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26024 * cgraphunit.c (symbol_table::process_new_functions): Update.
26025 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26026 (inline_generate_summary): Rename to ...
26027 (ipa_fn_summary_generate): ... this one.
26028 (inline_read_summary): Rename to ...
26029 (ipa_fn_summary_read): ... this one.
26030 (inline_write_summary): Rename to ...
26031 (ipa_fn_summary_write): ... this one.
26032 (inline_free_summary): Rename to ...
26033 (ipa_free_fn_summary): ... this one.
26034 (pass_data_local_fn_summary, pass_local_fn_summary,
26035 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26036 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26037 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26038 make_pass_ipa_fn_summary): New.
26039 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26040 inline_write_summary, inline_free_summary): Remove.
26041 (ipa_free_fn_summary) : New.
26042 * ipa-inline.c (ipa_inline): Update.
26043 (pass_ipa_inline): Do not generate summaries.
26044 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26046 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26047 and add pass_ipa_fn_summary.
26048 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26050 (make_pass_inline_parameters): Remove.
26052 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26054 * omp-low.c (struct omp_context): Remove "default_kind" member.
26057 * omp-offload.c (execute_oacc_device_lower): Remove the
26058 parallelism dimensions function attributes for unparallelized
26059 OpenACC kernels constructs.
26061 2017-05-23 Martin Liska <mliska@suse.cz>
26063 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
26065 (cgraph_edge::make_speculative): Likewise.
26066 (cgraph_edge::resolve_speculation): Likewise.
26067 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26068 (cgraph_node::dump): Likewise.
26069 * cgraph.h: Likewise.
26070 * cgraphunit.c (analyze_functions): Likewise.
26071 (symbol_table::compile): Likewise.
26072 * ipa-cp.c (print_all_lattices): Likewise.
26073 (determine_versionability): Likewise.
26074 (initialize_node_lattices): Likewise.
26075 (ipcp_verify_propagated_values): Likewise.
26076 (estimate_local_effects): Likewise.
26077 (update_profiling_info): Likewise.
26078 (create_specialized_node): Likewise.
26079 (perhaps_add_new_callers): Likewise.
26080 (decide_about_value): Likewise.
26081 (decide_whether_version_node): Likewise.
26082 (identify_dead_nodes): Likewise.
26083 (ipcp_store_bits_results): Likewise.
26084 * ipa-devirt.c (dump_targets): Likewise.
26085 (ipa_devirt): Likewise.
26086 * ipa-icf.c (sem_item::dump): Likewise.
26087 (sem_function::equals): Likewise.
26088 (sem_variable::equals): Likewise.
26089 (sem_item_optimizer::read_section): Likewise.
26090 (sem_item_optimizer::execute): Likewise.
26091 (congruence_class::dump): Likewise.
26092 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
26093 (dump_inline_summary): Likewise.
26094 (estimate_node_size_and_time): Likewise.
26095 (inline_analyze_function): Likewise.
26096 * ipa-inline-transform.c (inline_call): Likewise.
26097 * ipa-inline.c (report_inline_failed_reason): Likewise.
26098 (want_early_inline_function_p): Likewise.
26099 (edge_badness): Likewise.
26100 (update_edge_key): Likewise.
26101 (inline_small_functions): Likewise.
26102 * ipa-profile.c (ipa_profile): Likewise.
26103 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
26104 (ipa_make_edge_direct_to_target): Likewise.
26105 (remove_described_reference): Likewise.
26106 (ipa_impossible_devirt_target): Likewise.
26107 (propagate_controlled_uses): Likewise.
26108 (ipa_print_node_params): Likewise.
26109 (ipcp_transform_function): Likewise.
26110 * ipa-pure-const.c (pure_const_read_summary): Likewise.
26111 (propagate_pure_const): Likewise.
26112 * ipa-reference.c (generate_summary): Likewise.
26113 (read_write_all_from_decl): Likewise.
26114 (propagate): Likewise.
26115 (ipa_reference_read_optimization_summary): Likewise.
26116 * ipa-utils.c (ipa_merge_profiles): Likewise.
26117 * ipa.c (walk_polymorphic_call_targets): Likewise.
26118 (symbol_table::remove_unreachable_nodes): Likewise.
26119 (ipa_single_use): Likewise.
26120 * passes.c (execute_todo): Likewise.
26121 * predict.c (drop_profile): Likewise.
26122 * symtab.c (symtab_node::get_dump_name): New function.
26123 (symtab_node::dump_name): Likewise.
26124 (symtab_node::dump_asm_name): Likewise.
26125 (symtab_node::dump_references): Likewise.
26126 (symtab_node::dump_referring): Likewise.
26127 (symtab_node::dump_base): Likewise.
26128 (symtab_node::debug_symtab): Likewise.
26129 * tree-sra.c (convert_callers_for_node): Likewise.
26130 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
26131 * value-prof.c (init_node_map): Likewise.
26133 2017-05-23 Martin Liska <mliska@suse.cz>
26135 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
26136 and symtab_node::debug_symtab to symbol_table::debug.
26137 * cgraphunit.c (analyze_functions): Use the renamed function.
26138 (symbol_table::compile): Likewise.
26139 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
26140 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
26141 * passes.c (execute_todo): Likewise.
26142 * symtab.c (symbol_table::dump): New function.
26143 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
26145 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26147 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
26148 that nonconst implies exec.
26150 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26152 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26153 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26154 (inline_edge_summary_vec): Turn into ...
26155 (ipa_call_summaries): ... this one.
26156 (redirect_to_unreachable, edge_set_predicate,
26157 evaluate_properties_for_edge, inline_summary_alloc,
26158 reset_ipa_call_summary, reset_inline_summary,
26159 inline_summary_t::duplicate): Update.
26160 (inline_edge_duplication_hook): Turn to ...
26161 (ipa_call_summary_t::duplicate): ... this one.
26162 (inline_edge_removal_hook): Turn to ...
26163 (ipa_call_summary_t::remove): ... this one.
26164 (dump_inline_edge_summary): Turn to ...
26165 (dump_ipa_call_summary): ... this one.
26166 (estimate_function_body_sizes): Update.
26167 (inline_update_callee_summaries): Update.
26168 (remap_edge_change_prob): Update.
26169 (remap_edge_summaries): Update.
26170 (inline_merge_summary): Update.
26171 (do_estimate_edge_time): Update.
26172 (inline_generate_summary): Update.
26173 (inline_read_section): Update.
26174 (inline_read_summary): Update.
26175 (inline_free_summary): Update.
26176 * ipa-inline.c (can_inline_edge_p): Update.
26177 (compute_inlined_call_time): Update.
26178 (want_inline_small_function_p): Update.
26179 (edge_badness): Update.
26180 (early_inliner): Update.
26181 * ipa-inline.h (inline_edge_summary): Turn to ...
26182 (ipa_call_summary): ... this one.
26183 (ipa_call_summary_t): New class.
26184 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26185 (ipa_call_summaries): New.
26186 (inline_edge_summary): Remove.
26187 (estimate_edge_growth): Update.
26188 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26189 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26190 * ipa-split.c (execute_split_functions): Update.
26191 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26193 2017-05-23 Tom de Vries <tom@codesourcery.com>
26195 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26196 attributes): Document rdrand effective target.
26198 2017-05-23 Tom de Vries <tom@codesourcery.com>
26200 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26201 attributes): Sort alphabetically.
26203 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
26205 * config/avr/genmultilib.awk: Use gsub instead of gensub.
26207 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
26210 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
26211 V2DF/V2DI splat into two separate patterns, one that handles
26212 registers, and the other that only handles memory. Drop support
26213 for splatting from a GPR on ISA 2.07 and then splitting the
26214 splat into direct move and splat.
26215 (vsx_splat_<mode>_reg): Likewise.
26216 (vsx_splat_<mode>_mem): Likewise.
26218 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
26220 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
26222 2017-05-22 Jakub Jelinek <jakub@redhat.com>
26224 PR middle-end/80809
26225 * omp-low.c (finish_taskreg_remap): New function.
26226 (finish_taskreg_scan): If unit size of ctx->record_type
26227 is non-constant, unshare the size expression and replace
26228 decls in it with possible outer var refs.
26230 PR middle-end/80809
26231 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
26232 GOVD_SHARED rather than GOVD_PRIVATE with it.
26233 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
26234 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
26236 PR middle-end/80853
26237 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
26238 as last argument to build_outer_var_ref for pointer bases of array
26239 section reductions.
26241 2017-05-19 Martin Sebor <msebor@redhat.com>
26243 * print-tree.c (print_node): Print DECL_READ_P flag.
26245 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26247 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
26248 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
26249 * cgraph.c: Likewise.
26250 * cgraphunit.c: Likewise.
26251 * gengtype.c: Likewise.
26252 * ipa-cp.c: Likewise.
26253 * ipa-devirt.c: Likewise.
26254 * ipa-icf.c: Likewise.
26255 * ipa-predicate.c: Likewise.
26256 * ipa-profile.c: Likewise.
26257 * ipa-prop.c: Likewise.
26258 * ipa-split.c: Likewise.
26260 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
26261 edge_predicate_pool, dump_inline_hints,
26262 inline_summary::account_size_time, redirect_to_unreachable,
26263 edge_set_predicate, set_hint_predicate,
26264 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
26265 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
26266 inline_summary_t::remove, remap_hint_predicate_after_duplication,
26267 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
26268 ipa_call_summary_t::remove, initialize_growth_caches,
26269 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
26270 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
26271 mark_modified, unmodified_parm_1, unmodified_parm,
26272 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
26273 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
26274 compute_bb_predicates, will_be_nonconstant_expr_predicate,
26275 will_be_nonconstant_predicate, record_modified_bb_info,
26276 get_minimal_bb, record_modified, param_change_prob,
26277 phi_result_unknown_predicate, predicate_for_phi_result,
26278 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
26279 estimate_function_body_sizes, compute_inline_parameters,
26280 compute_inline_parameters_for_curren, pass_data_inline_parameters,
26281 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
26282 inline_update_callee_summaries, remap_edge_change_prob,
26283 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
26284 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
26285 inline_analyze_function, inline_summary_t::insert,
26286 inline_generate_summary, read_ipa_call_summary, inline_read_section,
26287 inline_read_summary, write_ipa_call_summary, inline_write_summary,
26288 inline_free_summary): Move to ipa-fnsummary.h
26289 (predicate_t): Remove.
26290 * ipa-fnsummary.c: New file.
26291 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
26292 (enum inline_hints_vals, inline_hints, agg_position_info,
26293 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
26294 inline_summaries, ipa_call_summary, ipa_call_summary_t,
26295 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
26296 dump_inline_summary, dump_inline_hints, inline_generate_summary,
26297 inline_read_summary, inline_write_summary, inline_free_summary,
26298 inline_analyze_function, initialize_inline_failed,
26299 inline_merge_summary, inline_update_overall_summary,
26300 compute_inline_parameters): Move to ipa-fnsummary.h
26301 * ipa-fnsummary.h: New file.
26302 * ipa-inline-transform.h: Include ipa-inline.h.
26303 * ipa-inline.c: LIkewise.
26305 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26307 * ipa-inline.c (edge_badness): Use inlined_time instead of
26308 inline_summaries->get.
26310 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26312 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
26314 2017-05-22 Nathan Sidwell <nathan@acm.org>
26316 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
26317 (fdump-lang): Document 'raw' option.
26318 * dumpfile.h (TDI_tu): Delete.
26319 * dumpfile.c (dump_files): Remove translation-unit.
26320 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
26322 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
26324 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
26325 command option from $(AWK) call.
26326 * config/avr/genmultilib.awk: Simplify and rewrite so that it
26327 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
26328 [FORMAT]: Remove handling of variable.
26329 * config/avr/t-multilib: Regenerate.
26331 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26333 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
26335 (dump_inline_summary): Do not print self_time.
26336 (estimate_function_body_sizes): Do not set self_time.
26337 (compute_inline_parameters): Likewise.
26338 (inline_read_section, inline_write_summary): Do not stream self_time.
26339 * ipa-inline.h (inline_summary): Drop self_time.
26341 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26343 * ipa-inline-analysis.c (account_size_time): Rename to ...
26344 (inline_summary::account_size_time): ... this one.
26345 (reset_ipa_call_summary): Turn to ...
26346 (ipa_call_summary::reset): ... this one.
26347 (reset_inline_summary): Turn to ...
26348 (inline_summary::reset): ... this one.
26349 (inline_summary_t::remove): Update.
26350 (inline_summary_t::duplicate): Update.
26351 (ipa_call_summary_t::remove): Update.
26352 (dump_inline_summary): Update.
26353 (estimate_function_body_sizes): Update.
26354 (compute_inline_parameters): Update.
26355 (estimate_node_size_and_time): Update.
26356 (inline_merge_summary): Update.
26357 (inline_update_overall_summary): Update.
26358 (inline_read_section): Update.
26359 (inline_write_summary): Update.
26360 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
26361 add account_size_time and reset member functions.
26362 (ipa_call_summary): Add reset function.
26363 * ipa-predicate.h (predicate::operator &): Constify.
26365 2017-05-22 Richard Biener <rguenther@suse.de>
26367 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
26369 2017-05-19 Jason Merrill <jason@redhat.com>
26371 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
26373 2017-05-19 Marek Polacek <polacek@redhat.com>
26376 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
26377 TYPE_OVERFLOW_WRAPS checks.
26379 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
26381 * tree-core.h (enum omp_clause_default_kind): Add
26382 "OMP_CLAUSE_DEFAULT_PRESENT".
26383 * tree-pretty-print.c (dump_omp_clause): Handle it.
26384 * gimplify.c (enum gimplify_omp_var_data): Add
26385 "GOVD_MAP_FORCE_PRESENT".
26386 (gimplify_adjust_omp_clauses_1): Map it to
26387 "GOMP_MAP_FORCE_PRESENT".
26388 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
26390 * gimplify.c (oacc_default_clause): Clarify.
26392 2017-05-19 Nathan Sidwell <nathan@acm.org>
26394 LANG_HOOK_REGISTER_DUMPS
26395 * toplev.c (general_init): Call register dump lang hook.
26396 * doc/invoke.texi: Document -fdump-lang option family.
26397 * dumpfile.c (dump_files): Remove class dump here.
26398 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
26399 * dumpfile.h (tree_dump_index): Remove TDI_class.
26400 * langhooks-def.h (lhd_register_dumps): Declare.
26401 (LANG_HOOKS_REGISTER_DUMPS): Define.
26402 (LANG_HOOKS_INITIALIZER): Add it.
26403 * langhooks.c (lhd_register_dumps): Define.
26404 * langhooks.h (struct lang_hooks): Add register_dumps.
26406 2017-05-19 Nathan Sidwell <nathan@acm.org>
26408 * context.h (context::set_passes): New.
26409 * context.c (context::context): Do not create pass manager.
26410 * toplev.c (general_init): Create pass manager here.
26412 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
26414 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
26415 use this splitter if two add or or instructions would also work for
26416 the constant we want to generate.
26418 2017-05-19 Richard Biener <rguenther@suse.de>
26421 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
26422 predicate evaluation.
26424 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26426 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
26428 * ipa-inline.c (want_inline_small_function_p): Do not cast to
26431 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26433 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26434 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26435 (inline_edge_summary_vec): Turn into ...
26436 (ipa_call_summaries): ... this one.
26437 (redirect_to_unreachable, edge_set_predicate,
26438 evaluate_properties_for_edge, inline_summary_alloc,
26439 reset_ipa_call_summary, reset_inline_summary,
26440 inline_summary_t::duplicate): Update.
26441 (inline_edge_duplication_hook): Turn to ...
26442 (ipa_call_summary_t::duplicate): ... this one.
26443 (inline_edge_removal_hook): Turn to ...
26444 (ipa_call_summary_t::remove): ... this one.
26445 (dump_inline_edge_summary): Turn to ...
26446 (dump_ipa_call_summary): ... this one.
26447 (estimate_function_body_sizes): Update.
26448 (inline_update_callee_summaries): Update.
26449 (remap_edge_change_prob): Update.
26450 (remap_edge_summaries): Update.
26451 (inline_merge_summary): Update.
26452 (do_estimate_edge_time): Update.
26453 (inline_generate_summary): Update.
26454 (inline_read_section): Update.
26455 (inline_read_summary): Update.
26456 (inline_free_summary): Update.
26457 * ipa-inline.c (can_inline_edge_p): Update.
26458 (compute_inlined_call_time): Update.
26459 (want_inline_small_function_p): Update.
26460 (edge_badness): Update.
26461 (early_inliner): Update.
26462 * ipa-inline.h (inline_edge_summary): Turn to ...
26463 (ipa_call_summary): ... this one.
26464 (ipa_call_summary_t): New class.
26465 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26466 (ipa_call_summaries): New.
26467 (inline_edge_summary): Remove.
26468 (estimate_edge_growth): Update.
26469 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26470 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26471 * ipa-split.c (execute_split_functions): Update.
26472 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26474 2017-05-19 Richard Biener <rguenther@suse.de>
26476 PR middle-end/80764
26477 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
26479 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
26481 * config/rs6000/rs6000.c (struct machine_function): Add field
26482 fpr_is_wrapped_separately.
26483 (rs6000_get_separate_components): Use 64 components. Handle the
26484 new FPR components.
26485 (rs6000_components_for_bb): Handle the FPR components.
26486 (rs6000_emit_prologue_components): Handle the FPR components.
26487 (rs6000_emit_epilogue_components): Handle the FPR components.
26488 (rs6000_set_handled_components): Handle the FPR components.
26489 (rs6000_emit_prologue): Don't output prologue code for those FPRs
26490 that are already separately shrink-wrapped.
26491 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
26492 that are already separately shrink-wrapped.
26494 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26497 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
26500 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
26501 (define_peephole2 for Altivec d-form load): Add peepholes to catch
26502 cases where the register allocator uses a move and an offsettable
26503 memory operation to/from a FPR register on ISA 2.06/2.07.
26504 (define_peephole2 for Altivec d-form store): Likewise.
26506 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
26509 * config/i386/mmx.md (*mov<mode>_internal): Enable
26510 alternatives 11, 12, 13 and 14 also for 32bit targets.
26511 Remove alternatives 15, 16, 17 and 18.
26512 * config/i386/sse.md (vec_concatv2di): Change
26513 alternative (!x, *y) to (x, ?!*Yn).
26515 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
26517 * dumpfile.h (enum dump_kind): Remove stray comma.
26519 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26521 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
26522 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
26523 predicate::num_conditions
26524 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
26525 (CHANGED): turn into predicate::changed.
26526 (agg_position_info): Move to ipa-predicate.h
26527 (add_condition, predicate::add_clause, predicate::operator &=,
26528 predicate::or_with, predicate::evaluate, predicate::probability,
26529 dump_condition, dump_clause, predicate::dump,
26530 predicate::remap_after_duplication, predicate::remap_after_inlining,
26531 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
26532 (evaluate_conditions_for_known_args): Update.
26533 (set_cond_stmt_execution_predicate): Update.
26534 * ipa-inline.h: Include ipa-predicate.h
26535 (condition, inline_param_summary, conditions, agg_position_info,
26536 predicate): Move to ipa-predicate.h
26537 * ipa-predicate.c: New file.
26538 * ipa-predicate.h: New file.
26540 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
26542 * final.c (leaf_function_p): Check we are not in a sequence.
26544 2017-05-18 Martin Liska <mliska@suse.cz>
26546 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
26547 * dumpfile.c (dump_register): Use new enum dump_kind.
26548 (get_dump_file_name): Likewise.
26549 (dump_enable_all): Likewise.
26550 (dump_switch_p_1): Likewise.
26551 (enable_rtl_dump_file): Remove usage of TDF_RTL.
26552 * dumpfile.h (enum dump_kind): New enum type.
26553 (struct dump_file_info): Create constructor and
26554 format fields and comments.
26555 * passes.c (pass_manager::register_one_dump_file):
26557 * statistics.c (statistics_early_init): Likewise.
26558 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
26559 TDF_TREE with TDF_SLIM.
26560 (gather_memory_references_ref): Likewise.
26562 2017-05-18 Martin Liska <mliska@suse.cz>
26564 * vec.h (struct vnull): Use it.
26566 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26568 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
26569 (true_predicate, false_predicate, true_predicate_p,
26570 false_predicate_p): Remove.
26571 (single_cond_predicate, not_inlined_predicate): Turn to member function
26573 (add_condition): Update.
26574 (add_clause): Turn to...
26575 (predicate::add_clause): ... this one; update; allow passing NULL
26577 (and_predicates): Turn to ...
26578 (predicate::operator &=): ... this one.
26579 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
26580 (or_predicates): Turn to ...
26581 (predicate::or_with): ... this one.
26582 (evaluate_predicate): Turn to ...
26583 (predicate::evaluate): ... this one.
26584 (predicate_probability): Turn to ...
26585 (predicate::probability): ... this one.
26586 (dump_condition): Update.
26587 (dump_predicate): Turn to ...
26588 (predicate::dump): ... this one.
26589 (account_size_time): Update.
26590 (edge_set_predicate): Update.
26591 (set_hint_predicate): UPdate.
26592 (evaluate_conditions_for_known_args): Update.
26593 (evaluate_properties_for_edge): Update.
26594 (remap_predicate_after_duplication): Turn to...
26595 (predicate::remap_after_duplication): ... this one.
26596 (remap_hint_predicate_after_duplication): Update.
26597 (inline_summary_t::duplicate): UPdate.
26598 (dump_inline_edge_summary): Update.
26599 (dump_inline_summary): Update.
26600 (set_cond_stmt_execution_predicate): Update.
26601 (set_switch_stmt_execution_predicate): Update.
26602 (compute_bb_predicates): Update.
26603 (will_be_nonconstant_expr_predicate): Update.
26604 (will_be_nonconstant_predicate): Update.
26605 (phi_result_unknown_predicate): Update.
26606 (predicate_for_phi_result): Update.
26607 (array_index_predicate): Update.
26608 (estimate_function_body_sizes): Update.
26609 (estimate_node_size_and_time): Update.
26610 (estimate_ipcp_clone_size_and_time): Update.
26611 (remap_predicate): Rename to ...
26612 (predicate::remap_after_inlining): ... this one.
26613 (remap_hint_predicate): Update.
26614 (inline_merge_summary): Update.
26615 (inline_update_overall_summary): Update.
26616 (estimate_size_after_inlining): Update.
26617 (read_predicate): Rename to ...
26618 (predicate::stream_in): ... this one.
26619 (read_inline_edge_summary): Update.
26620 (write_predicate): Rename to ...
26621 (predicate::stream_out): ... this one.
26622 (write_inline_edge_summary): Update.
26623 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
26624 (clause_t): Turn to uint32_t
26625 (predicate): Turn to class; implement constructor and operators
26627 (size_time_entry): Update.
26628 (inline_summary): Update.
26629 (inline_edge_summary): Update.
26631 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
26633 * fold-const.c (fold_binary_loc): Move transformation...
26634 * match.pd (C - X CMP X): ... here.
26636 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
26638 * config/sparc/sparc.c (sparc_option_override): Set function
26639 alignment for -mcpu=niagara7 to 64 to match the I$ line.
26640 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
26642 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
26644 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
26646 2017-05-18 Marek Polacek <polacek@redhat.com>
26649 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
26650 (pass_ubsan::execute): Call gimple_assign_single_p instead of
26651 gimple_assign_load_p.
26653 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
26655 PR middle-end/80692
26656 * real.c (do_compare): Give decimal_do_compare preference over
26657 comparing just the signs.
26659 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
26661 * doc/md.texi (Canonicalization of Instructions): Describe the
26662 canonical form of instructions that inherently set a condition
26665 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
26667 PR middle-end/80775
26668 * tree-cfg.c: Move deletion of unreachable case statements to after
26669 the merging of consecutive case labels.
26671 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
26673 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
26674 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
26675 restoring of callee-saved registers.
26677 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
26679 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
26680 * config/visium/visium.c (single_set_and_flags): Likewise.
26681 * config/visium/visium.md (Substitutions): Likewise.
26683 2017-05-17 Martin Liska <mliska@suse.cz>
26685 * cfg.c: Introduce dump_flags_t type and
26686 use it instead of int type.
26688 * cfghooks.c: Likewise.
26689 * cfghooks.h (struct cfg_hooks): Likewise.
26690 * cfgrtl.c: Likewise.
26691 * cfgrtl.h: Likewise.
26692 * cgraph.c (cgraph_node::get_body): Likewise.
26693 * coretypes.h: Likewise.
26694 * domwalk.c: Likewise.
26695 * domwalk.h: Likewise.
26696 * dumpfile.c (struct dump_option_value_info): Likewise.
26697 (dump_enable_all): Likewise.
26698 (dump_switch_p_1): Likewise.
26699 (opt_info_switch_p): Likewise.
26700 * dumpfile.h (enum tree_dump_index): Likewise.
26701 (struct dump_file_info): Likewise.
26702 * genemit.c: Likewise.
26703 * generic-match-head.c: Likewise.
26704 * gengtype.c (open_base_files): Likewise.
26705 * gimple-pretty-print.c: Likewise.
26706 * gimple-pretty-print.h: Likewise.
26707 * graph.c (print_graph_cfg): Likewise.
26708 * graphite-scop-detection.c (dot_all_sese): Likewise.
26709 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
26710 * loop-unroll.c (report_unroll): Likewise.
26711 * passes.c (pass_manager::register_one_dump_file): Likewise.
26712 * print-tree.c: Likewise.
26713 * statistics.c: Likewise.
26714 * tree-cfg.c: Likewise.
26715 * tree-cfg.h: Likewise.
26716 * tree-dfa.c: Likewise.
26717 * tree-dfa.h: Likewise.
26718 * tree-dump.c (dump_function): Likewise.
26719 * tree-dump.h (struct dump_info): Likewise.
26720 * tree-pretty-print.c: Likewise.
26721 * tree-pretty-print.h: Likewise.
26722 * tree-ssa-live.c: Likewise.
26723 * tree-ssa-live.h: Likewise.
26724 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
26725 * tree-vect-loop.c: Likewise.
26726 * tree-vect-slp.c: Likewise.
26728 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
26729 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26731 PR tree-optimization/80457
26732 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
26733 of all arguments to a statement as scalar_to_vec operations.
26734 (vectorizable_call): Adjust call to vect_model_simple_cost for
26736 (vectorizable_conversion): Likewise.
26737 (vectorizable_assignment): Likewise.
26738 (vectorizable_shift): Likewise.
26739 (vectorizable_operation): Likewise.
26740 (vectorizable_comparison): Likewise.
26741 (vect_is_simple_cond): Record the def types for operands.
26742 (vectorizable_condition): Likewise, call vect_model_simple_cost.
26743 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
26744 for statement argument count.
26746 2017-05-16 Carl Love <cel@us.ibm.com>
26748 * config/rs6000/rs6000-c: Add support for built-in functions
26749 vector unsigned long long vec_bperm (vector unsigned long long,
26750 vector unsigned char)
26751 vector signed long long vec_mule (vector signed int,
26753 vector unsigned long long vec_mule (vector unsigned int,
26754 vector unsigned int)
26755 vector signed long long vec_mulo (vector signed int,
26757 vector unsigned long long vec_mulo (vector unsigned int,
26758 vector unsigned int)
26759 vector signed char vec_sldw (vector signed char,
26760 vector signed char,
26762 vector unsigned char vec_sldw (vector unsigned char,
26763 vector unsigned char,
26765 vector signed short vec_sldw (vector signed short,
26766 vector signed short,
26768 vector unsigned short vec_sldw (vector unsigned short,
26769 vector unsigned short,
26771 vector signed int vec_sldw (vector signed int,
26774 vector unsigned int vec_sldw (vector unsigned int,
26775 vector unsigned int,
26777 vector signed long long vec_sldw (vector signed long long,
26778 vector signed long long,
26780 vector unsigned long long vec_sldw (vector unsigned long long,
26781 vector unsigned long long,
26783 * config/rs6000/rs6000-c: Add support for built-in functions
26784 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
26785 * config/rs6000/altivec.h: Add defintion for vec_sldw.
26786 * doc/extend.texi: Update the built-in documentation for the
26787 new built-in functions.
26789 2017-05-16 Marek Polacek <polacek@redhat.com>
26793 * tree.c (save_expr): Don't fold the expression.
26795 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
26797 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
26798 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
26799 and (?*y,m). Update insn attributes.
26801 2017-05-16 Martin Liska <mliska@suse.cz>
26803 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
26804 flags argument of print_gimple_stmt, print_gimple_expr,
26805 print_generic_stmt and print_generic_expr.
26806 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
26807 * coretypes.h: Likewise.
26808 * except.c (dump_eh_tree): Likewise.
26809 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
26810 * gimple-pretty-print.h: Likewise.
26811 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
26812 (backprop::push_to_worklist): Likewise.
26813 (backprop::pop_from_worklist): Likewise.
26814 (backprop::process_use): Likewise.
26815 (backprop::intersect_uses): Likewise.
26816 (note_replacement): Likewise.
26817 * gimple-ssa-store-merging.c
26818 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
26819 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
26820 (pass_store_merging::execute): Likewise.
26821 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
26822 (ssa_base_cand_dump_callback): Likewise.
26823 (dump_incr_vec): Likewise.
26824 (replace_refs): Likewise.
26825 (replace_mult_candidate): Likewise.
26826 (create_add_on_incoming_edge): Likewise.
26827 (create_phi_basis): Likewise.
26828 (insert_initializers): Likewise.
26829 (all_phi_incrs_profitable): Likewise.
26830 (introduce_cast_before_cand): Likewise.
26831 (replace_one_candidate): Likewise.
26832 * gimplify.c (gimplify_expr): Likewise.
26833 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
26834 (set_rename): Likewise.
26835 (rename_uses): Likewise.
26836 (copy_loop_phi_nodes): Likewise.
26837 (add_close_phis_to_merge_points): Likewise.
26838 (copy_loop_close_phi_args): Likewise.
26839 (copy_cond_phi_args): Likewise.
26840 (graphite_copy_stmts_from_block): Likewise.
26841 (translate_pending_phi_nodes): Likewise.
26842 * graphite-poly.c (print_pdr): Likewise.
26843 (dump_gbb_cases): Likewise.
26844 (dump_gbb_conditions): Likewise.
26845 (print_scop_params): Likewise.
26846 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
26847 (build_cross_bb_scalars_use): Likewise.
26848 (gather_bbs::before_dom_children): Likewise.
26849 * hsa-dump.c (dump_hsa_immed): Likewise.
26850 * ipa-cp.c (print_ipcp_constant_value): Likewise.
26851 (get_replacement_map): Likewise.
26852 * ipa-inline-analysis.c (dump_condition): Likewise.
26853 (estimate_function_body_sizes): Likewise.
26854 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
26855 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
26856 * ipa-prop.c (ipa_dump_param): Likewise.
26857 (ipa_print_node_jump_functions_for_edge): Likewise.
26858 (ipa_modify_call_arguments): Likewise.
26859 (ipa_modify_expr): Likewise.
26860 (ipa_dump_param_adjustments): Likewise.
26861 (ipa_dump_agg_replacement_values): Likewise.
26862 (ipcp_modif_dom_walker::before_dom_children): Likewise.
26863 * ipa-pure-const.c (check_stmt): Likewise.
26864 (pass_nothrow::execute): Likewise.
26865 * ipa-split.c (execute_split_functions): Likewise.
26866 * omp-offload.c (dump_oacc_loop_part): Likewise.
26867 (dump_oacc_loop): Likewise.
26868 * trans-mem.c (tm_log_emit): Likewise.
26869 (tm_memopt_accumulate_memops): Likewise.
26870 (dump_tm_memopt_set): Likewise.
26871 (dump_tm_memopt_transform): Likewise.
26872 * tree-cfg.c (gimple_verify_flow_info): Likewise.
26873 (print_loop): Likewise.
26874 * tree-chkp-opt.c (chkp_print_addr): Likewise.
26875 (chkp_gather_checks_info): Likewise.
26876 (chkp_get_check_result): Likewise.
26877 (chkp_remove_check_if_pass): Likewise.
26878 (chkp_use_outer_bounds_if_possible): Likewise.
26879 (chkp_reduce_bounds_lifetime): Likewise.
26880 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
26881 (chkp_mark_completed_bounds): Likewise.
26882 (chkp_register_incomplete_bounds): Likewise.
26883 (chkp_mark_invalid_bounds): Likewise.
26884 (chkp_maybe_copy_and_register_bounds): Likewise.
26885 (chkp_build_returned_bound): Likewise.
26886 (chkp_get_bound_for_parm): Likewise.
26887 (chkp_build_bndldx): Likewise.
26888 (chkp_get_bounds_by_definition): Likewise.
26889 (chkp_generate_extern_var_bounds): Likewise.
26890 (chkp_get_bounds_for_decl_addr): Likewise.
26891 * tree-chrec.c (chrec_apply): Likewise.
26892 * tree-data-ref.c (dump_data_reference): Likewise.
26893 (dump_subscript): Likewise.
26894 (dump_data_dependence_relation): Likewise.
26895 (analyze_overlapping_iterations): Likewise.
26896 * tree-inline.c (expand_call_inline): Likewise.
26897 (tree_function_versioning): Likewise.
26898 * tree-into-ssa.c (dump_defs_stack): Likewise.
26899 (dump_currdefs): Likewise.
26900 (dump_names_replaced_by): Likewise.
26901 (dump_update_ssa): Likewise.
26902 (update_ssa): Likewise.
26903 * tree-object-size.c (pass_object_sizes::execute): Likewise.
26904 * tree-parloops.c (build_new_reduction): Likewise.
26905 (try_create_reduction_list): Likewise.
26906 (ref_conflicts_with_region): Likewise.
26907 (oacc_entry_exit_ok_1): Likewise.
26908 (oacc_entry_exit_single_gang): Likewise.
26909 * tree-pretty-print.h: Likewise.
26910 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
26911 (get_scalar_evolution): Likewise.
26912 (add_to_evolution): Likewise.
26913 (get_loop_exit_condition): Likewise.
26914 (analyze_evolution_in_loop): Likewise.
26915 (analyze_initial_condition): Likewise.
26916 (analyze_scalar_evolution): Likewise.
26917 (instantiate_scev): Likewise.
26918 (number_of_latch_executions): Likewise.
26919 (gather_chrec_stats): Likewise.
26920 (final_value_replacement_loop): Likewise.
26921 (scev_const_prop): Likewise.
26922 * tree-sra.c (dump_access): Likewise.
26923 (disqualify_candidate): Likewise.
26924 (create_access): Likewise.
26925 (reject): Likewise.
26926 (maybe_add_sra_candidate): Likewise.
26927 (create_access_replacement): Likewise.
26928 (analyze_access_subtree): Likewise.
26929 (analyze_all_variable_accesses): Likewise.
26930 (sra_modify_assign): Likewise.
26931 (initialize_constant_pool_replacements): Likewise.
26932 (find_param_candidates): Likewise.
26933 (decide_one_param_reduction): Likewise.
26934 (replace_removed_params_ssa_names): Likewise.
26935 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
26936 * tree-ssa-copy.c (dump_copy_of): Likewise.
26937 (copy_prop_visit_cond_stmt): Likewise.
26938 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
26939 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
26940 (record_equivalences_from_stmt): Likewise.
26941 * tree-ssa-dse.c (compute_trims): Likewise.
26942 (delete_dead_call): Likewise.
26943 (delete_dead_assignment): Likewise.
26944 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
26945 (forward_propagate_into_cond): Likewise.
26946 (pass_forwprop::execute): Likewise.
26947 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
26948 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
26950 (move_computations_worker): Likewise.
26951 (execute_sm): Likewise.
26952 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
26953 (remove_exits_and_undefined_stmts): Likewise.
26954 (remove_redundant_iv_tests): Likewise.
26955 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
26956 (adjust_iv_update_pos): Likewise.
26957 * tree-ssa-math-opts.c (bswap_replace): Likewise.
26958 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
26959 (value_replacement): Likewise.
26960 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
26961 * tree-ssa-pre.c (print_pre_expr): Likewise.
26962 (get_representative_for): Likewise.
26963 (create_expression_by_pieces): Likewise.
26964 (insert_into_preds_of_block): Likewise.
26965 (eliminate_insert): Likewise.
26966 (eliminate_dom_walker::before_dom_children): Likewise.
26967 (eliminate): Likewise.
26968 (remove_dead_inserted_code): Likewise.
26969 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
26970 * tree-ssa-reassoc.c (get_rank): Likewise.
26971 (eliminate_duplicate_pair): Likewise.
26972 (eliminate_plus_minus_pair): Likewise.
26973 (eliminate_not_pairs): Likewise.
26974 (undistribute_ops_list): Likewise.
26975 (eliminate_redundant_comparison): Likewise.
26976 (update_range_test): Likewise.
26977 (optimize_range_tests_var_bound): Likewise.
26978 (optimize_vec_cond_expr): Likewise.
26979 (rewrite_expr_tree): Likewise.
26980 (rewrite_expr_tree_parallel): Likewise.
26981 (linearize_expr): Likewise.
26982 (break_up_subtract): Likewise.
26983 (linearize_expr_tree): Likewise.
26984 (attempt_builtin_powi): Likewise.
26985 (attempt_builtin_copysign): Likewise.
26986 (transform_stmt_to_copy): Likewise.
26987 (transform_stmt_to_multiply): Likewise.
26988 (dump_ops_vector): Likewise.
26989 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
26990 (print_scc): Likewise.
26991 (set_ssa_val_to): Likewise.
26992 (visit_reference_op_store): Likewise.
26993 (visit_use): Likewise.
26994 (sccvn_dom_walker::before_dom_children): Likewise.
26995 (run_scc_vn): Likewise.
26996 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
26998 (expr_hash_elt::print): Likewise.
26999 (const_and_copies::pop_to_marker): Likewise.
27000 (const_and_copies::record_const_or_copy_raw): Likewise.
27001 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27002 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27003 (dump_predicates): Likewise.
27004 (find_uninit_use): Likewise.
27005 (warn_uninitialized_phi): Likewise.
27006 (pass_late_warn_uninitialized::execute): Likewise.
27007 * tree-ssa.c (verify_vssa): Likewise.
27008 (verify_ssa): Likewise.
27009 (maybe_optimize_var): Likewise.
27010 * tree-vrp.c (dump_value_range): Likewise.
27011 (dump_all_value_ranges): Likewise.
27012 (dump_asserts_for): Likewise.
27013 (register_edge_assert_for_2): Likewise.
27014 (vrp_visit_cond_stmt): Likewise.
27015 (vrp_visit_switch_stmt): Likewise.
27016 (vrp_visit_stmt): Likewise.
27017 (vrp_visit_phi_node): Likewise.
27018 (simplify_cond_using_ranges_1): Likewise.
27019 (fold_predicate_in): Likewise.
27020 (evrp_dom_walker::before_dom_children): Likewise.
27021 (evrp_dom_walker::push_value_range): Likewise.
27022 (evrp_dom_walker::pop_value_range): Likewise.
27023 (execute_early_vrp): Likewise.
27025 2017-05-16 Richard Biener <rguenther@suse.de>
27027 * dwarf2out.c (loc_list_from_tree_1): Do not create
27028 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27030 2017-05-16 Richard Biener <rguenther@suse.de>
27032 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27034 (note_variable_value_in_expr): If we resolved the decl ref
27035 do not push to the stack.
27037 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27039 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27040 operations in fast-math mode.
27041 (vaddq_f16): Likewise.
27042 (vmul_f16): Likewise.
27043 (vmulq_f16): Likewise.
27044 (vsub_f16): Likewise.
27045 (vsubq_f16): Likewise.
27046 * config/arm/neon.md (add<mode>3): New.
27048 (fma:<VH:mode>3): New. Also remove outdated comment.
27051 2017-05-16 Martin Liska <mliska@suse.cz>
27055 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27056 (odr_types_equivalent_p): Likewise.
27058 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
27060 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
27062 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
27065 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
27066 non-interunit SSE move alternatives with '?'.
27067 (zero-extendsidi peephole2): New peephole to skip intermediate
27068 general register in SSE zero-extend sequence.
27070 2017-05-15 Jeff Law <law@redhat.com>
27072 * reorg.c (relax_delay_slots): Create a new variable to hold
27073 the temporary target rather than clobbering TARGET_LABEL.
27075 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
27076 missing argument to extract_bit_field call.
27077 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
27079 2017-05-15 Martin Liska <mliska@suse.cz>
27082 * gcc.c (process_command): Do not allow empty argument of -o option.
27084 2017-05-15 Renlin Li <renlin.li@arm.com>
27086 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
27087 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
27088 * config/aarch64/constraints.md (Usf): Add long call check.
27089 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
27090 (call_value): Likewise.
27091 (sibcall): Likewise.
27092 (sibcall_value): Likewise.
27094 (call_value_insn): New.
27095 (sibcall_insn): Update rtx pattern.
27096 (sibcall_value_insn): Likewise.
27097 (call_internal): Remove.
27098 (call_value_internal): Likewise.
27099 (sibcall_internal): Likewise.
27100 (sibcall_value_internal): Likewise.
27101 (call_reg): Likewise.
27102 (call_symbol): Likewise.
27103 (call_value_reg): Likewise.
27104 (call_value_symbol): Likewise.
27106 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
27109 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
27111 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
27113 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
27114 compatible with CCGOCmode and with CCZmode.
27116 2017-05-14 Martin Sebor <msebor@redhat.com>
27118 PR middle-end/77671
27119 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
27120 (gimple_fold_builtin_snprintf): Same.
27121 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
27122 (gimple_fold_builtin_snprintf): Same.
27123 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
27124 of character types.
27125 (is_call_safe): New function.
27126 (try_substitute_return_value): Call it.
27127 (try_simplify_call): New function.
27128 (pass_sprintf_length::handle_gimple_call): Call it.
27130 2017-05-14 Martin Sebor <msebor@redhat.com>
27132 PR middle-end/80669
27133 * builtins.c (expand_builtin_stpncpy): Simplify.
27135 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
27137 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
27138 * config/i386/i386.h
27139 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
27140 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
27141 (struct machine_function): Add new members call_ms2sysv,
27142 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
27143 (struct machine_frame_state): New fields sp_realigned and
27144 sp_realigned_offset.
27145 * config/i386/i386.c
27146 (enum xlogue_stub): New enum.
27147 (enum xlogue_stub_sets): New enum.
27148 (class xlogue_layout): New class.
27149 (struct ix86_frame): New fields stack_realign_allocate_offset,
27150 stack_realign_offset and outlined_save_offset. Modify comments to
27151 detail stack layout when using out-of-line stubs.
27152 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
27153 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
27154 -mcall-ms2sysv-xlogues.
27155 (stub_managed_regs): New static variable.
27156 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
27157 registers managed by out-of-line stub.
27158 (disable_call_ms2sysv_xlogues): New function.
27159 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
27160 m->call_ms2sysv when appropriate and compute frame layout for
27162 (sp_valid_at, fp_valid_at): New inline functions.
27163 (choose_basereg): New function.
27164 (choose_baseaddr): Add align parameter, use choose_basereg and modify
27166 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
27167 Use align parameter of choose_baseaddr to generated aligned SSE movs
27169 (pro_epilogue_adjust_stack): Modify to track
27170 machine_frame_state::sp_realigned.
27171 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
27172 (ix86_nsaved_sseregs): Likewise.
27173 (ix86_emit_save_regs): Likewise.
27174 (ix86_emit_save_regs_using_mov): Likewise.
27175 (ix86_emit_save_sse_regs_using_mov): Likewise.
27176 (get_scratch_register_on_entry): Likewise.
27177 (gen_frame_set): New function.
27178 (gen_frame_load): Likewise.
27179 (gen_frame_store): Likewise.
27180 (emit_outlined_ms2sysv_save): Likewise.
27181 (emit_outlined_ms2sysv_restore): Likewise.
27182 (ix86_expand_prologue): Modify stack re-alignment code and call
27183 emit_outlined_ms2sysv_save when appropriate.
27184 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
27185 parameter rtx_insn *insn, which allows the function to be used to only
27186 generate the notes.
27187 (ix86_expand_epilogue): Modify validity checks of frame and stack
27188 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
27189 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
27190 * config/i386/predicates.md
27191 (save_multiple): New predicate.
27192 (restore_multiple): Likewise.
27193 * config/i386/sse.md
27194 (save_multiple<mode>): New pattern.
27195 (save_multiple_realign<mode>): Likewise.
27196 (restore_multiple<mode>): Likewise.
27197 (restore_multiple_and_return<mode>): Likewise.
27198 (restore_multiple_leave_return<mode>): Likewise.
27199 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
27201 2017-05-14 Julia Koval <julia.koval@intel.com>
27203 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
27204 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
27205 (__builtin_ia32_xsetbv): New builtins.
27206 * config/i386/i386.c (ix86_expand_special_args_builtin):
27208 (ix86_expand_builtin): Special expand for new intrinsics.
27209 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
27210 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
27211 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
27213 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27215 * cfganal.c (inverted_post_order_compute): Change argument type
27217 * cfganal.h (inverted_post_order_compute): Adjust prototype.
27218 * df-core.c (rest_of_handle_df_initialize): Adjust.
27219 (rest_of_handle_df_finish): Likewise.
27220 (df_analyze_1): Likewise.
27221 (df_analyze): Likewise.
27222 (loop_inverted_post_order_compute): Change argument to be a vec *.
27223 (df_analyze_loop): Adjust.
27224 (df_get_n_blocks): Likewise.
27225 (df_get_postorder): Likewise.
27226 * df.h (struct df_d): Change field to be a vec.
27227 * lcm.c (compute_laterin): Adjust.
27228 (compute_available): Likewise.
27229 * lra-lives.c (lra_create_live_ranges_1): Likewise.
27230 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
27231 * tree-ssa-pre.c (compute_antic): Likewise.
27233 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27235 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
27236 (depth_first_search::depth_first_search): Change structure init
27237 function to this constructor.
27238 (depth_first_search::add_bb): Rename function to this member.
27239 (depth_first_search::execute): Likewise.
27240 (flow_dfs_compute_reverse_finish): Adjust.
27242 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27244 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
27245 (longest_simple_path): Likewise.
27246 * shrink-wrap.c (spread_components): Likewise.
27247 (disqualify_problematic_components): Likewise.
27248 (emit_common_heads_for_components): Likewise.
27249 (emit_common_tails_for_components): Likewise.
27250 (insert_prologue_epilogue_for_components): Likewise.
27252 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27254 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
27257 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27259 * df-core.c (df_set_blocks): Start using auto_bitmap.
27260 (df_compact_blocks): Likewise.
27261 * df-problems.c (df_rd_confluence_n): Likewise.
27262 * df-scan.c (df_insn_rescan_all): Likewise.
27263 (df_process_deferred_rescans): Likewise.
27264 (df_update_entry_block_defs): Likewise.
27265 (df_update_exit_block_uses): Likewise.
27266 (df_entry_block_bitmap_verify): Likewise.
27267 (df_exit_block_bitmap_verify): Likewise.
27268 (df_scan_verify): Likewise.
27269 * lra-constraints.c (lra_constraints): Likewise.
27270 (undo_optional_reloads): Likewise.
27271 (lra_undo_inheritance): Likewise.
27272 * lra-remat.c (calculate_gen_cands): Likewise.
27273 (do_remat): Likewise.
27274 * lra-spills.c (assign_spill_hard_regs): Likewise.
27275 (spill_pseudos): Likewise.
27276 * tree-ssa-pre.c (bitmap_set_and): Likewise.
27277 (bitmap_set_subtract_values): Likewise.
27279 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27281 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
27282 management with auto_bitmap.
27283 (fix_inter_tick): Likewise.
27284 (fix_recovery_deps): Likewise.
27285 * ira.c (add_store_equivs): Likewise.
27286 (find_moveable_pseudos): Likewise.
27287 (split_live_ranges_for_shrink_wrap): Likewise.
27288 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
27289 (rtx_reuse_manager::seen_def_p): Likewise.
27290 (rtx_reuse_manager::set_seen_def): Likewise.
27291 * print-rtl.h (class rtx_reuse_manager): Likewise.
27293 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27295 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
27297 (migrate_btr_def): Likewise.
27298 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
27299 * df-core.c (loop_post_order_compute): Likewise.
27300 (loop_inverted_post_order_compute): Likewise.
27301 * hsa-common.h: Likewise.
27302 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
27303 * init-regs.c (initialize_uninitialized_regs): Likewise.
27304 * ipa-inline.c (resolve_noninline_speculation): Likewise.
27305 (inline_small_functions): Likewise.
27306 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
27307 * ira.c (combine_and_move_insns): Likewise.
27308 (build_insn_chain): Likewise.
27309 * loop-invariant.c (find_invariants): Likewise.
27310 * lower-subreg.c (propagate_pseudo_copies): Likewise.
27311 * predict.c (tree_predict_by_opcode): Likewise.
27312 (predict_paths_leading_to): Likewise.
27313 (predict_paths_leading_to_edge): Likewise.
27314 (estimate_loops_at_level): Likewise.
27315 (estimate_loops): Likewise.
27316 * shrink-wrap.c (try_shrink_wrapping): Likewise.
27317 (spread_components): Likewise.
27318 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
27319 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
27320 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
27321 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
27322 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
27323 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
27324 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
27325 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
27326 (mark_threaded_blocks): Likewise.
27327 (thread_through_all_blocks): Likewise.
27328 * tree-ssa.c (verify_ssa): Likewise.
27329 (execute_update_addresses_taken): Likewise.
27330 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
27332 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27334 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
27336 (post_order_compute): Likewise.
27337 (inverted_post_order_compute): Likewise.
27338 (pre_and_rev_post_order_compute_fn): Likewise.
27340 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27342 * genrecog.c (int_set::int_set): Explicitly construct our
27343 auto_vec base class.
27344 * vec.h (auto_vec::auto_vec): New constructor.
27346 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27348 * bitmap.h (class auto_bitmap): New constructor taking
27349 bitmap_obstack * argument.
27351 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27353 * bitmap.h (class auto_bitmap): Change type of m_bits to
27354 bitmap_head, and adjust ctor / dtor and member operators.
27356 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
27358 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
27359 when returned register mode doesn't match original mode.
27361 2017-05-12 Jeff Law <law@redhat.com>
27362 Jakub Jelinek <jakub@redhat.com>
27364 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
27365 we look for cc setter after the compare-elim changes.
27366 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
27367 within the vector to match what compare-elim now expects.
27368 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
27369 (xorsi3_flags, one_cmplsi2_flags): Likewise.
27371 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
27372 after the compare-elim changes.
27373 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
27374 the vector to match what compare-elim now expects.
27375 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
27376 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
27377 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
27378 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
27379 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
27381 * config/visium/visium.c (single_set_and_flags): Fix where
27382 we look for cc setter after the compare-elim changes.
27383 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
27384 with the vector to match what compare-elim now expects.
27385 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
27386 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
27387 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
27388 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
27389 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
27390 (neg<mode>2_insn_set_overflow): Likewise.
27392 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
27394 PR middle-end/79794
27395 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
27396 maybe_expand_insn call, set ops[0].target. If still set after call,
27397 set alt_rtl. Add extra arg to recursive calls.
27398 (extract_bit_field): Add alt_rtl argument. Pass to
27400 * expmed.h (extract_bit_field): Fix prototype.
27401 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
27402 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
27403 to extract_bit_field_calls.
27404 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
27405 Pass alt_rtl to extract_bit_field calls.
27406 * calls.c (store_unaligned_arguments_into_psuedos)
27407 load_register_parameters): Pass extra NULL to extract_bit_field calls.
27408 * optabs.c (maybe_legitimize_operand): Clear op->target when call
27410 * optabs.h (struct expand_operand): Add target bitfield.
27412 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27414 * compare-elim.c (try_eliminate_compare): Canonicalize
27415 operation with embedded compare to
27416 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
27417 (set (reg) (operation)].
27419 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
27421 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27424 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
27425 cost of adding a carry flag for ADC instruction.
27426 [case MINUS]: Ignore the cost of subtracting a carry flag
27427 for SBB instruction.
27429 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
27431 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
27433 * config/rs6000/bmiintrin.h: New file.
27434 * config/rs6000/bmi2intrin.h: New file.
27435 * config/rs6000/x86intrin.h: New file.
27437 2017-05-12 Jeff Law <law@redhat.com>
27439 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
27442 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
27444 PR middle-end/80707
27445 * tree-cfg.c: Remove cfg edges of unreachable case statements.
27447 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27449 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27450 early expansion of vector divide builtins.
27451 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
27452 builtins identified as having unsigned arguments.
27454 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27456 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
27457 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
27458 expansion of vector logical operations (and, andc, or, xor,
27461 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27463 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
27464 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
27466 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27468 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27469 early GIMPLE expansion of vector multiplies.
27471 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27473 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
27474 TARGET_HAVE_MOVT conditional.
27475 (movt splitter): Likewise.
27477 2017-05-12 Richard Biener <rguenther@suse.de>
27479 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
27480 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27481 Fold all stmts not inplace.
27483 2017-05-12 Richard Biener <rguenther@suse.de>
27485 PR tree-optimization/80713
27486 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
27487 inserted_exprs bit for not removed stmts.
27489 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
27491 PR middle-end/69921
27492 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
27493 parallelized" attribute for parallelized OpenACC kernels.
27494 * omp-offload.c (execute_oacc_device_lower): Use it.
27496 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
27497 Set "oacc kernels" attribute.
27498 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
27499 parameter. Adjust all users.
27500 (oacc_fn_attrib_kernels_p): Remove function.
27501 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
27502 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
27503 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
27504 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
27505 assert "oacc kernels" attribute is set.
27507 2017-05-11 Carl Love <cel@us.ibm.com>
27509 * config/rs6000/rs6000-c: Add support for built-in functions
27510 vector unsigned char vec_popcnt (vector signed char)
27511 vector unsigned char vec_popcnt (vector unsigned char)
27512 vector unsigned short vec_popcnt (vector signed short)
27513 vector unsigned short vec_popcnt (vector unsigned short)
27514 vector unsigned int vec_popcnt (vector signed int)
27515 vector unsigned int vec_popcnt (vector unsigned int)
27516 vector unsigned long long vec_popcnt (vector signed long long)
27517 vector unsigned long long vec_popcnt (vector unsigned long long)
27518 vector signed long long vec_slo (vector signed long long,
27519 vector signed char)
27520 vector signed long long vec_slo (vector signed long long,
27521 vector unsigned char)
27522 vector unsigned long long vec_slo (vector unsigned long long,
27523 vector signed char)
27524 vector unsigned long long vec_slo (vector unsigned long long,
27525 vector unsigned char)
27526 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
27527 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
27528 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
27529 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
27530 * doc/extend.texi: Update the built-in documentation file for the
27531 new built-in functions.
27533 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
27535 * attribs.h (sorted_attr_string): Move machine independent
27536 functions for target clone support from the i386 port to common
27537 code. Rename ix86_function_versions to common_function_versions.
27538 Rename make_name to make_unique_name.
27539 (common_function_versions): Likewise.
27540 (make_unique_name): Likewise.
27541 (make_dispatcher_decl): Likewise.
27542 (is_function_default_version): Likewise.
27543 * attribs.c (attr_strcmp): Likewise.
27544 (sorted_attr_string): Likewise.
27545 (common_function_versions): Likewise.
27546 (make_unique_name): Likewise.
27547 (make_dispatcher_decl): Likewise.
27548 (is_function_default_version): Likewise.
27549 * config/i386/i386.c (attr_strcmp): Likewise.
27550 (sorted_attr_string): Likewise.
27551 (ix86_function_versions): Likewise.
27552 (make_name): Likewise.
27553 (make_dispatcher_decl): Likewise.
27554 (is_function_default_version): Likewise.
27555 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
27557 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27560 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
27561 Account for direct move costs for vec_construct of integer
27564 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
27567 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
27568 (UNSPEC_STX_ATOMIC): Ditto.
27569 (loaddi_via_sse): New insn.
27570 (storedi_via_sse): Ditto.
27571 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
27572 Update corresponding peephole2 patterns.
27573 (atomic_storedi_fpu): Ditto.
27575 2017-05-11 Julia Koval <julia.koval@intel.com>
27577 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
27578 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
27580 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
27581 (__builtin_ia32_rsqrt14ss_mask): New builtins.
27582 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
27584 2017-05-11 Nathan Sidwell <nathan@acm.org>
27586 * graphite-poly.c: Include dumpfile.h.
27588 * dumpfle.h (dump_function): Declare here ...
27589 * tree-dump.h (dump_function): ... not here.
27590 * dumpfile.c: #include tree-cfg.h.
27591 (dump_function): Move here from ...
27592 * tree-dump.c (dump_function): ... here.
27593 * gimplify.c: #include splay-tree.h, not tree-dump.h.
27594 * graphite-poly.c: Don't include tree-dump.h.
27595 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
27596 * print-tree.c: Likewise.
27597 * stor-layout.c: Likewise.
27598 * tree-nested.c: Likewise.
27600 * dumpfile.c (dump_start): Use TDF_FLAGS.
27601 (dump_enable_all): Fix TDF_KIND check thinko.
27603 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
27605 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27606 array entries to represent two legal parameterizations of the
27607 overloaded __builtin_cmpb function, as represented by the
27608 P6_OV_BUILTIN_CMPB constant.
27609 (altivec_resolve_overloaded_builtin): Add special case handling
27610 for the __builtin_cmpb function, as represented by the
27611 P6_OV_BUILTIN_CMPB constant.
27612 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
27613 (BU_P6_64BIT_2): New macro.
27614 (BU_P6_OVERLOAD_2): New macro
27615 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
27616 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
27617 (CMPB): Add overload support to represent both 32-bit and 64-bit
27618 compare-bytes function.
27619 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
27620 support for TARGET_CMPB.
27621 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
27622 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27623 documentation of the __builtin_cmpb overloaded built-in function.
27625 2017-05-11 Richard Biener <rguenther@suse.de>
27627 PR tree-optimization/80705
27628 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
27629 bases are not vectorizable.
27631 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27633 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
27634 when counting register pressure.
27636 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27638 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
27639 (iv_ca_more_deps): Renamed to ...
27640 (iv_ca_compare_deps): ... this.
27641 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
27643 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27645 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
27647 (determine_group_iv_costs): ... here.
27648 (find_inv_vars_cb): Record inv var if it's not recorded before.
27650 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27652 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
27653 (get_shiftadd_cost): Ditto.
27655 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27657 * tree-ssa-address.c: Include header file.
27658 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
27660 (add_to_parts): Refactor.
27661 (addr_to_parts): New parameter. Update use of move_hint_to_base.
27662 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
27665 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27667 PR tree-optimization/53090
27668 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
27670 (extract_cond_operands): Detect condition with IV on both sides
27671 and return COMP_IV_EXPR_2.
27672 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
27673 (rewrite_use_compare): Simplify by removing call to function
27674 extract_cond_operands.
27676 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27678 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
27679 (extract_cond_operands): Detect condition comparing against non-
27680 invariant bound and return appropriate enum value.
27681 (find_interesting_uses_cond): Update use of extract_cond_operands.
27682 Handle its return value accordingly.
27683 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
27685 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27687 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
27688 nonlinear iv_use computation in loop invariant sensitive way.
27690 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27692 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
27693 (find_iv_candidates): Call relate_compare_use_with_all_cands.
27695 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27697 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
27698 (dump_cand): Support iv_cand.inv_exprs.
27699 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
27701 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
27704 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27706 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
27708 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
27709 as local function. Include necessary header files.
27710 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
27712 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27714 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
27716 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27718 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
27719 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
27720 RSHIFT_EXPR and BIT_NOT_EXPR.
27722 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27724 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
27725 (adjust_setup_cost): New parameter supporting round up adjustment.
27726 (struct address_cost_data): Delete.
27727 (force_expr_to_var_cost): Don't bound cost with spill_cost.
27728 (split_address_cost, ptr_difference_cost): Delete.
27729 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
27730 (struct ainc_cost_data): New struct.
27731 (get_address_cost_ainc): New function.
27732 (get_address_cost, get_computation_cost): Reimplement.
27733 (determine_group_iv_cost_address): Record inv_expr for all uses of
27735 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
27736 (iv_ca_has_deps): Reimplemented to ...
27737 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
27739 (iv_ca_extend): Call iv_ca_more_deps.
27741 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27743 * tree-ssa-address.c (struct mem_address): Move to header file.
27744 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
27745 * tree-ssa-address.h (struct mem_address): Move from C file.
27746 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
27748 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27750 * tree-affine.h (aff_combination_type): New interface.
27751 (aff_combination_zero_p): Remove static.
27752 (aff_combination_const_p): New interface.
27753 (aff_combination_singleton_var_p): New interfaces.
27755 2017-05-11 Richard Biener <rguenther@suse.de>
27757 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27758 Skip unreachable blocks and destinations.
27759 (eliminate): Move stmt removal and fixup ...
27760 (fini_eliminate): ... here. Skip inserted exprs.
27761 (pass_pre::execute): Move fini_pre after fini_eliminate.
27762 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
27763 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
27764 PRE to get rid of dead code that has invalid SSA form and
27765 split critical edges again.
27767 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27769 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27771 2017-05-11 Richard Biener <rguenther@suse.de>
27773 * passes.c (execute_function_todo): Verify loops if they are
27774 said to be up-to-date.
27775 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
27776 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
27778 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
27781 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
27782 handle calling assemble_external ourself.
27785 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
27786 modes with zero size. Enhance comment.
27788 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27790 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
27791 built-ins for vec_xl and vec_xst with short and char pointer
27794 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27796 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
27797 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
27798 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
27799 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
27800 (_mm_maskz_min_round_ss): New intrinsics.
27801 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
27802 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
27803 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
27804 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
27805 (__builtin_ia32_minss_mask_round): New builtins.
27806 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27807 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27808 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
27810 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
27811 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
27813 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
27816 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27818 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
27819 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
27820 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
27821 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
27822 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
27823 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
27824 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
27825 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27826 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27827 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
27828 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
27829 (__builtin_ia32_mulss_mask_round): New builtins.
27830 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27831 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27832 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
27834 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
27835 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27837 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27840 2017-05-10 Julia Koval <julia.koval@intel.com>
27842 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
27843 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
27844 (_mm256_setr_m128i): New intrinsics.
27846 2017-05-10 Julia Koval <julia.koval@intel.com>
27848 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
27849 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
27850 (_mm_maskz_rcp14_ss): New intrinsics.
27851 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
27852 (__builtin_ia32_rcp14ss_mask): New builtins.
27853 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
27855 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
27857 PR tree-optimization/51513
27858 * tree-cfg.c (gimple_seq_unreachable_p): New function.
27859 (assert_unreachable_fallthru_edge_p): Use it.
27860 (group_case_labels_stmt): Likewise.
27861 * tree-cfg.h: Prototype it.
27862 * stmt.c: Include cfghooks.h and tree-cfg.h.
27863 (emit_case_dispatch_table) <gap_label>: New local variable.
27864 Use it to fill dispatch table gaps.
27865 Test for default_label before updating probabilities.
27866 (expand_case) <default_label>: Remove unneeded initialization.
27867 Test for unreachable default case statement and remove its edge.
27868 Set default_label accordingly.
27869 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
27871 2017-05-10 Carl Love <cel@us.ibm.com>
27873 * config/rs6000/rs6000-c: Add support for built-in functions
27874 vector signed char vec_neg (vector signed char)
27875 vector signed short int vec_neg (vector short int)
27876 vector signed int vec_neg (vector signed int)
27877 vector signed long long vec_neg (vector signed long long)
27878 vector float vec_neg (vector float)
27879 vector double vec_neg (vector double)
27880 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
27882 * config/rs6000/altivec.h: Add define for vec_neg
27883 * doc/extend.texi: Update the built-in documentation for the
27884 new built-in functions.
27886 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27888 PR tree-optimization/77644
27889 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
27891 2017-05-10 Nathan Sidwell <nathan@acm.org>
27893 * dumpfile.h (TDI_lang_all): New.
27894 (TDF_KIND): New. Renumber others
27895 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
27897 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
27899 (get_dump_file_name): Adjust suffix generation.
27900 (dump_enable_all): Use TDF_KIND.
27901 * doc/invoke.texi (-fdump-lang-all): Document.
27903 * dumpfile.h: Tabify.
27905 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
27908 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
27909 Move member access before delete.
27911 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
27913 * tree-inline.c (expand_call_inline): Split block at stmt
27916 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
27919 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
27920 are now unused after splitting mov{sf,sd}_hardfloat.
27921 (f32_lr2): Likewise.
27922 (f32_lm): Likewise.
27923 (f32_lm2): Likewise.
27924 (f32_li): Likewise.
27925 (f32_li2): Likewise.
27926 (f32_lv): Likewise.
27927 (f32_sr): Likewise.
27928 (f32_sr2): Likewise.
27929 (f32_sm): Likewise.
27930 (f32_sm2): Likewise.
27931 (f32_si): Likewise.
27932 (f32_si2): Likewise.
27933 (f32_sv): Likewise.
27934 (f32_dm): Likewise.
27935 (f32_vsx): Likewise.
27936 (f32_av): Likewise.
27937 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
27938 For movsf, order stores so the VSX stores occur before the GPR
27939 store which encourages the register allocator to use a traditional
27940 FPR instead of a GPR. For movsd, order the stores so that the GPR
27941 store comes before the VSX stores to allow the power6 to work.
27942 This is due to the power6 not having a 32-bit integer store
27943 instruction from a FPR.
27944 (movsf_hardfloat): Likewise.
27945 (movsd_hardfloat): Likewise.
27947 2017-05-09 Martin Sebor <msebor@redhat.com>
27949 PR translation/80280
27950 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
27953 PR translation/80280
27954 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
27955 data member added in r247778.
27956 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
27958 2017-05-09 Nathan Sidwell <nathan@acm.org>
27960 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
27962 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
27965 2017-05-09 Marek Polacek <polacek@redhat.com>
27967 * doc/invoke.texi: Fix typo.
27969 2017-05-09 Richard Biener <rguenther@suse.de>
27971 * tree-vrp.c (vrp_val_is_max): Adjust comment.
27972 (vrp_val_is_min): Likewise.
27973 (set_value_range_to_value): Likewise.
27974 (set_value_range_to_nonnegative): Likewise.
27975 (gimple_assign_nonzero_p): Likewise.
27976 (gimple_stmt_nonzero_p): Likewise.
27977 (vrp_int_const_binop): Likewise. Remove unreachable case.
27978 (adjust_range_with_scev): Adjust comments.
27979 (compare_range_with_value): Likewise.
27980 (extract_range_from_phi_node): Likewise.
27981 (test_for_singularity): Likewise.
27983 2017-05-09 Richard Biener <rguenther@suse.de>
27985 * tree-vrp.c (get_single_symbol): Add assert that we don't
27986 get overflowed constants as invariant part.
27987 (compare_values_warnv): Add comment before the TREE_NO_WARNING
27988 checks. Use wi::cmp instead of recursing for integer constants.
27989 (compare_values): Just ignore whether we assumed undefined
27990 overflow instead of failing the compare.
27991 (extract_range_for_var_from_comparison_expr): Add comment before the
27992 TREE_NO_WARNING sets.
27993 (test_for_singularity): Likewise.
27994 (extract_range_from_comparison): Do not disable optimization
27995 when we assumed undefined overflow.
27996 (extract_range_basic): Remove init of unused var.
27998 2017-05-09 Richard Biener <rguenther@suse.de>
28000 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28001 (extract_range_from_multiplicative_op_1): Adjust.
28002 (extract_range_from_binary_expr_1): Use int_const_binop.
28004 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28007 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28008 rs6000_store_data_bypass_p in seven define_bypass directives and
28009 in several comments.
28010 * config/rs6000/rs6000-protos.h: Add prototype for
28011 rs6000_store_data_bypass_p function.
28012 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28013 function implements slightly different (rs6000-specific) semantics
28014 than store_data_bypass_p, returning false rather than aborting
28015 with assertion error when arguments do not satisfy the
28016 requirements of store data bypass.
28017 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28018 rs6000_store_data_bypass_p.
28020 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28022 * config/xtensa/xtensa-protos.h
28023 (xtensa_initial_elimination_offset): New declaration.
28024 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28025 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28026 macro definition, add case for FRAME_POINTER_REGNUM when
28027 FRAME_GROWS_DOWNWARD.
28028 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28029 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28030 xtensa_initial_elimination_offset.
28032 2017-05-08 Nathan Sidwell <nathan@acm.org>
28034 * doc/invoke.texi: Alphabetize -fdump options.
28036 2017-05-08 Martin Sebor <msebor@redhat.com>
28038 PR translation/80280
28039 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28041 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28043 * target.def (compute_frame_layout): New optional target hook.
28044 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28045 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28046 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28048 * reload1.c (verify_initial_elim_offsets): Likewise.
28049 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28050 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28051 (arm_get_frame_offsets): Split up into this ...
28052 (arm_compute_frame_layout): ... and this function.
28054 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28056 * config/aarch64/constraints.md (Usa): New constraint.
28057 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
28059 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
28061 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
28062 with_multilib_list after it has been checked.
28064 2017-05-08 Richard Biener <rguenther@suse.de>
28066 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
28067 (bitmap_set_subtract_values): Likewise.
28069 2017-05-08 Richard Biener <rguenther@suse.de>
28071 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
28072 (gimple_assign_nonzero): ... this and remove strict_overflow_p
28074 (gimple_stmt_nonzero_warnv_p): Rename to ...
28075 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
28077 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
28078 (extract_range_basic): Adjust, do not disable propagation on
28079 strict overflow sensitive simplification.
28080 (vrp_visit_cond_stmt): Likewise.
28082 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
28084 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
28085 body size unconditionally.
28087 2017-05-07 Jeff Law <law@redhat.com>
28090 2017-05-06 Jeff Law <law@redhat.com>
28091 PR tree-optimization/78496
28092 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28095 PR tree-optimization/78496
28096 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28097 (simplify_stmt_using_ranges): Call it.
28098 (vrp_dom_walker::before_dom_children): Extract equivalences
28099 from an ASSERT_EXPR with an equality comparison against a
28102 2017-05-06 Jeff Law <law@redhat.com>
28104 PR tree-optimization/78496
28105 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28108 PR tree-optimization/78496
28109 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28110 (simplify_stmt_using_ranges): Call it.
28111 (vrp_dom_walker::before_dom_children): Extract equivalences
28112 from an ASSERT_EXPR with an equality comparison against a
28115 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28117 * lra-constraints.c (lra_copy_reg_equiv): New function.
28118 (split_reg): Use it to copy equivalence information from the
28119 original register to the spill register.
28121 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28123 PR rtl-optimization/75964
28124 * simplify-rtx.c (simplify_const_relational_operation): Remove
28125 invalid handling of comparisons of integer ABS.
28127 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
28129 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
28130 initialize to zero.
28131 (init_regs): Remove declaration.
28132 (function_arg_advance_32): Initialize error_p as boolean variable.
28134 2017-05-05 Nathan Sidwell <nathan@acm.org>
28136 * store-motion.c (remove_reachable_equiv_notes): Reformat long
28137 lines. Use for (;;).
28139 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28141 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
28142 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
28143 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
28144 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
28145 VF=2 that require versioning.
28147 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28149 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
28152 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28154 * diagnostic.h (diagnostic_override_option_index): Convert from
28155 macro to inline function.
28157 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28159 * diagnostic.c (last_module_changed_p): New function.
28160 (set_last_module): New function.
28161 (diagnostic_report_current_module): Convert macro usage to
28162 the above functions.
28163 * diagnostic.h (diagnostic_context::last_module): Strengthen
28164 from const line_map * to const line_map_ordinary *.
28165 (diagnostic_last_module_changed): Delete macro.
28166 (diagnostic_set_last_module): Delete macro.
28168 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28170 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
28171 with diagnostic_report_diagnostic.
28172 (diagnostic_n_impl_richloc): Likewise.
28173 * diagnostic.h (report_diagnostic): Delete macro.
28174 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
28175 with diagnostic_report_diagnostic.
28176 * substring-locations.c (format_warning_va): Likewise.
28178 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28180 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
28181 save/restor of format_spec. Move option-printing code to...
28182 (print_option_information): ...this new function, and
28183 reimplement by simply printing to the pretty_printer,
28184 rather than appending to the format string.
28186 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28188 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
28189 handling logic into...
28190 (update_effective_level_from_pragmas): ...this new function.
28192 2017-05-04 Andrew Waterman <andrew@sifive.com>
28194 * config/riscv/riscv.opt (mstrict-align): New option.
28195 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
28196 (SLOW_UNALIGNED_ACCESS): Define.
28197 (riscv_slow_unaligned_access): Declare.
28198 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
28200 (riscv_slow_unaligned_access): New variable.
28201 (rocket_tune_info): Set slow_unaligned_access to true.
28202 (optimize_size_tune_info): Set slow_unaligned_access to false.
28203 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
28204 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
28205 (riscv_option_override): Set riscv_slow_unaligned_access.
28206 * doc/invoke.texi: Add -mstrict-align to RISC-V.
28208 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
28210 * config/riscv/riscv.md: Unify indentation.
28212 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28217 * config/rs6000/rs6000.md (u code attribute): Add FIX and
28219 (extendsi<mode>2): Add support for doing sign extension via
28220 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
28221 don't have ISA 3.0 instructions.
28222 (extendsi<mode>2 splitter): Likewise.
28223 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
28224 generate the normal insns since SImode can now go in vector
28225 registers. Disallow the special UNSPECs needed for previous
28226 machines to hide SImode being used. Add new insns
28227 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
28228 (fix_trunc<mode>si2_stfiwx): Likewise.
28229 (fix_trunc<mode>si2_internal): Likewise.
28230 (fixuns_trunc<mode>si2): Likewise.
28231 (fixuns_trunc<mode>si2_stfiwx): Likewise.
28232 (fctiw<u>z_<mode>_smallint): Likewise.
28233 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
28234 of floating point to 32-bit integer from doing a direct move to
28235 the GPR registers to do a store.
28236 (fctiwz_<mode>): Break long line.
28238 2017-05-05 Bin Cheng <bin.cheng@arm.com>
28240 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
28241 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
28242 (addr_list, addr_offset_valid_p): New.
28243 (split_address_groups): Check offset validity with above function.
28244 (gt-tree-ssa-loop-ivopts.h): Include header file.
28246 2017-05-05 Nathan Sidwell <nathan@acm.org>
28248 * config.gcc (arm*-*-*): Add missing 'fi'.
28250 2017-05-05 Steve Ellcey <sellcey@cavium.com>
28252 * doc/invoke.texi (-fopt-info): Explicitly say order of options
28253 included in -fopt-info does not matter.
28254 * doc/optinfo.texi (-fopt-info): Fix description of default
28255 behavour. Explicitly say order of options included in -fopt-info
28258 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28260 * config.gcc: Allow combinations of aprofile and rmprofile values for
28261 --with-multilib-list.
28262 * config/arm/t-multilib: New file.
28263 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
28264 variables. Remove setting of ISA and floating-point ABI in
28265 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
28266 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
28267 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
28268 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
28270 * config/arm/t-rmprofile: Likewise except for the matches changes.
28271 * doc/install.texi (--with-multilib-list): Document the combination of
28272 aprofile and rmprofile values and warn about pitfalls in doing that.
28274 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28276 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
28277 (movdi_aarch64): Likewise.
28279 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28281 PR tree-optimization/80632
28282 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
28284 (build_arrays): Initialize it for virtual phis.
28285 (fix_phi_nodes): Use it for virtual phis.
28287 PR tree-optimization/80558
28288 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
28289 [x, y] op z into [x op, y op z] for op & or | if conditions
28292 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
28293 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28296 * config/arm/arm.md (use_literal_pool): Remove.
28297 (64-bit immediate split): No longer takes cost into consideration
28298 if arm_disable_literal_pool is enabled.
28299 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
28300 used when arm_disable_literal_pool is enabled.
28301 (arm_max_const_double_inline_cost): Remove use of
28302 arm_disable_literal_pool.
28303 (push_minipool_fix): Add assert.
28304 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
28305 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
28306 (no_literal_pool_sf_immediate): New.
28308 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28310 PR tree-optimization/80613
28311 * tree-ssa-dce.c (propagate_necessity): Remove cases for
28312 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
28314 2017-05-05 Richard Biener <rguenther@suse.de>
28316 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
28318 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
28320 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
28321 of this flag from insn conditions due to removal from r247495.
28323 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28325 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
28327 (arm_early_store_addr_dep_ptr): Likewise.
28328 * config/arm/aarch-common-protos.h
28329 (arm_early_load_addr_dep_ptr): Add prototype.
28330 (arm_early_store_addr_dep_ptr): Likewise.
28331 * config/arm/cortex-a53.md: Add new bypasses.
28333 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28335 * tree.c (next_type_uid): Change type to unsigned.
28336 (type_hash_canon): Decrement back next_type_uid if
28337 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
28338 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
28341 2017-05-04 Martin Sebor <msebor@redhat.com>
28343 * builtins.c: Fix a trivial typo in a comment.
28345 PR middle-end/79234
28346 * builtins.c (check_sizes): Adjust to handle reading past the end.
28347 Avoid printing excessive upper bound of ranges. Use %E to print
28348 tree nodes instead of converting them to %wu.
28349 (expand_builtin_memchr): New function.
28350 (compute_dest_size): Rename...
28351 (compute_objsize): ...to this.
28352 (expand_builtin_memcpy): Adjust.
28353 (expand_builtin_mempcpy): Adjust.
28354 (expand_builtin_strcat): Adjust.
28355 (expand_builtin_strcpy): Adjust.
28356 (check_strncat_sizes): Adjust.
28357 (expand_builtin_strncat): Adjust.
28358 (expand_builtin_strncpy): Adjust and simplify.
28359 (expand_builtin_memset): Adjust.
28360 (expand_builtin_bzero): Adjust.
28361 (expand_builtin_memcmp): Adjust.
28362 (expand_builtin): Handle memcmp.
28363 (maybe_emit_chk_warning): Check strncat just once.
28365 2017-05-04 Martin Sebor <msebor@redhat.com>
28367 PR preprocessor/79214
28368 PR middle-end/79222
28369 PR middle-end/79223
28370 * builtins.c (check_sizes): Add inlining context and issue
28371 warnings even when -Wno-system-headers is set.
28372 (check_strncat_sizes): Same.
28373 (expand_builtin_strncat): Same.
28374 (expand_builtin_memmove): New function.
28375 (expand_builtin_stpncpy): Same.
28376 (expand_builtin): Handle memmove and stpncpy.
28378 2017-05-04 Bin Cheng <bin.cheng@arm.com>
28380 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
28381 which is not used any more.
28383 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28385 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
28387 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28389 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
28390 (cortexa53_tunings): Likewise.
28391 (cortexa57_tunings): Likewise.
28392 (cortexa72_tunings): Likewise.
28393 (cortexa73_tunings): Likewise.
28395 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28397 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
28398 Set loop alignment to 8.
28400 2017-05-04 Martin Sebor <msebor@redhat.com>
28402 PR translation/80280
28403 * builtins.c (expand_builtin_object_size): Add missing quoting to
28404 %D and like directives.
28405 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
28406 (hsa_type_for_tree_type): Same.
28407 (verify_function_arguments): Same.
28408 * symtab.c (symbol_table::change_decl_assembler_name): Same.
28409 * varasm.c (get_section): Same.
28412 2017-05-04 Martin Sebor <msebor@redhat.com>
28414 PR translation/80280
28415 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
28417 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28419 * config/aarch64/aarch64.c (generic_addrcost_table):
28420 Change HI/TI mode setting.
28422 2017-05-04 Martin Jambor <mjambor@suse.cz>
28424 PR tree-optimization/80622
28425 * tree-sra.c (comes_initialized_p): New function.
28426 (build_accesses_from_assign): Only set write lazily when
28427 comes_initialized_p is false.
28428 (analyze_access_subtree): Use comes_initialized_p.
28429 (propagate_subaccesses_across_link): Assert !comes_initialized_p
28430 instead of testing for PARM_DECL.
28432 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28434 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
28435 constraint on operand 0 to allow more general addressing modes.
28436 Adjust output template.
28437 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
28439 * config/aarch64/aarch64-protos.h
28440 (aarch64_address_valid_for_prefetch_p): Declare prototype.
28441 * config/aarch64/constraints.md (Dp): New address constraint.
28442 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
28445 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
28447 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
28448 update use of estimate_ipcp_clone_size_and_time.
28449 (estimate_local_effects): Update use of
28450 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
28451 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
28452 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
28453 Return nonspecialized time.
28455 2017-05-04 Richard Biener <rguenther@suse.de>
28457 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
28458 for the last VUSE which def dominates the PHI. Directly call
28460 (get_continuation_for_phi_1): Remove.
28462 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
28464 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
28465 to explain the use of truncating division. Cap the number of
28466 iterations to the maximum given by nb_iterations_upper_bound,
28469 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28471 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
28472 * configure: Regenerate.
28473 * config.in: Regenerate.
28474 * config/i386/driver-mingw32.c: new file.
28475 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
28476 * config.host: Link driver-mingw32.o on MinGW host.
28477 * doc/install.texi: Document new --enable-mingw-wildcard configure
28480 2017-05-04 Marek Polacek <polacek@redhat.com>
28482 PR tree-optimization/80612
28483 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
28485 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28486 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
28488 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
28489 (movt splitter): Likewise.
28490 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
28491 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
28492 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
28493 block for Thumb-1 with MOVT.
28494 (thumb2_legitimate_address_p): Move code block ...
28495 (can_avoid_literal_pool_for_label_p): ... into this new function.
28496 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
28498 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
28499 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
28500 "M-profile targets with the MOVT instruction".
28502 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28504 * config/arm/arm-builtins.c (arm_init_builtins): Rename
28505 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
28506 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
28508 2017-05-04 Martin Liska <mliska@suse.cz>
28510 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
28511 variable cond_code.
28513 2017-05-04 Richard Biener <rguenther@suse.de>
28515 * tree.c (array_at_struct_end_p): Handle arrays at struct
28516 end with flexarrays more conservatively. Refactor and treat
28517 arrays of arrays or aggregates more strict. Fix
28518 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
28519 * tree.c (array_at_struct_end_p): Adjust prototype.
28520 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
28521 * gimple-fold.c (get_range_strlen): Likewise.
28522 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
28524 2017-05-04 Richard Biener <rguenther@suse.de>
28526 PR tree-optimization/31130
28527 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
28529 (supports_overflow_infinity): Likewise.
28530 (is_negative_overflow_infinity): Likewise.
28531 (is_positive_overflow_infinity): Likewise.
28532 (is_overflow_infinity): Likewise.
28533 (stmt_overflow_infinity): Likewise.
28534 (overflow_infinity_range_p): Likewise.
28535 (usable_range_p): Remove as always returning true.
28536 (make_overflow_infinity): Remove.
28537 (negative_overflow_infinity): Likewise.
28538 (positive_overflow_infinity): Likewise.
28539 (avoid_overflow_infinity): Likewise.
28540 (set_value_range): Adjust accordingly.
28541 (set_value_range_to_nonnegative): Likewise, remove now unused
28542 overflow_infinity arg.
28543 (vrp_operand_equal_p): Adjust.
28544 (update_value_range): Likewise.
28545 (range_int_cst_singleton_p): Likewise.
28546 (operand_less_p): Likewise.
28547 (compare_values_warnv): Likewise.
28548 (extract_range_for_var_from_comparison_expr): Likewise.
28549 (vrp_int_const_binop): Likewise.
28550 (zero_nonzero_bits_from_vr): Likewise.
28551 (extract_range_from_multiplicative_op_1): Likewise.
28552 (extract_range_from_binary_expr_1): Likewise.
28553 (extract_range_from_unary_expr): Likewise.
28554 (extract_range_from_comparison): Likewise.
28555 (extract_range_basic): Likewise.
28556 (adjust_range_with_scev): Likewise.
28557 (compare_ranges): Likewise.
28558 (compare_range_with_value): Likewise.
28559 (dump_value_range): Likewise.
28560 (test_for_singularity): Likewise, remove strict_overflow_p parameter
28562 (simplify_cond_using_ranges): Adjust.
28564 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28566 * brig-builtins.def: Added a builtin for class_f64.
28567 * builtin-types.def: Added a builtin type needed by class_f64.
28569 2017-05-03 Jason Merrill <jason@redhat.com>
28571 * timevar.def: Add TV_CONSTEXPR.
28573 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28575 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
28577 2017-05-03 Martin Jambor <mjambor@suse.cz>
28579 * ipa-prop.c (ipa_update_after_lto_read): Removed.
28580 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
28581 * ipa-cp.c (ipcp_propagate_stage): Do not call
28582 ipa_update_after_lto_read.
28583 * ipa-inline.c (ipa_inline): Likewise.
28585 2017-05-03 Martin Jambor <mjambor@suse.cz>
28587 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
28588 tag. Added a default constructor and a destructor.
28589 (ipa_edge_args_sum_t): New class;
28590 (ipa_edge_args_sum): Declare.
28591 (ipa_edge_args_vector): Remove declaration.
28592 (IPA_EDGE_REF): Use ipa_edge_args_sum.
28593 (ipa_free_edge_args_substructures): Remove declaration.
28594 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
28595 (ipa_edge_args_info_available_for_edge_p): Likewise.
28596 * ipa-prop.c (ipa_edge_args_vector): Removed.
28597 (edge_removal_hook_holder): Likewise.
28598 (edge_duplication_hook_holder): Likewise.
28599 (ipa_edge_args_sum): New variable.
28600 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
28601 ipa_edge_args_vector.
28602 (ipa_free_edge_args_substructures): Likewise.
28603 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
28604 ipa_edge_args_vector.
28605 (ipa_edge_removal_hook): Turned into method
28606 ipa_edge_args_sum_t::remove.
28607 (ipa_edge_duplication_hook): Turned into method
28608 ipa_edge_args_sum_t::duplicate.
28609 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
28610 registering edge hooks.
28611 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
28612 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
28613 ipa_edge_args_sum instead of ipa_edge_args_vector.
28614 * ipa-profile.c (ipa_profile): Likewise.
28616 2017-05-03 Martin Jambor <mjambor@suse.cz>
28618 * symbol-summary.h (function_summary): New method exists.
28619 (function_summary::symtab_removal): Deallocate through release.
28620 (call_summary): New class.
28621 (gt_ggc_mx): New overload.
28622 (gt_pch_nx): Likewise.
28623 (gt_pch_nx): Likewise.
28625 2017-05-03 Jeff Law <law@redhat.com>
28627 PR tree-optimization/78496
28628 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
28629 from simplify_cond_using_ranges. Split off code to walk
28630 backwards through casts into ...
28631 (simplify_cond_using_ranges_2): New function.
28632 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
28633 (execute_vrp): After identifying jump threads, call
28634 simplify_cond_using_ranges_2.
28636 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
28639 * ipa-inline.h (inline_summary): Add ctor.
28640 (create_ggc): Do not use ggc_cleared_alloc.
28642 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
28643 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28645 * gcc.c (handle_braces): Support escaping in switch matching
28647 * doc/invoke.texi (Spec Files): Document it.
28648 Remove superfluous @code markup in items.
28650 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28652 * diagnostic-show-locus.c (struct column_range): New struct.
28653 (get_affected_columns): New function.
28654 (get_printed_columns): New function.
28655 (struct correction): New struct.
28656 (correction::ensure_capacity): New function.
28657 (correction::ensure_terminated): New function.
28658 (struct line_corrections): New struct.
28659 (line_corrections::~line_corrections): New dtor.
28660 (line_corrections::add_hint): New function.
28661 (layout::print_trailing_fixits): Reimplement in terms of the new
28663 (selftest::test_overlapped_fixit_printing): New function.
28664 (selftest::diagnostic_show_locus_c_tests): Call it.
28666 2017-05-03 Nathan Sidwell <nathan@acm.org>
28668 Canonicalize canonical type hashing
28669 * tree.h (type_hash_canon_hash): Declare.
28670 * tree.c (type_hash_list, attribute_hash_list): Move into
28671 type_hash_canon_hash.
28672 (build_type_attribute_qual_variant): Break out hash code calc into
28673 type_hash_canon_hash.
28674 (type_hash_canon_hash): New. Generic type hash computation.
28675 (build_range_type_1, build_array_type_1, build_function_type,
28676 build_method_type_directly, build_offset_type, build_complex_type,
28677 make_vector_type): Call it.
28679 2017-05-03 Richard Biener <rguenther@suse.de>
28681 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
28682 When all DRs have unknown misaligned do not always peel
28683 when there is a store but apply the same costing model as if
28684 there were only loads.
28686 2017-05-03 Richard Biener <rguenther@suse.de>
28689 PR tree-optimization/80492
28690 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28691 compare_base_decls returning dont-know properly.
28693 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
28695 * config/arm/iterators.md (CCSI): New mode iterator.
28696 (arch): New mode attribute.
28697 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
28698 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
28699 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
28700 code iterator for success result mode.
28701 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
28702 the corresponding new insn generators.
28704 2017-05-03 Bin Cheng <bin.cheng@arm.com>
28707 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28708 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28710 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
28712 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
28713 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
28714 (DDR_A): Wrap DDR argument in brackets.
28715 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
28716 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
28717 (DDR_REVERSED_P): Likewise.
28719 2017-05-03 Jakub Jelinek <jakub@redhat.com>
28721 PR tree-optimization/79472
28722 * tree-switch-conversion.c (struct switch_conv_info): Add
28723 contiguous_range and default_case_nonstandard fields.
28724 (collect_switch_conv_info): Compute contiguous_range and
28725 default_case_nonstandard fields, don't clear final_bb if
28726 contiguous_range and only the default case doesn't have the required
28728 (check_all_empty_except_final): Set default_case_nonstandard instead
28729 of failing if contiguous_range and the default case doesn't have empty
28731 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
28732 and only the default case doesn't have the required constants. Skip
28734 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
28735 if default_case_nonstandard.
28736 (build_constructors): Build constant 1 just once. Assert that default
28737 values aren't inserted in between cases if contiguous_range. Skip
28739 (build_arrays): Skip virtual phis.
28740 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
28741 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
28742 Handle virtual phis.
28743 (gen_inbound_check): Handle default_case_nonstandard case.
28744 (process_switch): Adjust check_final_bb caller. Call
28745 gather_default_values with the first non-default case instead of
28746 default case if default_case_nonstandard.
28748 2017-05-02 Nathan Sidwell <nathan@acm.org>
28750 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
28751 check. Fix formatting.
28753 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
28755 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
28756 errors when comparing specialized and unspecialized times.
28758 2017-05-02 David Malcolm <dmalcolm@redhat.com>
28760 * diagnostic-show-locus.c
28761 (layout::should_print_annotation_line_p): Make private.
28762 (layout::print_annotation_line): Make private.
28763 (layout::annotation_line_showed_range_p): Make private.
28764 (layout::show_ruler): Make private.
28765 (layout::print_source_line): Make private. Pass in line and
28766 line_width, rather than calling location_get_source_line. Drop
28768 (layout::print_leading_fixits): New method.
28769 (layout::print_any_fixits): Rename to...
28770 (layout::print_trailing_fixits): ...this, and make private.
28771 Don't print newline fixits.
28772 (diagnostic_show_locus): Move logic for printing one row into...
28773 (layout::print_line): ...this new function. Move the
28774 location_get_source_line call and error-handling from
28775 print_source_line to here. Call print_leading_fixits, and rename
28776 print_any_fixits to print_trailing_fixits.
28777 (selftest::test_fixit_insert_containing_newline): Update now that
28778 newlines are partially supported.
28779 (selftest::test_fixit_insert_containing_newline_2): New test.
28780 (selftest::test_fixit_replace_containing_newline): Update comments.
28781 (selftest::diagnostic_show_locus_c_tests): Call the new test.
28782 * edit-context.c (class added_line): New class.
28783 (class edited_line): Describe newline handling in comment.
28784 (edited_line::actually_edited_p): New method.
28785 (edited_line::print_content): Delete redundant decl.
28786 (edited_line::m_predecessors): New field.
28787 (edited_file::print_content): Call edited_line::print_content.
28788 (edited_file::print_diff): Update to support newlines.
28789 (edited_file::print_diff_hunk): Likewise.
28790 (edited_file::print_run_of_changed_lines): New function.
28791 (edited_file::print_diff_line): Convert to...
28792 (print_diff_line): ...this.
28793 (edited_file::get_effective_line_count): New function.
28794 (edited_line::edited_line): Initialize new field m_predecessors.
28795 (edited_line::~edited_line): Clean up m_predecessors.
28796 (edited_line::apply_fixit): Handle newlines.
28797 (edited_line::get_effective_line_count): New function.
28798 (edited_line::print_content): New function.
28799 (edited_line::print_diff_lines): New function.
28800 (selftest::test_applying_fixits_insert_containing_newline): New
28802 (selftest::test_applying_fixits_replace_containing_newline): New
28804 (selftest::insert_line): New function.
28805 (selftest::test_applying_fixits_multiple_lines): Add example of
28807 (selftest::edit_context_c_tests): Call the new tests.
28809 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28811 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
28812 parameter cand. Update dump information.
28813 (get_computation_cost): Update uses.
28815 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28817 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
28818 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
28819 (get_computation_at, rewrite_use_address): Update use of
28820 get_computation_aff.
28822 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28824 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
28825 (get_computation): Delete.
28826 (get_computation_cost): Implement like get_computation_cost_at.
28827 Use get_computation_at.
28828 (get_computation_cost_at): Delete.
28829 (rewrite_use_nonlinear_expr): Use get_computation_at.
28830 (rewrite_use_compare, remove_unused_ivs): Ditto.
28832 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28834 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
28836 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28838 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
28839 (ivopts_global_cost_for_size): Rename parameter and update uses.
28840 (iv_ca_recount_cost): Update uses.
28841 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
28842 candidates seperately in n_invs and n_cands.
28843 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
28845 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28847 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
28848 (find_inv_vars_cb): New.
28849 (find_depends): Renamed to ...
28850 (find_inv_vars): ... this.
28851 (add_candidate_1, force_var_cost): Call find_inv_vars.
28852 (split_address_cost, determine_group_iv_cost_cond): Ditto.
28854 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28856 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
28857 inv_vars. Add inv_exprs.
28858 (struct iv_cand): Rename depends_on to inv_vars.
28859 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
28860 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
28861 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
28862 (dump_cand): Dump inv_vars.
28863 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
28864 (record_invariant, find_depends, add_candidate_1): Ditto.
28865 (set_group_iv_cost, force_var_cost): Ditto.
28866 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
28867 (get_computation_cost_at, get_computation_cost): Ditto.
28868 (determine_group_iv_cost_generic): Ditto.
28869 (determine_group_iv_cost_address): Ditto.
28870 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
28871 (determine_group_iv_costs): Ditto.
28872 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
28873 (iv_ca_set_remove_invariants): Renamed to ...
28874 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
28875 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
28876 (iv_ca_set_add_invariants): Renamed to ...
28877 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
28878 (iv_ca_set_cp): Use iv_ca_set_add_invs.
28879 (iv_ca_has_deps): Support inv_vars and inv_exprs.
28880 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
28881 (create_new_ivs): Remove useless dump.
28883 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28885 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
28887 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
28888 (iv_ca_set_no_cp, create_new_iv): Ditto.
28890 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28892 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28894 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28896 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
28897 function tree_check2.
28899 2017-05-02 Martin Liska <mliska@suse.cz>
28901 * doc/gcov.texi: Add missing preposition.
28902 * gcov.c (function_info::function_info): Properly fill up
28903 all member variables.
28905 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28907 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
28909 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28911 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
28913 2017-05-02 Martin Liska <mliska@suse.cz>
28916 * lto-streamer-in.c (lto_read_tree_1): Remove
28917 LTO_STREAMER_DEBUG.
28918 * lto-streamer.c (struct tree_hash_entry): Likewise.
28919 (struct tree_entry_hasher): Likewise.
28920 (tree_entry_hasher::hash): Likewise.
28921 (tree_entry_hasher::equal): Likewise.
28922 (lto_streamer_init): Likewise.
28923 (lto_orig_address_map): Likewise.
28924 (lto_orig_address_get): Likewise.
28925 (lto_orig_address_remove): Likewise.
28926 * lto-streamer.h: Likewise.
28927 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
28928 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
28930 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
28932 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
28933 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
28934 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
28935 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
28936 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
28937 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
28938 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
28939 (mm_maskz_sub_ss): New intrinsics.
28940 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28941 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28942 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
28943 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
28944 (__builtin_ia32_subss_mask_round): New builtins.
28945 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28946 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28947 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
28949 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
28950 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28952 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28955 2017-05-02 Martin Jambor <mjambor@suse.cz>
28957 PR tree-optimization/78687
28958 * tree-sra.c (access): New field parent.
28959 (process_subtree_disqualification): New function.
28960 (disqualify_candidate): Call it.
28961 (build_accesses_from_assign): Reset write flag if creating an
28963 (build_access_subtree): Fill in parent field and also prpagate
28964 down grp_write flag.
28965 (create_artificial_child_access): New parameter set_grp_write, set
28966 grp_write to its value.
28967 (propagate_subaccesses_across_link): Also propagate grp_write flag
28969 (propagate_all_subaccesses): Push the closest parent back to work
28970 queue if add_access_to_work_queue returned true.
28972 2017-05-02 Richard Biener <rguenther@suse.de>
28974 * common.opt (fstrict-overflow): Alias negative to fwrapv.
28975 * doc/invoke.texi (fstrict-overflow): Remove all traces of
28976 -fstrict-overflow documentation.
28977 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
28978 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
28979 flag_strict_overflow.
28980 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
28981 * lto-opts.c (lto_write_options): Do not stream it.
28982 * lto-wrapper.c (merge_and_complain): Do not handle it.
28983 * opts.c (default_options_table): Do not set -fstrict-overflow.
28984 (finish_options): Likewise do not clear it when sanitizing.
28985 * simplify-rtx.c (simplify_const_relational_operation): Do not
28986 test flag_strict_overflow.
28988 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
28990 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
28991 using enabled attribute.
28992 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
28993 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
28994 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
28995 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
28996 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
28997 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
28998 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
28999 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29000 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29001 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29003 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29005 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29007 2017-05-02 Richard Biener <rguenther@suse.de>
29009 PR tree-optimization/80591
29011 2017-04-10 Richard Biener <rguenther@suse.de>
29013 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29016 2017-05-02 Richard Biener <rguenther@suse.de>
29018 PR tree-optimization/80549
29019 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29020 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29021 headers if they do not have a preheader.
29023 2017-05-02 Martin Liska <mliska@suse.cz>
29026 * common.opt: Fix typo.
29027 * doc/invoke.texi: Likewise.
29029 2017-05-01 Jan Beulich <jbeulich@suse.com>
29031 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29032 swapping, add (x,x,m,x,n) alternative.
29034 2017-05-01 Nathan Sidwell <nathan@acm.org>
29036 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29037 unnecessary unadjusted_alignment check.
29039 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29042 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29044 * gimplify.c (gimplify_cilk_detach): New function.
29045 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29046 * tree-core.h: Document EXPR_CILK_SPAWN.
29047 * tree.h (EXPR_CILK_SPAWN): Define.
29049 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29051 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29052 to use new fixit_hint representation, using the "replace" logic.
29053 (get_line_span_for_fixit_hint): Likewise.
29054 (layout::print_any_fixits): Likewise.
29055 (selftest::test_one_liner_many_fixits): Rename to...
29056 (selftest::test_one_liner_many_fixits_1): ...this, and update
29057 comment and expected output to reflect that the multiple fix-it
29058 hints are now consolidated into one insertion.
29059 (selftest::test_one_liner_many_fixits_2): New test.
29060 (selftest::test_diagnostic_show_locus_one_liner): Update for
29062 (selftest::test_fixit_consolidation): Update for fix-it API
29064 * diagnostic.c (print_parseable_fixits): Likewise.
29065 * edit-context.c (edited_line::m_line_events): Convert from
29066 auto_vec <line_event *> to auto_vec <line_event>.
29067 (class line_event): Convert from abstract base class to a concrete
29068 class, taking over the role of replace_event.
29069 (class insert_event): Delete.
29070 (class replace_event): Rename to class line_event. Convert to
29072 (edit_context::add_fixits): Reimplement.
29073 (edit_context::apply_insert): Delete.
29074 (edit_context::apply_replace): Rename to...
29075 (edit_context::apply_fixit): ...this. Convert to half-open range.
29076 (edited_file::apply_insert): Delete.
29077 (edited_file::apply_replace): Rename to...
29078 (edited_file::apply_fixit): ...this.
29079 (edited_line::~edited_line): Drop deletion of events.
29080 (edited_line::apply_insert): Delete.
29081 (edited_line::apply_replace): Rename to...
29082 (edited_line::apply_fixit): ...this. Convert to half-open range.
29083 Update for change to type of m_line_events.
29084 * edit-context.h (edit_context::apply_insert): Delete.
29085 (edit_context::apply_replace): Rename to...
29086 (edit_context::apply_fixit): ...this.
29088 2017-05-01 Martin Sebor <msebor@redhat.com>
29090 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
29093 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
29096 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
29097 __get_cpuid_max returns 0.
29098 (__get_cpuid_count): Ditto.
29100 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
29102 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
29103 replacement expression is another instance of one of its arguments.
29105 2017-05-01 Jakub Jelinek <jakub@redhat.com>
29108 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
29109 check for stack push/pop autoinc.
29110 * config/i386/i386.c (ix86_agi_dependent): Return false
29111 if the only reason why modified_in_p returned true is that
29112 addr is SP based and set_insn is a push or pop.
29114 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29116 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
29119 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29122 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
29123 (account_size_time): Use two predicates - exec_pred and
29125 (evaluate_conditions_for_known_args): Compute both clause and
29127 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
29128 (inline_summary_t::duplicate): Update.
29129 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
29131 (compute_inline_parameters): Likewise.
29132 (estimate_edge_size_and_time): Update caluclation of time.
29133 (estimate_node_size_and_time): Compute both time and nonspecialized
29135 (estimate_ipcp_clone_size_and_time): Update.
29136 (inline_merge_summary): Update.
29137 (do_estimate_edge_time): Update.
29138 (do_estimate_edge_size): Update.
29139 (do_estimate_edge_hints): Update.
29140 (inline_read_section, inline_write_summary): Stream both new predicates.
29141 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
29143 (compute_inlined_call_time): Cleanup.
29144 (big_speedup_p): Update.
29145 (edge_badness): Update.
29146 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
29147 (size_time_entry): Replace predicate by exec_predicate and
29148 nonconst_predicate.
29149 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
29150 (estimate_edge_time): Return also nonspec_time.
29151 (reset_edge_growth_cache): Update.
29153 2017-04-29 Jakub Jelinek <jakub@redhat.com>
29155 PR rtl-optimization/80491
29156 * ifcvt.c (noce_process_if_block): When looking for x setter
29157 with missing else_bb, don't check only the insn right before
29158 cond_earliest, but look for the last insn that x is modified in
29159 within the same bb.
29161 PR rtl-optimization/80491
29162 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
29164 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
29166 PR tree-optimization/80487
29167 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
29169 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29171 PR tree-optimization/79697
29172 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
29173 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
29174 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
29176 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
29177 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
29179 2017-04-28 Martin Sebor <msebor@redhat.com>
29181 PR tree-optimization/80523
29182 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
29183 (init_target_to_host_charmap, target_to_host, target_strtol10): New
29185 (maybe_warn, format_directive, parse_directive): Use new functions.
29186 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
29188 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
29190 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
29192 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
29194 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
29195 target_header_dir): Set correctly.
29196 * configure: Regenerated.
29197 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29198 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29199 instead of SYSTEM_HEADER_DIR.
29201 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
29203 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
29204 (estimate_local_effects): Likewise.
29205 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
29206 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
29207 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
29208 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
29209 do_estimate_edge_time, estimate_edge_time): Likewise.
29210 * ipa-inline-analysis.c (estimate_node_size_and_time,
29211 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
29212 (estimate_time_after_inlining): Remove.
29214 2017-04-28 Martin Liska <mliska@suse.cz>
29216 * doc/gcov.texi: Enhance documentation of gcov.
29218 2017-04-28 Martin Liska <mliska@suse.cz>
29220 * doc/gcov.texi: Sort options in alphabetic order.
29221 * doc/gcov-dump.texi: Likewise.
29222 * doc/gcov-tool.texi: Likewise.
29223 * gcov.c (print_usage): Likewise.
29224 * gcov-dump.c (print_usage): Likewise.
29225 * gcov-tool.c (print_merge_usage_message): Likewise.
29226 (print_rewrite_usage_message): Likewise.
29227 (print_overlap_usage_message): Likewise.
29229 2017-04-28 Martin Liska <mliska@suse.cz>
29231 PR gcov-profile/53915
29232 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
29234 2017-04-28 Martin Liska <mliska@suse.cz>
29236 PR gcov-profile/79891
29237 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
29238 is marked by compiler as living on a line.
29239 (get_cycles_count): Remove usage of the union.
29240 (output_intermediate_file): Likewise.
29241 (find_source): Fix GNU coding style.
29242 (accumulate_line_counts): Remove old non-all block mode.
29243 (output_lines): Remove usage of the union.
29244 * profile.c (output_location): Include all BBs, even if
29245 belonging to a same line (and file) as a previous BB.
29247 2017-04-28 Martin Liska <mliska@suse.cz>
29249 * gcov.c (process_args): Handle new argument 'w'.
29250 (read_graph_file): Assign ID to BBs.
29251 (output_branch_count): Display BB # if verbose flag is set.
29252 (output_lines): Likewise for arcs.
29253 (print_usage): Add '--verbose' option help.
29254 * doc/gcov.texi: Document --verbose (-w) option.
29256 2017-04-28 Martin Liska <mliska@suse.cz>
29258 * gcov.c (struct block_location_info): New struct.
29259 (process_file): Fill up the new structure.
29260 (read_graph_file): Replace usage of encoding by the newly added
29262 (add_line_counts): Likewise.
29263 (accumulate_line_counts): Remove usage of the union.
29264 (function_info::function_info): New function.
29265 (function_info::~function_info): Likewise.
29266 (process_file): Call delete instead of release_function.
29267 (release_function): Release the function.
29268 (release_structures): Call delete instead of release_function.
29269 (solve_flow_graph): Replace usage of num_blocks.
29270 (find_exception_blocks): Likewise.
29271 (output_lines): Fix GNU coding style.
29273 2017-04-28 Martin Liska <mliska@suse.cz>
29276 * coverage.c (coverage_remove_note_file): New function.
29277 * coverage.h: Declare the function.
29278 * toplev.c (finalize): Clean if an error has been seen.
29280 2017-04-28 Martin Liska <mliska@suse.cz>
29282 PR gcov-profile/80031
29283 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
29284 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
29285 * gcov.c (read_graph_file): Read just number of blocks.
29286 * profile.c (branch_prob): Do not stream 0 flags per a basic
29289 2017-04-28 Martin Liska <mliska@suse.cz>
29291 * gcov-dump.c (tag_*): Add new argument to declarations.
29292 (dump_gcov_file): Likewise.
29293 (tag_blocks): Add and use new argument depth.
29294 (tag_arcs): Likewise.
29295 (tag_lines): Likewise.
29296 (tag_counters): Likewise.
29297 (tag_summary): Likewise.
29298 (dump_working_sets): Use depth to do a proper indentation.
29300 2017-04-28 Jakub Jelinek <jakub@redhat.com>
29303 * cgraph.h (symtab_node::debug_symtab): No longer inline.
29304 * symtab.c (symtab_node::debug_symtab): Move definition here.
29306 2017-04-28 Richard Biener <rguenther@suse.de>
29308 * lto-streamer.h (LTO_major_version): Bump to 7.
29310 2017-04-28 Richard Biener <rguenther@suse.de>
29312 * tree-vrp.c (assert_info): New struct.
29313 (add_assert_info): New helper.
29314 (register_edge_assert_for_2): Refactor to add asserts to a vector
29316 (register_edge_assert_for_1): Likewise.
29317 (register_edge_assert_for): Likewise.
29318 (finish_register_edge_assert_for): New helper actually registering
29319 asserts where live on edge.
29320 (find_conditional_asserts): Adjust.
29321 (find_switch_asserts): Likewise.
29322 (evrp_dom_walker::try_find_new_range): Generalize.
29323 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
29325 2017-04-27 Marek Polacek <polacek@redhat.com>
29328 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
29329 arg10 and arg11 to itype.
29331 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
29333 * doc/extend.texi (Object Size Checking): Improve grammar.
29335 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
29338 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
29339 that the logic for permitting reciprocal estimates matches that
29342 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29345 * tree.c (type_cache_hasher::equal): Only compare
29346 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
29347 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
29348 non-aggregate element types.
29349 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
29350 about the flag on ARRAY_TYPEs in the comment, formatting fix.
29352 2017-04-27 Richard Biener <rguenther@suse.de>
29354 PR middle-end/80533
29355 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
29356 stripping ARRAY_REFs from MEM_EXPR make sure we're not
29357 keeping a reference to a trailing array.
29359 2017-04-27 Richard Biener <rguenther@suse.de>
29361 PR middle-end/80539
29362 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
29363 being in loop-closed SSA form conservatively.
29364 (chrec_fold_multiply_poly_poly): Likewise.
29366 2017-04-27 Tamar Christina <tamar.christina@arm.com>
29368 PR middle-end/79665
29369 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
29370 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
29372 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29375 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
29376 (aarch64_function_arg_alignment): Return unsigned int again, but still
29377 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
29378 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29379 Don't emit -Wpsabi note.
29380 (aarch64_function_arg_boundary): Likewise.
29381 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29384 2017-04-26 Nathan Sidwell <nathan@acm.org>
29386 * tree.h (crc32_unsigned_n): Declare.
29387 (crc32_unsigned, crc32_unsigned): Make inline.
29388 * tree.c (crc32_unsigned_bits): Replace with ...
29389 (crc32_unsigned_n): ... this.
29390 (crc32_unsigned, crc32_byte): Remove.
29391 (crc32_string): Remove unnecessary braces.
29393 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29395 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
29396 * ipa-inline-analysis.c (MAX_TIME): Remove.
29397 (account_size_time): Use sreal for time.
29398 (dump_inline_summary): Update.
29399 (estimate_function_body_sizes): Update.
29400 (estimate_edge_size_and_time): Update.
29401 (estimate_calls_size_and_time): Update.
29402 (estimate_node_size_and_time): Update.
29403 (inline_merge_summary): Update.
29404 (inline_update_overall_summary): Update.
29405 (estimate_time_after_inlining): Update.
29406 (inline_read_section): Update.
29407 (inline_write_summary): Update.
29408 * ipa-inline.c (compute_uninlined_call_time): Update.
29409 (compute_inlined_call_time): Update.
29410 (recursive_inlining): Update.
29411 (inline_small_functions): Update.
29412 (dump_overall_stats): Update.
29413 * ipa-inline.h: Include sreal.h.
29414 (size_time_entry): Turn time to sreal.
29415 (inline_summary): Turn self_time nad time to sreal.
29417 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29419 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
29421 (sreal::stream_out, sreal::stream_in): New.
29422 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
29424 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29426 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
29429 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
29432 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
29434 (dimode_scalar_chain::compute_convert_gain): Ditto.
29435 (dimode_scalar_chain::make_vector_copies): Ditto.
29436 (dimode_scalar_chain::convert_reg): Ditto.
29437 (dimode_scalar_chain::convert_insn): Ditto.
29438 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
29439 (VI248_AVX512BW_1): New mode iterator.
29440 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
29441 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
29444 2017-04-25 Martin Sebor <msebor@redhat.com>
29446 PR tree-optimization/80497
29447 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
29448 constants are representable in HOST_WIDE_INT.
29449 (parse_directive): Ditto.
29451 2017-04-25 Martin Sebor <msebor@redhat.com>
29454 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
29455 (new_zero_array): Adjust signature.
29456 (dom_info::dom_init): Used unsigned rather that size_t.
29457 (dom_info::dom_info): Same.
29459 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29460 Jakub Jelinek <jakub@redhat.com>
29463 * config/arm/arm.c: Include gimple.h.
29464 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29465 returns negative, increment ncrn only if it returned positive.
29466 (arm_needs_doubleword_align): Return int instead of bool,
29467 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
29468 members, but if there is any such non-FIELD_DECL
29469 > PARM_BOUNDARY aligned decl, return -1 instead of false.
29470 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29471 returns negative, increment nregs only if it returned positive.
29472 (arm_setup_incoming_varargs): Likewise.
29473 (arm_function_arg_boundary): Emit -Wpsabi note if
29474 arm_needs_doubleword_align returns negative, return
29475 DOUBLEWORD_ALIGNMENT only if it returned positive.
29477 2017-04-25 Marek Polacek <polacek@redhat.com>
29480 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29481 first argument to type.
29483 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
29486 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
29487 type checks to test for compatibility instead of equality.
29489 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29490 Jakub Jelinek <jakub@redhat.com>
29493 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
29495 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
29496 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
29497 the alignment computation, but return their maximum in warn_alignment.
29498 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29499 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
29501 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
29502 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29505 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29507 * config/arc/simdext.md (dmpyh): Fix typo.
29509 2017-04-25 Richard Biener <rguenther@suse.de>
29511 PR tree-optimization/80492
29512 * alias.c (compare_base_decls): Handle registers with asm
29513 specification conservatively.
29514 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29515 compare_base_decls returning dont-know properly.
29517 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29519 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
29520 (legitimate_offset_address_p): New function.
29521 (arc_legitimate_address_p): Use above function.
29523 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29525 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
29527 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29529 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
29530 ACCH registers whenever they are available.
29532 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29534 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
29535 double regs fix when not used.
29537 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29539 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
29541 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
29542 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
29544 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29546 * config/arc/arc.c (arc_output_addsi): Check for h-register class
29547 when emitting short ADD instructions.
29549 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29551 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
29553 (cmpsi_cc_c_insn): Likewise.
29554 (cbranchsi4_scratch): Compute proper instruction length using
29555 compact_hreg_operand.
29556 * config/arc/predicates.md (compact_hreg_operand): New predicate.
29558 2017-04-25 Richard Biener <rguenther@suse.de>
29560 PR middle-end/80509
29561 * passes.c (pass_manager::pass_manager): Initialize
29562 m_name_to_pass_map.
29564 2017-04-25 Richard Biener <rguenther@suse.de>
29566 PR tree-optimization/79201
29567 * tree-ssa-sink.c (statement_sink_location): Handle calls.
29569 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29572 * config/s390/vector.md: Split MEM->GPR vector moves for
29573 non-s_operand addresses.
29575 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29578 * config/s390/predicates.md (reload_const_wide_int_operand): New
29580 * config/s390/s390.md ("movti"): Remove d/P alternative.
29581 ("movti_bigconst"): New pattern definition.
29583 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29586 * s390-protos.h (s390_expand_cs_hqi): Removed.
29587 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
29588 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
29589 modes as well as CCZ1mode and CCZmode.
29590 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
29591 signature of s390_emit_compare_and_swap.
29592 (s390_expand_cs_hqi): Likewise, make static.
29593 (s390_expand_cs_tdsi): Generate an explicit compare before trying
29594 compare-and-swap, in some cases.
29595 (s390_expand_cs): Wrapper function.
29596 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
29598 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
29599 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
29600 patterns for small and large integers. Forbid symref memory operands.
29601 Move expander to s390.c. Require cc register.
29602 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
29603 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
29604 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
29605 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
29606 symref memory operands. Remove CC mode and call s390_match_ccmode
29608 ("atomic_exchange<mode>"): Allow and implement all integer modes.
29610 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29612 * config/s390/s390.md (define_peephole2): New peephole to help
29613 combining the load-and-test pattern with volatile memory.
29615 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29617 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
29618 with CCZmode for TARGET_Z196.
29620 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29622 PR rtl-optimization/80501
29623 * combine.c (make_compound_operation_int): Set subreg_code to SET
29624 even for AND with mask of the sign bit of mode.
29626 PR rtl-optimization/80500
29627 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
29628 sum's initial value.
29630 2017-04-25 Julian Brown <julian@codesourcery.com>
29631 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29633 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
29635 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
29637 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
29639 2017-04-25 Julian Brown <julian@codesourcery.com>
29640 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29642 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
29643 (thunderx2t99_sha): New Reservation.
29645 2017-04-25 Julian Brown <julian@codesourcery.com>
29646 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29648 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
29649 type for 1-element load.
29651 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
29653 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
29655 2017-04-24 Martin Jambor <mjambor@suse.cz>
29657 PR tree-optimization/80293
29658 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
29659 char arrays not totally scalarizable if it is false.
29660 (analyze_all_variable_accesses): Pass correct value in the new
29661 parameter. Add a statistics counter.
29663 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
29665 PR middle-end/79931
29666 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
29668 2017-04-24 Richard Biener <rguenther@suse.de>
29670 PR tree-optimization/80494
29671 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
29672 out for complex types.
29674 2017-04-24 Richard Biener <rguenther@suse.de>
29676 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
29677 * tree-ssa-sccvn.c (print_scc): Print SCC size.
29678 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
29679 (DFS): Adjust and never fail.
29680 (sccvn_dom_walker::fail): Remove.
29681 (sccvn_dom_walker::before_dom_children): Adjust.
29682 (run_scc_vn): Likewise and never fail.
29683 * tree-ssa-pre.c (pass_pre::execute): Adjust.
29684 (pass_fre::execute): Likewise.
29686 2017-04-24 Richard Biener <rguenther@suse.de>
29688 PR tree-optimization/79725
29689 * tree-ssa-sink.c (statement_sink_location): Return whether
29690 failure reason was zero uses. Move that check later.
29691 (sink_code_in_bb): Deal with zero uses by removing the stmt
29694 2017-04-24 Richard Biener <rguenther@suse.de>
29697 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
29698 pointer-based references.
29700 2017-04-24 Richard Biener <rguenther@suse.de>
29703 * pass_manager.h (pass_manager::operator new): Remove.
29704 (pass_manager::operator delete): Likewise.
29705 * passes.c (pass_manager::operator new): Remove.
29706 (pass_manager::operator delete): Likewise.
29707 (pass_manager::pass_manager): Zero individual pass members.
29709 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
29712 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
29713 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
29714 Check "XEXP (src, 1)" operand here.
29715 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
29716 Check "XEXP (src, 1)" operand here.
29717 (dimode_scalar_chain::make_vector_copies): Detect count register
29718 of a shift instruction. Zero extend count register from QImode
29719 to DImode to satisfy vector shift pattern count operand predicate.
29720 Substitute vector shift count operand with a DImode copy.
29721 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
29724 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
29726 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
29727 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29728 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
29729 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29730 (UNSPEC_NOREX_MEM): Remove definition.
29732 2017-04-21 Richard Biener <rguenther@suse.de>
29734 PR tree-optimization/79547
29735 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
29736 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
29737 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
29738 without any constraints.
29740 2017-04-21 Richard Biener <rguenther@suse.de>
29742 PR tree-optimization/78847
29743 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
29745 2017-04-21 Richard Biener <rguenther@suse.de>
29747 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
29748 (build_distinct_type_copy): Likewise.
29749 (build_variant_type_copy): Likewise.
29750 * tree.c (build_qualified_type): Pass down mem-stat info.
29751 (build_distinct_type_copy): Likewise.
29752 (build_variant_type_copy): Likewise.
29754 2017-04-21 Richard Biener <rguenther@suse.de>
29756 PR tree-optimization/80237
29757 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
29759 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
29760 for a simplified result.
29762 2016-04-21 Richard Biener <rguenther@suse.de>
29764 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
29765 sth as strict as a simple_iv but a chrec without symbols and an
29766 operand defined in the loop we are peeling (and not some subloop).
29767 (propagate_constants_for_unrolling): Propagate all constants.
29769 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29772 * config/i386/i386.c (print_reg): Remove assert for disalowed
29773 regno values, call output_operand_lossage instead.
29775 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29778 * config/i386/constraints.md (Yc): New register constraint.
29779 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
29780 Use Yc constraint for alternative 2 of operand 0. Remove
29781 preferred_for_speed attribute.
29783 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29785 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
29786 lastprivate clauses in SIMT case.
29788 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
29790 * doc/invoke.texi (-Wextra-semi): Document new warning option.
29792 2017-04-20 Richard Biener <rguenther@suse.de>
29794 PR tree-optimization/57796
29795 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
29796 as N scalar stores.
29797 (vect_model_load_cost): Cost gathers as N scalar loads.
29799 2017-04-20 Richard Biener <rguenther@suse.de>
29801 * ggc-page.c (ggc_allocated_p): Rename to ...
29802 (safe_lookup_page_table_entry): ... this and return the lookup
29804 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
29806 2017-04-20 Richard Biener <rguenther@suse.de>
29808 PR tree-optimization/80453
29809 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
29810 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
29811 from the conditions.
29812 (vn_phi_eq): Pass them down.
29813 (vn_phi_lookup): Record them.
29814 (vn_phi_insert): Likewise.
29816 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
29818 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
29819 uninitialized variable warning to avoid buffer overrun.
29821 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29824 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
29825 is suppressed for '{ 0 }' in C.
29827 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29829 * BASE-VER: Set to 8.0.0.
29831 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
29833 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
29834 priority .init_array and .fini_array section with SECTION_NOTYPE
29837 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29839 PR middle-end/80423
29840 * tree.h (build_array_type): Add typeless_storage default argument.
29841 * tree.c (type_cache_hasher::equal): Also compare
29842 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
29843 (build_array_type): Add typeless_storage argument, set
29844 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
29846 (build_nonshared_array_type): Adjust build_array_type_1 caller.
29847 (build_array_type): Likewise. Add typeless_storage argument.
29849 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29850 Jakub Jelinek <jakub@redhat.com>
29852 PR tree-optimization/80426
29853 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
29854 operation on symbolic operands, also compute the overflow for the
29855 invariant part when the operation degenerates into a negation.
29857 2017-04-19 Jakub Jelinek <jakub@redhat.com>
29860 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
29861 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
29864 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
29866 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
29869 * config/avr/avr.c (tree.h): Include it.
29870 (cgraph.h): Include it.
29871 (avr_encode_section_info): Don't warn for uninitialized progmem
29872 variable if it's just an alias.
29874 2017-04-19 Richard Biener <rguenther@suse.de>
29877 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
29878 when needed by AutoPGO.
29880 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
29883 * doc/lto.texi: Remove an extra 'that'.
29885 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
29887 PR rtl-optimization/80429
29888 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
29889 are only used in debug insns.
29891 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29892 Vladimir Makarov <vmakarov@redhat.com>
29894 * config/sparc/predicates.md (input_operand): Add comment. Return
29895 true for any memory operand when LRA is in progress.
29896 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
29898 2017-04-18 Jeff Law <law@redhat.com>
29901 * mips.md ({return,simple_return}_internal): Do not overwrite
29904 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29906 PR tree-optimization/80443
29907 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
29908 instead of adding 1, subtract -1 and similarly instead of subtracting
29911 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
29913 PR rtl-optimization/80357
29914 * haifa-sched.c (tmp_bitmap): New variable.
29915 (model_recompute): Handle duplicate use records.
29916 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
29917 (free_global_sched_pressure_data): Free it.
29919 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
29922 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29923 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29924 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29925 instead of SYSTEM_HEADER_DIR.
29927 2017-04-18 Jeff Law <law@redhat.com>
29929 PR middle-end/80422
29930 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
29931 predecessors after walking up the insn chain.
29933 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29936 * dwarf2out.c (modified_type_die): Try harder not to emit internal
29937 sizetype type into debug info.
29939 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29942 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
29943 unneeded test for TARGET_UPPER_REGS_SF.
29944 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
29946 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29949 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
29950 instead of gsi_after_labels.
29952 2017-04-18 Jeff Law <law@redhat.com>
29954 * regcprop.c (maybe_mode_change): Avoid creating copies of the
29958 2017-04-13 Jeff Law <law@redhat.com>
29959 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
29960 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29962 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
29965 * config/avr/avr.c (intl.h): Include it.
29966 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
29968 2017-04-18 Martin Liska <mliska@suse.cz>
29970 PR gcov-profile/78783
29971 * gcov-tool.c (gcov_output_files): Validate that destination
29972 file is either removed by the tool or by a user.
29974 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29975 Guy Benyei <guybe@mellanox.com>
29977 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
29978 block, and do not negate it, the stored id is already negative.
29980 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29982 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
29984 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29987 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
29988 masks of options that should be turned off if the VSX vector
29989 options are turned off.
29990 (OTHER_P8_VECTOR_MASKS): Likewise.
29991 (OTHER_VSX_VECTOR_MASKS): Likewise.
29992 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
29993 rs6000_disable_incompatible_switches to validate no type switches
29995 (rs6000_incompatible_switch): New function to disallow turning on
29996 other vector options if -mno-vsx, -mno-power8-vector, or
29997 -mno-power9-vector are specified.
29999 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30001 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30003 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30005 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30006 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30007 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30008 (ARG_POINTER_CFA_OFFSET): Likewise.
30010 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30012 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30013 conditions to take advantage of various optimizations.
30015 2017-04-13 Jeff Law <law@redhat.com>
30017 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30018 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30019 (zero_extendsidi2_dext): Likewise.
30021 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30024 * fold-const.c (fold_ternary_loc): Revert
30025 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30028 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30030 PR rtl-optimization/80343
30031 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30032 new scratch pseudo.
30034 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30037 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30038 to ubsan_encode_value.
30040 2017-04-13 Jeff Law <law@redhat.com>
30042 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30043 appearing in DEBUG_INSNs.
30045 2017-04-13 Martin Liska <mliska@suse.cz>
30047 PR gcov-profile/80413
30048 * gcov-io.c (gcov_write_string): Copy to buffer just when
30049 allocated size is greater than zero.
30051 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30054 * dwarf2out.c (decls_for_scope): Ignore declarations of
30055 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30057 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
30060 * ipa-visibility.c (non_local_p): Fix typos.
30061 (localize_node): When localizing symbol in same comdat group,
30062 dissolve the group only when we know external symbols are going
30064 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
30066 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30068 PR tree-optimization/79390
30069 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
30070 order does not result in usable sequence, retry with reversed operand
30076 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
30077 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
30078 op0 instead of fold_convert_loc (loc, type, arg0).
30080 2017-04-12 Jeff Law <law@redhat.com>
30082 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
30083 has a delay slot in the generated code.
30085 * config/cris/cris.md (cris_preferred_reload_class): Return
30086 GENNONACR_REGS rather than GENERAL_REGS.
30088 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30091 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
30092 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
30093 signedness of the result type.
30095 2017-04-12 Richard Biener <rguenther@suse.de>
30096 Jeff Law <law@redhat.com>
30098 PR tree-optimization/80359
30099 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
30100 trim stores to TARGET_MEM_REFs.
30102 2017-04-12 Richard Biener <rguenther@suse.de>
30104 PR tree-optimization/79390
30105 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
30106 threading case even more.
30108 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
30111 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
30112 for quad_address_p for TImode, instead of just not indexed_address.
30114 2017-04-12 Richard Biener <rguenther@suse.de>
30115 Bernd Edlinger <bernd.edlinger@hotmail.de>
30117 PR middle-end/79671
30118 * alias.c (component_uses_parent_alias_set_from): Handle
30119 TYPE_TYPELESS_STORAGE.
30120 (get_alias_set): Likewise.
30121 * tree-core.h (tree_type_common): Add typeless_storage flag.
30122 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
30123 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
30124 for types containing members with TYPE_TYPELESS_STORAGE.
30125 (place_field): Likewise.
30126 (layout_type): Likewise for ARRAY_TYPE.
30127 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
30128 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
30129 TYPE_TYPELESS_STORAGE.
30130 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
30132 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30135 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30136 first argument to type.
30138 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30142 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
30143 CONST0_RTX (mode) rather than const0_rtx where appropriate.
30144 (rs6000_expand_binop_builtin): Likewise.
30145 (rs6000_expand_ternop_builtin): Likewise; also add missing
30146 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
30147 vshasigma built-ins.
30148 * doc/extend.texi: Document that vec_xxpermdi's third argument
30149 must be a constant.
30151 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
30153 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
30154 Use shift_const cost parameter when calculating gain of STV shifts.
30156 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
30158 PR rtl-optimization/70478
30159 * lra-constraints.c (process_alt_operands): Check memory for
30160 disfavoring memory insn operand.
30162 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30164 PR middle-end/80100
30165 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
30166 left shift in unsigned HOST_WIDE_INT type.
30168 PR rtl-optimization/80385
30169 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
30170 (not (neg X)) into (plus X -1) for complex or non-integral modes.
30173 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
30174 if they have any depend clauses.
30176 2017-04-11 Martin Liska <mliska@suse.cz>
30179 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
30180 * ipa-split.c (split_function): Create a local comdat symbol
30181 if caller is in a comdat group.
30183 2017-04-11 Martin Liska <mliska@suse.cz>
30186 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
30189 2017-04-11 Martin Sebor <msebor@redhat.com>
30191 PR middle-end/80364
30192 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
30193 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
30195 (directive::set_width, directive::set_precision, format_character):
30197 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
30200 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
30203 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
30204 conflict, set target->arch_name instead of target->cpu_name.
30206 2017-04-11 Richard Biener <rguenther@suse.de>
30208 PR tree-optimization/80374
30209 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
30210 build_zero_cst, remove fold_convertible_p check again.
30212 2017-04-11 Martin Liska <mliska@suse.cz>
30215 * ubsan.c (instrument_object_size): Do not instrument register
30218 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30221 * config/i386/i386-builtin-types.def
30222 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
30223 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
30224 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
30225 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
30226 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
30227 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
30228 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
30229 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
30230 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
30231 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
30232 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
30233 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
30234 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
30235 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
30236 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
30237 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
30238 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
30239 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
30240 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
30241 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
30242 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
30243 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
30244 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
30245 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
30246 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
30247 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
30248 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
30249 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
30250 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
30251 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
30252 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
30253 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
30254 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
30255 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
30256 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
30257 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
30258 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
30259 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
30260 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
30261 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
30262 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
30263 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
30264 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
30265 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
30266 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
30267 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
30269 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
30270 flag to second_arg_count, handle 4 argument function type _COUNT
30271 aliases, handle second_arg_count on second argument rather than last.
30273 2017-04-10 Jeff Law <law@redhat.com>
30275 PR tree-optimization/80374
30276 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
30277 record anything if we can not convert integer_zero_node to the
30280 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30283 * config/rs6000/rs6000.c (rs6000_option_override_internal):
30284 Enhance special handling given to the TARGET_P9_MINMAX option in
30285 relation to certain other options.
30287 2017-04-10 Bin Cheng <bin.cheng@arm.com>
30289 PR tree-optimization/80153
30290 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
30291 remove POINTER_PLUS_EXPR's base part directly, rather than through
30294 2017-04-10 Richard Biener <rguenther@suse.de>
30295 Bin Cheng <bin.cheng@arm.com>
30297 PR tree-optimization/80153
30298 * tree-affine.c (aff_combination_to_tree): Get base pointer from
30299 the first element of pointer type aff_tree. Build result expr in
30301 (add_elt_to_tree): Convert to type unconditionally. Remove other
30302 fold_convert calls.
30303 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
30304 (rewrite_use_nonlinear_expr): Check invariant using iv information.
30306 2017-04-10 Richard Biener <rguenther@suse.de>
30308 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30311 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
30313 PR rtl-optimization/70478
30314 * lra-constraints.c (curr_small_class_check): New.
30315 (update_and_check_small_class_inputs): New.
30316 (process_alt_operands): Update curr_small_class_check. Disfavor
30317 alternative insn memory operands. Check available regs for small
30320 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30323 * config/mips/mips.opt (-mvirt): Update description.
30324 * doc/invoke.texi (-mvirt): Likewise.
30326 2017-04-10 Richard Biener <rguenther@suse.de>
30328 PR middle-end/80362
30329 * fold-const.c (fold_binary_loc): Look at unstripped ops when
30330 looking for NEGATE_EXPR in -A / -B to A / B folding.
30332 2017-04-10 Martin Liska <mliska@suse.cz>
30334 PR gcov-profile/80224
30335 * gcov.c (print_usage): Fix usage string.
30336 (get_gcov_intermediate_filename): Remove.
30337 (output_gcov_file): Use both for normal and intermediate format.
30338 (generate_results): Do not initialize special file for
30339 intermediate format.
30341 2017-04-10 Richard Biener <rguenther@suse.de>
30343 PR tree-optimization/80304
30344 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
30347 2017-04-10 Nathan Sidwell <nathan@acm.org>
30350 * config/rs6000/rs6000.c (rs6000_vector_type): New.
30351 (rs6000_init_builtins): Use it.
30353 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30355 * config/arm/arm.md (<mrc>): Add mode to SET source.
30356 (<mrrc>): Likewise.
30358 2017-04-10 Richard Biener <rguenther@suse.de>
30360 PR middle-end/80344
30361 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
30363 2017-04-10 Jakub Jelinek <jakub@redhat.com>
30366 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
30367 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
30368 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
30369 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
30370 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
30371 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
30372 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
30373 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
30374 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
30375 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
30376 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
30377 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
30378 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
30379 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
30380 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
30381 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
30382 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
30383 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
30384 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
30385 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
30386 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
30387 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
30388 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
30390 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
30392 PR rtl-optimization/70478
30393 * lra-constraints.c: Reverse the last patch.
30395 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
30397 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
30398 Add comment for WCHAR_T.
30400 2017-04-08 Martin Liska <mliska@suse.cz>
30403 2017-04-07 Martin Liska <mliska@suse.cz>
30406 * ipa-split.c (split_function): Add function part to a same comdat
30409 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30412 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
30414 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
30416 * rs6000/rs6000.c (vec_load_pendulum): Rename...
30417 (vec_pairing): ...to this.
30418 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
30419 (rs6000_sched_init): Adjust for name change.
30420 (struct rs6000_sched_context): Likewise.
30421 (rs6000_init_sched_context): Likewise.
30422 (rs6000_set_sched_context): Likewise.
30424 2017-04-07 Jakub Jelinek <jakub@redhat.com>
30430 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
30432 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
30433 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
30434 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
30436 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
30438 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
30440 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30442 PR rtl-optimization/70703
30443 * ira-color.c (update_conflict_hard_regno_costs): Use
30444 int64_t instead of HOST_WIDE_INT.
30446 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30448 PR rtl-optimization/70478
30449 * lra-constraints.c (process_alt_operands): Disfavor alternative
30450 insn memory operands.
30452 2017-04-07 Jeff Law <law@redhat.com>
30454 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
30455 CALL and NOTE_INSN_CALL_ARG_LOCATION.
30457 2017-04-07 Martin Liska <mliska@suse.cz>
30460 * config/aarch64/aarch64.c (aarch64_process_target_attr):
30461 Show error message instead of an ICE.
30463 2017-04-07 Martin Liska <mliska@suse.cz>
30466 * ipa-split.c (split_function): Add function part to a same comdat
30469 2017-04-07 Richard Biener <rguenther@suse.de>
30471 PR middle-end/80341
30472 * tree.c (get_unwidened): Also handle ! for_type case for
30474 * convert.c (do_narrow): Split out from ...
30475 (convert_to_integer_1): ... here. Do not pass final truncation
30476 type to get_unwidened for TRUNC_DIV_EXPR.
30478 2017-04-07 Richard Biener <rguenther@suse.de>
30480 * tree-affine.c (wide_int_ext_for_comb): Take type rather
30482 (aff_combination_const): Adjust.
30483 (aff_combination_scale): Likewise.
30484 (aff_combination_add_elt): Likewise.
30485 (aff_combination_add_cst): Likewise.
30486 (aff_combination_convert): Likewise.
30487 (add_elt_to_tree): Likewise. Remove unused argument.
30488 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
30490 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
30492 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
30494 * config/arm/arm.c (arm_default_short_enums): Use
30495 ARM_DEFAULT_SHORT_ENUMS.
30496 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
30498 2017-04-06 Jakub Jelinek <jakub@redhat.com>
30501 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
30502 members with redundant out-of-class redeclaration.
30504 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30507 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
30508 * config/i386/i386.md (*zero_extendsidi2):
30509 Add (?*x,*x) and (?*v,*v) alternatives.
30511 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30514 * config/i386/i386.c (ix86_expand_builtin)
30515 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
30516 mode from insn data. Convert operands to insn operand mode.
30517 Copy operands that don't satisfy insn predicate to a register.
30519 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
30521 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
30524 2017-04-06 Richard Biener <rguenther@suse.de>
30526 PR tree-optimization/80334
30527 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
30528 preserve alignment of accesses.
30530 2017-04-06 Richard Biener <rguenther@suse.de>
30532 PR tree-optimization/80262
30533 * tree-sra.c (build_ref_for_offset): Preserve address-space
30535 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
30536 Drop useless address-space information on MEM_REF offsets.
30538 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
30540 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
30542 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30544 PR rtl-optimization/70703
30545 * ira-color.c (update_conflict_hard_regno_costs): Use
30546 HOST_WIDE_INT instead of long.
30548 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
30551 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
30552 not defined for x86_64 target. Add -mmmx target option when __SSE2__
30554 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
30555 for x86_64 target. Handle -m3dnowa option.
30557 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30559 PR rtl-optimization/70703
30560 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
30561 (update_conflict_hard_regno_costs): Use long instead of unsigned
30562 arithmetic for cost calculation.
30564 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30565 Bernd Edlinger <bernd.edlinger@hotmail.de>
30568 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
30571 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
30574 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
30575 ptr_mode with Pmode throughout.
30576 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
30577 into probe_stack_range and use DImode.
30579 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
30582 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
30583 call_eh_return is true.
30585 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30587 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
30588 Initialize last_match_fntype_index.
30590 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30593 * tree-nvr.c: Include internal-fn.h.
30594 (pass_return_slot::execute): Ignore internal calls without
30597 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30598 Richard Biener <rguenther@suse.de>
30601 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
30602 captures used multiple times, except for the last use.
30603 * generic-match-head.c: Include gimplify.h.
30605 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30607 PR tree-optimization/79390
30608 * target.h (struct noce_if_info): Declare.
30609 * targhooks.h (default_noce_conversion_profitable_p): Declare.
30610 * target.def (noce_conversion_profitable_p): New target hook.
30611 * ifcvt.h (struct noce_if_info): New type, moved from ...
30612 * ifcvt.c (struct noce_if_info): ... here.
30613 (noce_conversion_profitable_p): Renamed to ...
30614 (default_noce_conversion_profitable_p): ... this. No longer
30616 (noce_try_store_flag_constants, noce_try_addcc,
30617 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
30618 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
30619 instead of noce_conversion_profitable_p.
30620 * config/i386/i386.c: Include ifcvt.h.
30621 (ix86_option_override_internal): Don't override
30622 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
30623 (ix86_noce_conversion_profitable_p): New function.
30624 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
30625 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
30626 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
30627 * doc/tm.texi: Regenerated.
30629 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30631 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
30634 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
30637 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
30638 instructions for small multiply cores.
30640 2017-04-04 Jeff Law <law@redhat.com>
30642 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
30644 (mips_expand_vec_perm_const): Initialize elements in orig_perm
30645 that are not set by the loop over the elements.
30647 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30650 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
30651 int mode, convert_modes it to mode as unsigned, otherwise use
30652 lowpart_subreg to mode rather than SImode.
30653 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
30654 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
30655 Use DImode instead of SImode for the shift count operand.
30656 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
30659 2017-04-04 Richard Biener <rguenther@suse.de>
30661 PR middle-end/80281
30662 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
30663 arithmetic done for the negate or the plus. Simplify.
30664 (A - (-B) -> A + B): Likewise.
30665 * fold-const.c (split_tree): Make sure to not negate pointers.
30667 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
30669 PR rtl-optimization/60818
30670 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
30671 a compare of comparisons with the thing compared if this results
30672 in a different machine mode.
30674 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
30676 * alias.c (base_alias_check): Fix typo in comment.
30677 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
30678 * cgraphunit.c (symbol_table::compile): Likewise.
30679 * collect2.c (maybe_run_lto_and_relink): Likewise.
30680 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
30681 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
30682 * config/avr/avr.c (avr_map_op_t): Likewise.
30683 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
30684 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
30685 * config/epiphany/epiphany.md (movcc): Likewise.
30686 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
30687 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
30689 * config/mips/mips.c (mips_save_restore_reg): Likewise.
30690 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
30691 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
30692 * config/sh/sh.c (sh_rtx_costs): Likewise.
30693 * fold-const.c (fold_truth_andor): Likewise.
30694 * genautomata.c (collapse_flag): Likewise.
30695 * gengtype.h (struct type::u::s): Likewise.
30696 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
30697 * input.c (FORMAT_AMOUNT): Likewise.
30698 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
30699 (known_aggs_to_agg_replacement_list): Likewise.
30700 * ipa-inline-analysis.c: Likewise.
30701 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
30702 * ipa-polymorphic-call.c
30703 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
30704 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
30705 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
30707 * modulo-sched.c (apply_reg_moves): Likewise.
30708 * omp-expand.c (build_omp_regions_1): Likewise.
30709 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
30710 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
30711 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
30712 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
30713 * value-prof.c: Likewise.
30714 * var-tracking.c (val_reset): Likewise.
30716 2017-04-03 Richard Biener <rguenther@suse.de>
30718 PR tree-optimization/80275
30719 * fold-const.c (split_address_to_core_and_offset): Handle
30722 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
30724 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
30725 descriptors is at least equal to that of functions.
30727 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30729 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
30731 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30734 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
30735 (mov<IMOD4:mode>): New expander.
30736 (*mov<IMOD4:mode>_internal): New insn and split pattern.
30738 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
30740 PR rtl-optimization/79405
30741 * fwprop.c (propagations_left): New variable.
30742 (forward_propagate_into): Decrement it.
30743 (fwprop_init): Initialize it.
30744 (fw_prop): If the variable has reached zero, stop propagating.
30745 (fwprop_addr): Ditto.
30747 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30750 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
30751 a FUNCTION_DECL, pass it as decl instead of origin to
30754 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
30756 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
30759 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
30762 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
30763 TARGET_VSX_SMALL_INTEGER.
30765 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30767 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30768 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
30770 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30772 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
30773 extraction from odd-numbered MSA register.
30775 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30777 PR middle-end/80173
30778 * expmed.c (store_bit_field_1): Don't attempt to create
30779 a word subreg out of hard registers wider than word if they
30780 have HARD_REGNO_NREGS of 1 for their mode.
30782 PR middle-end/80163
30783 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
30784 conversions to integer types wider than word and pointer.
30787 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
30788 (rtx_equal_for_cselib_p): Pass 0 to it.
30789 * cselib.c (cselib_hasher::equal): Likewise.
30790 (rtx_equal_for_cselib_1): Add depth argument. If depth
30791 is 128, don't look up VALUE locs and punt. Increment
30792 depth in recursive calls when walking VALUE locs.
30794 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
30796 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
30797 (make_gcov_file_name): Use the canonical path name for generating
30799 (read_line): Fix handling of files with ascii null bytes.
30801 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
30803 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
30804 to initialise a vector register instead
30805 of using a const_int.
30807 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30809 PR translation/80189
30810 * gimplify.c (omp_default_clause): Use %qs instead of %s in
30811 diagnostic messages.
30813 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
30816 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
30817 (dfp_diex_<mode>): Update mode of operand 1.
30818 * doc/extend.texi (dxex, dxexq): Document change to return type.
30819 (diex, diexq): Document change to argument type.
30821 2017-03-30 Martin Jambor <mjambor@suse.cz>
30824 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
30825 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
30826 it reflects the signature changes performed at the callee side.
30827 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
30828 to cgraph_build_function_type_skip_args.
30829 (build_function_decl_skip_args): Adjust call to the above function.
30831 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30834 * config/i386/sse.md
30835 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
30836 register as dest whenever it is a MEM not rtx_equal_p to the
30837 corresponding dup operand, and when forcing into reg move the
30838 reg into the memory afterwards.
30839 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
30840 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
30841 for the force_reg mode.
30842 (avx512vl_vextractf128<mode>): Use register as dest either
30843 always when a MEM, or when it is a MEM not rtx_equal_p to the
30844 corresponding dup operand, or even not when it is a CONST_VECTOR
30845 depending on the mode and lo vs. hi.
30846 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
30848 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
30849 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
30850 Likewise. Require that operands[2] is even.
30851 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
30852 Remove extraneous parens. Require that operands[2] is a multiple
30854 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
30855 operands[0] is a MEM if <mask_applied>, the predicates/constraints
30856 disallow memory then.
30858 2017-03-30 Richard Biener <rguenther@suse.de>
30860 PR tree-optimization/77498
30861 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
30862 to non-constants over backedges.
30864 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
30866 PR rtl-optimization/80233
30867 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
30868 as last_combined_insn. Do not test for BARRIER_P separately.
30870 2017-03-29 Andreas Schwab <schwab@suse.de>
30873 * calls.c (prepare_call_address): Convert funexp to Pmode before
30874 copying to temp reg.
30876 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30878 PR tree-optimization/80158
30879 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
30880 Handle possible future case of more than one alternate
30882 (replace_rhs_if_not_dup): Likewise.
30883 (replace_one_candidate): Likewise.
30885 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
30887 PR rtl-optimization/80193
30888 * ira.c (ira): Do not check allocation for LRA.
30890 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
30892 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
30893 (nvptx_output_simt_exit): Declare.
30894 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
30895 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
30896 (init_softstack_frame): Move initialization of crtl->is_leaf to...
30897 (nvptx_declare_function_name): ...here. Emit declaration of local
30898 memory space buffer for omp_simt_enter insn.
30899 (nvptx_output_unisimt_switch): New.
30900 (nvptx_output_softstack_switch): New.
30901 (nvptx_output_simt_enter): New.
30902 (nvptx_output_simt_exit): New.
30903 * config/nvptx/nvptx.h (struct machine_function): New fields
30904 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
30905 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
30906 (UNSPECV_SIMT_EXIT): Ditto.
30907 (omp_simt_enter_insn): New insn.
30908 (omp_simt_enter): New expansion.
30909 (omp_simt_exit): New insn.
30910 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
30912 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
30913 (expand_GOMP_SIMT_ENTER_ALLOC): New.
30914 (expand_GOMP_SIMT_EXIT): New.
30915 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
30916 (GOMP_SIMT_ENTER_ALLOC): Ditto.
30917 (GOMP_SIMT_EXIT): Ditto.
30918 * target-insns.def (omp_simt_enter): New insn.
30919 (omp_simt_exit): Ditto.
30920 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
30922 (lower_rec_simd_input_clauses): Implement SIMT privatization.
30923 (lower_rec_input_clauses): Likewise.
30924 (lower_lastprivate_clauses): Handle SIMT privatization.
30926 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
30927 (ompdevlow_adjust_simt_enter): New.
30928 (find_simtpriv_var_op): New.
30929 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
30930 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
30932 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
30933 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
30934 (copy_decl_for_dup_finish): Ditto.
30936 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
30938 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
30941 * config/i386/i386.c (ix86_option_override_internal): Always
30942 allow -mpreferred-stack-boundary=3 for 64-bit targets.
30944 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30946 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
30948 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30950 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
30951 mark new edge's irreducible flag accordign to it.
30952 (vect_do_peeling): Check loop preheader edge's irreducible flag
30953 and pass it to function slpeel_add_loop_guard.
30955 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
30957 PR tree-optimization/80218
30958 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30959 Update block frequencies and counts.
30961 2017-03-28 Richard Biener <rguenther@suse.de>
30963 PR tree-optimization/78644
30964 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
30965 of a simplification result we may not use it at all.
30967 2017-03-28 Richard Biener <rguenther@suse.de>
30970 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
30971 without arguments, generate default definition of a SSA name.
30973 2017-03-28 Richard Biener <rguenther@suse.de>
30975 PR middle-end/80222
30976 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
30977 TYPE_REF_CAN_ALIAS_ALL references.
30978 * fold-const.c (fold_indirect_ref_1): Likewise.
30980 2017-03-28 Martin Liska <mliska@suse.cz>
30983 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
30984 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
30986 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30987 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
30989 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
30990 (EXTRA_SPECS): Define.
30991 (SUBTARGET_EXTRA_SPECS): Likewise.
30992 (SUBTARGET_CPP_SPEC): Likewise.
30993 * config/arc/elf.h (EXTRA_SPECS): Renamed to
30994 SUBTARGET_EXTRA_SPECS.
30995 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
30997 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
30999 * config/arc/simdext.md (vst64_insn): Update pattern.
31000 (vld32wh_insn): Likewise.
31001 (vld32wl_insn): Likewise.
31002 (vld64_insn): Likewise.
31003 (vld32_insn): Likewise.
31005 2017-03-28 Marek Polacek <polacek@redhat.com>
31008 * fold-const.c (fold_comparison): Use protected_set_expr_location
31009 instead of SET_EXPR_LOCATION.
31011 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31013 * tree.c (add_expr): Avoid name lookup warning.
31015 2017-03-27 Jeff Law <law@redhat.com>
31017 PR tree-optimization/80216
31018 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31019 function name. Limit recursion depth.
31020 (record_temporary_equivalences): Corresponding changes.
31022 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31024 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31027 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31030 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31032 * cfgcleanup.c (reg_note_cfa_p): New array.
31033 (insns_have_identical_cfa_notes): New function.
31034 (old_insns_match_p): Don't cross-jump in between /f
31035 and non-/f instructions. If both i1 and i2 are frame related,
31036 verify all CFA notes, their order and content.
31038 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31041 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31042 HImode and SImode with zero extend to DImode to one insn.
31043 (bswap<mode>2_extenddi): Likewise.
31044 (bswapsi2_extenddi): Likewise.
31045 (bswaphi2_extendsi): Likewise.
31046 (bswaphi2): Combine bswap HImode and SImode into one insn.
31047 Separate memory insns from swapping register.
31048 (bswapsi2): Likewise.
31049 (bswap<mode>2): Likewise.
31050 (bswaphi2_internal): Delete, no longer used.
31051 (bswapsi2_internal): Likewise.
31052 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31053 store, and gpr<-gpr swap insns.
31054 (bswap<mode>2_store): Likewise.
31055 (bswaphi2_reg): Register only splitter, combine with the splitter.
31056 (bswaphi2 splitter): Likewise.
31057 (bswapsi2_reg): Likewise.
31058 (bswapsi2 splitter): Likewise.
31059 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
31060 the insns into load, store, and register/register insns.
31061 (bswapdi2_ldbrx): Likewise.
31062 (bswapdi2_load): Likewise.
31063 (bswapdi2_store): Likewise.
31064 (bswapdi2_reg): Likewise.
31066 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
31068 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
31069 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
31071 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
31074 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
31076 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31077 special handling for target option conflicts between dform
31078 options (-mpower9-dform, -mpower9-dform-vector,
31079 -mpower9-dform-scalar) and -mno-direct-move.
31081 2017-03-27 Richard Biener <rguenther@suse.de>
31083 PR tree-optimization/80181
31084 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
31086 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31088 * config/arc/predicates.md (move_double_src_operand): Replace the
31089 call to move_double_src_operand with a call to address_operand.
31091 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31093 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
31094 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
31095 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
31097 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31099 * config/arc/predicates.md (long_immediate_loadstore_operand):
31100 Consider scaled addresses cases.
31102 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31104 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
31105 restored when in interrupt.
31106 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
31107 doesn't have delay slot.
31109 2017-03-27 Richard Biener <rguenther@suse.de>
31112 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
31113 inlined thunk clones.
31115 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31118 * asan.c (instrument_derefs): Copy over last operand from
31119 original COMPONENT_REF to the new COMPONENT_REF with
31120 DECL_BIT_FIELD_REPRESENTATIVE.
31121 * ubsan.c (instrument_object_size): Likewise.
31123 2017-03-27 Richard Biener <rguenther@suse.de>
31125 PR tree-optimization/80170
31126 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
31127 sure DR/SCEV didnt fold in constants we do not see when looking
31128 at the reference base alignment.
31130 2017-03-27 Richard Biener <rguenther@suse.de>
31132 PR middle-end/80171
31133 * gimple-fold.c (fold_ctor_reference): Properly guard against
31134 NULL return value from canonicalize_constructor_val.
31136 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
31139 * config/i386/i386.c (ix86_expand_builtin)
31140 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
31141 flags reg setting and flags reg using instructions.
31142 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
31143 clobbering instructions to zero extend op2.
31145 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
31147 * doc/install.texi (Configuration) <--with-aix-soname>:
31148 Update link to AIX ld.
31150 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
31152 PR rtl-optimization/80160
31153 PR rtl-optimization/80159
31154 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
31155 reg_alternate_class into account.
31157 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
31160 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
31161 to consider in curr_insn_transform.
31163 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31165 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
31166 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
31167 and emit_mode_inner.
31169 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31171 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
31172 argument to the overloaded builtin variants. Use the new flag to
31173 deprecate certain builtin variants.
31174 * config/s390/s390-builtin-types.def: Add new builtin types.
31175 * config/s390/s390-builtins.h: Support new flags field for
31176 overloaded builtins.
31177 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
31178 (s390_macro_to_expand): Enable vector float data type.
31179 (s390_cpu_cpp_builtins_internal): Indicate support of the new
31180 builtins by incrementing the __VEC__ version number.
31181 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
31183 (s390_resolve_overloaded_builtin): Emit error messages depending
31184 on the builtin flags.
31185 * config/s390/s390.c (s390_expand_builtin): Support additional
31186 flags argument. Change error message to match the messages
31187 emitted in s390-c.c.
31188 * config/s390/s390.md: New UNSPEC_* constants.
31189 (op_type): Add new instruction types.
31190 * config/s390/vecintrin.h: Add new builtins and test data class
31192 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
31193 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
31194 (VEC_INEXACT, VEC_NOINEXACT): New constants.
31195 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
31196 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
31197 ("vec_mergel<mode>"): V_HW -> VEC_HW.
31199 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
31200 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
31201 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
31202 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
31204 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
31205 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
31206 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
31207 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
31209 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
31210 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
31211 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
31212 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
31213 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
31214 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
31215 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
31217 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
31218 ("vec_scatter_element<V_HW_4:mode>_DI")
31219 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
31220 ("vec_fpint<mode>", "vflls")
31221 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
31222 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
31223 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
31224 ("*vec_cmphe<mode>_cc"): ... these.
31226 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
31227 mode constant instead of magic value.
31229 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31231 * config/s390/s390.c (s390_expand_vec_compare): Support other
31232 vector floating point modes than just V2DF.
31233 (s390_expand_vcond): Likewise.
31234 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
31235 (s390_cannot_change_mode_class): Prevent mode changes between TF
31236 and V1TF in vector registers.
31237 * config/s390/s390.md (DF, SF): New mode attributes.
31238 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
31239 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
31240 SFmode support for VRs.
31241 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
31243 (VFT, VF_HW): New mode iterators.
31244 (vw, sdx): New mode attributes.
31245 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
31246 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
31247 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
31248 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
31249 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
31250 also the new vector floating point modes. Renaming to ...
31252 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
31253 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
31254 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
31255 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
31256 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
31257 ("vec_unordered<mode>"): ... these.
31259 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
31260 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
31261 ("*vec_extendv2df"): New insn definitions.
31263 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31265 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
31266 ("mulditi3_2", "*muldi3_sign"): New patterns.
31267 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
31268 rename the pattern definition.
31270 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31272 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
31274 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
31276 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31278 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
31279 instruction if possible.
31280 * config/s390/vector.md (vec_halfnumelts): New mode
31282 ("*vec_vllezlf<mode>"): New pattern.
31284 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31286 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
31287 ("popcountv4si2", "popcountv2di2"): Rename to ...
31288 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
31289 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
31291 ("popcount<mode>2_vxe"): New pattern.
31293 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31295 * common/config/s390/s390-common.c (processor_flags_table): Add
31297 * config.gcc: Add arch12.
31298 * config/s390/driver-native.c (s390_host_detect_local_cpu):
31299 Default to arch12 for unknown CPU model numbers.
31300 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
31301 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
31302 PROCESSOR_max sanity check.
31303 * config/s390/s390-opts.h (enum processor_type): Add
31305 * config/s390/s390.c (processor_table): Add arch12.
31306 (s390_expand_builtin): Add check for B_VXE flag.
31307 (s390_issue_rate): Add PROCESSOR_ARCH12.
31308 (s390_get_sched_attrmask): Likewise.
31309 (s390_get_unit_mask): Likewise.
31310 (s390_sched_score): Enable z13 scheduling for arch12.
31311 (s390_sched_reorder): Likewise.
31312 (s390_sched_variable_issue): Likewise.
31313 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
31315 (s390_tune_attr): Use z13 scheduling also for arch12.
31316 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
31317 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
31318 (TARGET_VXE_P): New macros.
31319 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
31320 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
31321 * config/s390/s390.opt: Add arch12 as processor_type.
31323 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31325 * config/s390/s390.md
31326 ("fixuns_truncdddi2", "fixuns_trunctddi2")
31327 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
31328 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
31330 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
31331 Rename expanders to ...
31333 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
31334 ("fixuns_truncdddi2_emu"): ... these.
31336 ("fixuns_trunc<mode>si2_emu"): New expander.
31338 ("*fixuns_truncdfdi2_z13"): Rename to ...
31339 ("*fixuns_truncdfdi2_vx"): ... this.
31341 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31343 * config/s390/2964.md: Remove the single element vector compare
31344 instructions which are no longer used.
31345 * config/s390/s390.c (s390_select_ccmode): Remove handling of
31347 (s390_canonicalize_comparison): Remove handling of DFmode
31349 (s390_expand_vec_compare_scalar): Remove function.
31350 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
31351 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
31353 ("*cmp<mode>_ccs"): Add wfcdb instruction.
31355 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31357 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
31359 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
31360 will anyway by matched by mov<mode>_64dfp.
31362 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31364 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
31365 vlef/vstef. Add missing operand to vleif.
31367 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31369 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
31370 pair for all vector types with 64 bit elements.
31371 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
31372 * config/s390/vector.md (V_HW_64): ... here.
31373 (V_128_NOSINGLE): New mode iterator.
31374 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
31375 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
31376 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
31377 ("*vec_load_pairv2di"): Change to ...
31378 ("*vec_load_pair<mode>"): ... this one.
31380 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31382 * config/s390/constraints.md: Add comments.
31383 (jKK): Reject element sizes > 8 bytes.
31384 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
31386 * config/s390/s390.md: Add the s_operand checks formerly in
31387 s390_split_ok_p to various splitters where they are still
31389 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
31390 for 128 bit vectors. Plus two splitters.
31392 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31394 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
31397 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31400 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
31401 error if the boundary argument is not constant.
31403 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31405 PR rtl-optimization/80112
31406 * loop-doloop.c (doloop_condition_get): Don't check condition
31407 if cmp isn't SET with IF_THEN_ELSE src.
31409 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31411 PR tree-optimization/80158
31412 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
31413 replacing a candidate statement, also replace it for the
31414 candidate's alternate interpretation.
31415 (replace_rhs_if_not_dup): Likewise.
31416 (replace_one_candidate): Likewise.
31418 2017-03-24 Richard Biener <rguenther@suse.de>
31420 PR tree-optimization/80167
31421 * graphite-isl-ast-to-gimple.c
31422 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
31424 (translate_isl_ast_to_gimple::get_rename): Likewise.
31426 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31428 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
31429 handling of certain combinations of target options, including the
31430 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
31431 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
31433 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31436 * config/arm/arm.md (*load_multiple): Add reload_completed to
31437 matching condition.
31439 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31440 Richard Biener <rguenth@suse.de>
31442 PR tree-optimization/79908
31443 PR tree-optimization/80136
31444 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31445 been cast away, gimplify_and_add suffices.
31447 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
31449 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
31451 2017-03-23 Richard Biener <rguenther@suse.de>
31453 PR tree-optimization/80032
31454 * gimplify.c (gimple_push_cleanup): Forced unconditional
31455 cleanups still have to go to the conditional_cleanups
31458 2017-03-22 Jakub Jelinek <jakub@redhat.com>
31460 PR tree-optimization/80072
31461 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
31463 (next_operand_entry_id): Change type to unsigned int.
31464 (sort_by_operand_rank): Make sure to return the right return value
31465 even if unsigned fields are bigger than INT_MAX.
31466 (struct oecount): Change cnt and id type to unsigned int.
31467 (oecount_hasher::equal): Formatting fix.
31468 (oecount_cmp): Make sure to return the right return value
31469 even if unsigned fields are bigger than INT_MAX.
31470 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
31473 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
31474 TREE_READONLY on result if writing it more than once.
31477 * doc/invoke.texi (-fsanitize=thread): Document that with
31478 -fnon-call-exceptions atomics are not able to throw
31482 * tsan.c: Include tree-eh.h.
31483 (instrument_builtin_call): Call maybe_clean_eh_stmt or
31484 maybe_clean_or_replace_eh_stmt where needed.
31485 (instrument_memory_accesses): Add cfg_changed argument.
31486 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
31487 if it returned true.
31488 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
31490 PR rtl-optimization/63191
31491 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
31492 wrapper function, moved the whole old content into ...
31493 (ix86_delegitimize_address_1): ... this. New inline function.
31494 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
31495 true as last argument instead of ix86_delegitimize_address.
31497 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31499 * config/aarch64/aarch64.c (generic_branch_cost): Copy
31500 cortexa57_branch_cost.
31502 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31504 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
31506 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31509 * doc/md.texi (Constraints): Document wA constraint.
31510 * config/rs6000/constraints.md (wA): New.
31511 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
31512 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
31513 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
31514 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
31516 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
31519 * gimplify.c (is_oacc_declared): New function.
31520 (oacc_default_clause): Use it to set default flags for acc declared
31521 variables inside parallel regions.
31522 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
31523 declared variables.
31524 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
31525 declare attribute to any decl as necessary.
31527 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
31530 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
31531 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
31532 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
31533 (arm_arch_lpae): This.
31534 * config/arm/arm.c (arm_arch7ve): Rename into ...
31535 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
31536 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
31539 2017-03-22 Martin Liska <mliska@suse.cz>
31542 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
31543 error message instead of an ICE.
31545 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31547 * doc/extend.texi (6.11 Additional Floating Types): Revise.
31549 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
31551 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31553 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31556 2017-03-21 Martin Sebor <msebor@redhat.com>
31558 * doc/extend.texi: Use "cannot" instead of "can't."
31559 * doc/hostconfig.texi: Same.
31560 * doc/install.texi: Same.
31561 * doc/invoke.texi: Same.
31562 * doc/loop.texi: Same.
31563 * doc/md.texi: Same.
31564 * doc/objc.texi: Same.
31565 * doc/rtl.texi: Same.
31566 * doc/tm.texi: Same.
31567 * doc/tm.texi.in: Same.
31568 * doc/trouble.texi: Same.
31570 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
31573 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
31574 (collect_checksum_attributes): Set it.
31575 (die_checksum_ordered): Use it.
31577 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31579 PR tree-optimization/79908
31580 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
31581 change: For a VA_ARG whose LHS has been cast away, use
31582 force_gimple_operand to construct the side effects.
31584 2017-03-21 David Malcolm <dmalcolm@redhat.com>
31586 PR translation/80001
31587 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
31588 more amenable to translation.
31589 (oacc_loop_auto_partitions): Likewise.
31591 2017-03-21 Marek Polacek <polacek@redhat.com>
31592 Martin Sebor <msebor@redhat.com>
31594 PR tree-optimization/80109
31595 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
31596 on INTEGRAL_TYPE_P.
31598 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31599 Segher Boessenkool <segher@kernel.crashing.org>
31602 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
31603 check reg_used_between_p between insn and one of succ or succ2
31604 depending on if succ is artificial insn not inserted into insn
31607 2017-03-21 Martin Liska <mliska@suse.cz>
31609 PR gcov-profile/80081
31610 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
31611 * doc/gcc.texi: Include gcov-dump stuff.
31612 * doc/gcov-dump.texi: New file.
31614 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
31616 PR rtl-optimization/79150
31617 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
31618 conditional jump, if the jump is the last insn of the loop.
31620 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31621 Richard Biener <rguenth@suse.de>
31623 PR tree-optimization/79908
31624 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31625 been cast away, use force_gimple_operand to construct the side
31628 2017-03-21 Martin Liska <mliska@suse.cz>
31630 PR libfortran/79956
31631 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
31634 2017-03-21 Brad Spengler <spender@grsecurity.net>
31637 * plugin.c (htab_hash_plugin): New function.
31638 (add_new_plugin): Use it and adjust.
31639 (parse_plugin_arg_opt): Adjust.
31640 (init_one_plugin): Likewise.
31642 2017-03-21 Richard Biener <rguenther@suse.de>
31644 PR tree-optimization/80032
31645 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
31646 if set force the cleanup to happen unconditionally.
31647 (gimplify_target_expr): Push inserted clobbers with force_uncond
31648 to avoid them being removed by control-dependent DCE.
31650 2017-03-21 Richard Biener <rguenther@suse.de>
31652 PR tree-optimization/80122
31653 * tree-inline.c (copy_bb): Do not expans va-arg packs or
31654 va_arg_pack_len when the inlined call stmt requires pack
31656 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
31658 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31661 * tsan.c (instrument_builtin_call): If the memory model argument
31662 is not a constant, assume it is valid.
31665 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
31668 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31670 PR rtl-optimization/79910
31671 * combine.c (can_combine_p): Do not allow combining an I0 or I1
31672 if its dest is used by an insn before I2 (other than the combined
31673 insns themselves, which are properly handled already).
31675 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31678 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31680 * combine.c (record_used_regs): New static function.
31681 (try_combine): Handle situations where there is an additional
31682 instruction between I2 and I3 which needs to have a LOG_LINK
31686 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31688 * combine.c (try_combine): Delete redundant i1 test. Call
31689 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31691 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31694 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
31695 alternatives 13/14.
31697 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31699 PR tree-optimization/80054
31700 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
31701 the optimization if a PHI or any of its arguments is not dominated
31702 by the candidate's basis. Use gphi* rather than gimple* as
31704 (replace_profitable_candidates): Clean up a gimple* variable that
31705 should be a gphi* variable.
31707 2017-03-20 Martin Sebor <msebor@redhat.com>
31710 * doc/extend.texi (attribute constructor): Document present limitation.
31712 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
31715 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
31716 __POWER9_VECTOR__ #ifdef control, change template definition to
31717 use Power9-specific built-in function.
31718 (vec_any_eq): Likewise.
31719 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
31720 to control outcomes from this test.
31721 (vector_ae_<mode>p): For VEC_F modes, likewise.
31723 2017-03-20 Ian Lance Taylor <iant@google.com>
31725 * config/i386/i386.c (ix86_function_regparm): Save an extra
31726 register for -fsplit-stack with DECL_STATIC_CHAIN.
31728 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31731 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
31732 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
31734 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31736 * config/riscv/riscv.c (riscv_print_operand): Use "fence
31738 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
31741 2017-03-20 Marek Polacek <polacek@redhat.com>
31744 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
31746 2017-03-20 Richard Biener <rguenther@suse.de>
31748 PR tree-optimization/80113
31749 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
31750 allocate extra SSA name for PHI def.
31751 (add_close_phis_to_outer_loops): Likewise.
31752 (add_close_phis_to_merge_points): Likewise.
31753 (copy_loop_close_phi_args): Likewise.
31754 (copy_cond_phi_nodes): Likewise.
31756 2017-03-20 Martin Liska <mliska@suse.cz>
31758 PR middle-end/79753
31759 * tree-chkp.c (chkp_build_returned_bound): Do not build
31760 returned bounds for a LHS that's not a BOUNDED_P type.
31762 2017-03-20 Martin Liska <mliska@suse.cz>
31766 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
31767 COMPLEX_CST and VECTOR_CST.
31769 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31772 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
31773 target operand. A new splitter adds the clobber statement in case
31774 the target operand is dead anyway.
31776 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
31778 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
31779 to age-old versions of binutils and glibc.
31781 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
31783 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
31785 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31787 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
31789 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31791 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
31792 requirement for binutils 2.13.
31794 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31796 * combine.c (try_combine): Delete redundant i1 test. Call
31797 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31799 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
31801 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
31802 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
31804 <riscv64-*-elf>: Re-arrange section
31805 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
31806 <riscv32-*-linux>: Likewise.
31807 <riscv64-*-elf>: Likewise
31808 <riscv64-*-linux>: Likewise.
31810 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
31813 * aarch64.opt(verbose-cost-dump): Fix typo.
31815 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
31818 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
31819 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
31821 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31823 * reload.c (find_reloads): When reloading a nonoffsettable address,
31824 use RELOAD_OTHER for it and its address reloads.
31826 PR rtl-optimization/79910
31827 * combine.c (record_used_regs): New static function.
31828 (try_combine): Handle situations where there is an additional
31829 instruction between I2 and I3 which needs to have a LOG_LINK
31832 2017-03-17 Jeff Law <law@redhat.com>
31834 PR tree-optimization/71437
31835 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
31836 conditional in the hash table first.
31837 (vrp_dom_walker::before_dom_children): Extract condition from
31838 ASSERT_EXPR. Record condition, its inverion and any implied
31839 conditions as well.
31841 2017-03-17 Marek Polacek <polacek@redhat.com>
31842 Markus Trippelsdorf <markus@trippelsdorf.de>
31844 PR tree-optimization/80079
31845 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
31848 2017-03-17 Richard Biener <rguenther@suse.de>
31850 PR middle-end/80075
31851 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
31852 Properly verify the LHS before the RHS possibly claims to be
31854 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
31857 2017-03-17 Martin Jambor <mjambor@suse.cz>
31859 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
31860 (List of -O2 options): Likewise.
31861 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
31864 2017-03-17 Tom de Vries <tom@codesourcery.com>
31866 * gcov-dump.c (print_usage): Print bug_report_url.
31868 2017-03-17 Richard Biener <rguenther@suse.de>
31870 PR middle-end/80050
31871 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
31872 (parser::peek): Likewise.
31874 2017-03-17 Richard Biener <rguenther@suse.de>
31876 PR tree-optimization/80048
31877 * sese.c (free_sese_info): Properly release rename_map and
31878 copied_bb_map elements.
31880 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
31882 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
31883 Add linked-list forward and backlinks. Insert on
31884 construction, remove on destruction.
31885 (class pass_store_merging): Add m_stores_head field.
31886 (pass_store_merging::terminate_and_process_all_chains):
31887 Iterate over m_stores_head list.
31888 (pass_store_merging::terminate_all_aliasing_chains):
31890 (pass_store_merging::execute): Check for debug stmts first.
31891 Push new chains onto the m_stores_head stack.
31893 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
31896 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
31897 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
31898 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
31900 2017-03-16 Jeff Law <law@redhat.com>
31902 PR tree-optimization/71437
31903 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
31904 member function. Implementation moved into after_dom_children
31905 member function and into the threader's thread_outgoing_edges
31907 (dom_opt_dom_walker::after_dom_children): Simplify by moving
31908 some code into new thread_outgoing_edges.
31909 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
31910 definition. Simplify marker handling (do it here). Assume we always
31911 have the available expression and the const/copies tables.
31912 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
31914 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
31915 * tree-vrp.c (equiv_stack): No longer file scoped.
31916 (vrp_dom_walker): New class.
31917 (vrp_dom_walker::before_dom_children): New member function.
31918 (vrp_dom_walker::after_dom_children): Likewise.
31919 (identify_jump_threads): Setup domwalker. Use it rather than
31920 walking edges in a random order by hand. Simplify setup/finalization.
31921 (finalize_jump_threads): Remove.
31922 (vrp_finalize): Do not call identify_jump_threads here.
31923 (execute_vrp): Do it here instead and call thread_through_all_blocks
31926 PR tree-optimization/71437
31927 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
31929 (simplify_stmt_for_jump_threading): Add basic_block argument. All
31931 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
31932 (dom_opt_dom_walker::thread_across_edge): Remove
31933 handle_dominating_asserts argument. All callers changed.
31934 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
31935 changes. Remove calls to lhs_of_dominating_assert. Other
31936 uses of handle_dominating_asserts turn into unconditional code
31937 (simplify_control_stmt_condition_1): Likewise.
31938 (simplify_control_stmt_condition): Likewise.
31939 (thread_through_normal_block, thread_across_edge): Likewise.
31940 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
31941 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
31942 object if it is not an SSA_NAME.
31943 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
31944 before calling into the VRP specific simplifiers.
31945 (identify_jump_threads): Remove handle_dominating_asserts
31948 2017-03-16 Jakub Jelinek <jakub@redhat.com>
31951 * tree-diagnostic.c (default_tree_printer): No longer static.
31952 * tree-diagnostic.h (default_tree_printer): New prototype.
31954 2017-03-16 Tamar Christina <tamar.christina@arm.com>
31956 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
31957 Change ins into fmov.
31959 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31961 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
31962 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
31963 Use h_con constraint for operand 1.
31964 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
31965 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
31967 2017-03-15 Jeff Law <law@redhat.com>
31969 PR tree-optimization/71437
31970 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
31971 (record_temporary_equivalences): Use it.
31973 PR tree-optimization/71437
31974 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
31975 tree-ssa-scopedtables.
31976 (lookup_avail_expr, build_and_record_new_cond): Likewise.
31977 (record_conditions, record_cond, vuse_eq): Likewise.
31978 (record_edge_info): Adjust to API tweak of record_conditions.
31979 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
31980 (record_temporary_equivalences, optimize_stmt): Likewise.
31981 (eliminate_redundant_computations): Likewise.
31982 (record_equivalences_from_stmt): Likewise.
31983 * tree-ssa-scopedtables.c: Include options.h and params.h.
31984 (vuse_eq): New function, moved from tree-ssa-dom.c
31985 (build_and_record_new_cond): Likewise.
31986 (record_conditions): Likewise. Accept vector of conditions rather
31987 than edge_equivalence structure for first argument.
31988 for the first argument.
31989 (avail_exprs_stack::lookup_avail_expr): New member function, moved
31990 from tree-ssa-dom.c.
31991 (avail_exprs_stack::record_cond): Likewise.
31992 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
31993 from tree-ssa-dom.c.
31994 (avail_exprs_stack): Add new member functions lookup_avail_expr
31996 (record_conditions): Declare.
31998 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32001 * lra-constraints.c (process_alt_operands): Increase reject for
32002 reloading an input/output operand.
32004 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32007 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32008 insns to convert from signed/unsigned char/short to IEEE 128-bit
32010 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32012 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32015 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32016 subreg of inner mode for values already in registers.
32018 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32020 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32021 iteration reg is used after the loop.
32023 2017-03-14 Martin Sebor <msebor@redhat.com>
32025 PR tree-optimization/79800
32026 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32027 precision in negative-positive range.
32028 (format_floating): Call non-const overload with adjusted precision.
32030 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32033 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32036 2017-03-14 Martin Sebor <msebor@redhat.com>
32038 PR middle-end/80020
32039 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32040 * builtins.def (aligned_alloc): Use it.
32043 * Makefile.in (GTFILES): Add calls.c.
32044 * calls.c: Include "gt-calls.h".
32046 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32048 PR rtl-optimization/79728
32049 * regs.h (struct target_regs): New field
32050 x_contains_allocatable_regs_of_mode.
32051 (contains_allocatable_regs_of_mode): New macro.
32052 * reginfo.c (init_reg_sets_1): Initialize it, and change
32053 contains_reg_of_mode so it includes global regs as well.
32054 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32055 rather than contains_regs_of_mode.
32057 2017-03-14 Martin Liska <mliska@suse.cz>
32059 * doc/invoke.texi: Document options that can't be combined with
32060 -fcheck-pointer-bounds.
32062 2017-03-14 Martin Liska <mliska@suse.cz>
32064 PR middle-end/79831
32065 * doc/invoke.texi (-Wchkp): Document the option.
32067 2017-03-14 Martin Liska <mliska@suse.cz>
32069 * Makefile.in: Install gcov-dump.
32071 2017-03-14 Martin Liska <mliska@suse.cz>
32073 * multiple_target.c (expand_target_clones): Bail out for
32074 an invalid attribute.
32076 2017-03-14 Richard Biener <rguenther@suse.de>
32078 * alias.c (struct alias_set_entry): Pack properly.
32079 * cfgloop.h (struct loop): Likewise.
32080 * cse.c (struct set): Likewise.
32081 * ipa-utils.c (struct searchc_env): Likewise.
32082 * loop-invariant.c (struct invariant): Likewise.
32083 * lra-remat.c (struct cand): Likewise.
32084 * recog.c (struct change_t): Likewise.
32085 * rtl.h (struct address_info): Likewise.
32086 * symbol-summary.h (function_summary): Likewise.
32087 * tree-loop-distribution.c (struct partition): Likewise.
32088 * tree-object-size.c (struct object_size_info): Likewise.
32089 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
32090 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
32091 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
32092 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
32093 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
32094 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
32095 (struct _stmt_vec_info): Likewise.
32097 2017-03-14 Martin Liska <mliska@suse.cz>
32100 * multiple_target.c (create_dispatcher_calls): Check that
32101 a target can create a function dispatcher.
32103 2017-03-14 Martin Liska <mliska@suse.cz>
32106 * multiple_target.c (expand_target_clones): Drop local.local
32107 flag for default implementation.
32109 2017-03-14 Richard Biener <rguenther@suse.de>
32111 PR tree-optimization/80030
32112 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
32114 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
32116 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
32117 gcc_fallthrough() instead of __attribute__((fallthrough));
32119 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32121 * doc/gcc.texi: Remove "up" link to (DIR).
32122 * doc/gccint.texi: Ditto.
32124 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32126 * doc/install.texi (Specific) <avr>: Remove reference to
32129 2017-03-13 Jeff Law <law@redhat.com>
32131 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
32132 attribute rather than comments.
32134 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
32135 match_scratch operand is highest.
32137 2017-03-13 Martin Liska <mliska@suse.cz>
32139 PR middle-end/78339
32140 * ipa-pure-const.c (warn_function_noreturn): If the declarations
32141 is a CHKP clone, use original declaration.
32143 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32145 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
32146 (arc_conditional_register_usage): Use a different allocation order
32147 when optimizing for size.
32148 * common/config/arc/arc-common.c (arc_option_optimization_table):
32149 Section anchors default on when optimizing for size.
32151 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32153 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
32155 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32157 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
32158 * config/arc/arc.md (cpu_facility): Add cd variant.
32159 (*movqi_insn): Add code density variant.
32160 (*movhi_insn): Likewise.
32161 (*movqi_insn): Likewise.
32162 (*addsi3_mixed): Likewise.
32163 (subsi3_insn): Likewise.
32165 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32167 * config/arc/arc.md (movsi_cond_exec): Update constraint.
32169 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32171 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
32172 expressions with MINUS and UNARY ops.
32174 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32177 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
32179 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
32180 between vec_select and vector argument.
32181 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
32182 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
32183 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
32184 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
32185 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
32186 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
32188 2017-03-13 Richard Biener <rguenther@suse.de>
32191 * params.def (vect-max-peeling-for-alignment): Fix typo.
32193 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32195 * doc/install.texi (Specific) <mips-*-*>: Remove description of
32196 issue that only occurred with binutils below 2.18.
32198 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32200 * doc/install.texi (Specific) <cris-axis-elf>: No longer
32201 refer to binutils 2.11/2.12 minimum.
32203 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32205 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
32206 ftp.kernel.org and simplify binutils requirement.
32208 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
32210 * doc/invoke.texi (Warning Options): Fix spelling of link-time
32212 (Optimize Options): Ditto. Also remove redundancy.
32214 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32216 PR translation/79848
32217 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
32219 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
32220 to G_ to avoid double translation.
32222 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32224 PR translation/79923
32225 * auto-profile.c (get_combined_location): Convert leading
32226 character of diagnostics to lower case and remove trailing period.
32227 (read_profile): Likewise for various diagnostics.
32228 * config/arm/arm.c (arm_option_override): Remove trailing period
32229 from various diagnostics.
32230 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
32231 (msp430_expand_delay_cycles): Likewise.
32233 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32236 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
32237 full command-line argument, rather than just "str".
32238 (aarch64_validate_march): Likewise.
32239 (aarch64_validate_mtune): Likewise.
32241 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
32243 PR rtl-optimization/78911
32244 * lra-assigns.c (must_not_spill_p): New function.
32245 (spill_for): Use it.
32247 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32249 PR tree-optimization/79981
32250 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
32251 ATOMIC_COMPARE_EXCHANGE ifn result.
32252 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
32253 IFN_ATOMIC_COMPARE_EXCHANGE.
32255 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32258 * opts.c (parse_sanitizer_options): Add missing question mark to
32259 "did you mean" message.
32261 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32263 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
32265 (VMULEUH_UNS): Likewise.
32266 (VMULOUB_UNS): Likewise.
32267 (VMULOUH_UNS): Likewise.
32268 * config/rs6000/rs6000.c (builtin_function_type): Remove
32269 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
32271 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32274 * read-rtl-function.c (function_reader::read_rtx_operand): Update
32275 x with result of extra_parsing_for_operand_code_0.
32276 (function_reader::extra_parsing_for_operand_code_0): Convert
32277 return type from void to rtx, returning x. When reading
32278 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
32279 larger size containing struct block_symbol.
32281 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
32283 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
32284 -mfloat128-hardware without -m64.
32286 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
32289 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
32290 entries to the case statement that marks unsigned arguments to
32291 overloaded functions.
32293 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32295 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
32296 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
32298 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
32301 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
32302 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
32304 2017-03-10 Martin Liska <mliska@suse.cz>
32308 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
32309 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
32312 2017-03-10 Olivier Hainque <hainque@adacore.com>
32314 * tree-switch-conversion (array_value_type): Start by resetting
32315 candidate type to it's main variant.
32317 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32319 PR rtl-optimization/79909
32320 * combine.c (try_combine): Use simplify_replace_rtx on individual
32321 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
32322 of the whole CALL_INSN_FUNCTION_USAGE.
32324 PR tree-optimization/79972
32325 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
32326 get_range_info on SSA_NAMEs. Formatting fixes.
32328 2017-03-10 Richard Biener <rguenther@suse.de>
32329 Jakub Jelinek <jakub@redhat.com>
32331 PR tree-optimization/77975
32332 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
32333 edge to be constant.
32334 (get_val_for): For constant x return it. Formatting fix.
32335 (loop_niter_by_eval): Avoid pointless looping if the next iteration
32336 would use the same bases as the current one.
32338 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32340 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
32341 instead of vec_select for V1TImode.
32342 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
32344 (VSX_LE_128): Add V1TI to this mode iterator.
32345 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
32346 (*vsx_le_perm_store_<mode>): Likewise.
32347 (pre-reload splitter for VSX stores): Likewise.
32348 (post-reload splitter for VSX stores): Likewise.
32349 (*vsx_xxpermdi2_le_<mode>): Likewise.
32350 (*vsx_lxvd2x2_le_<mode>): Likewise.
32351 (*vsx_stxvd2x2_le_<mode>): Likewise.
32353 2017-03-09 Michael Eager <eager@eagercon.com>
32355 Correct failures with --enable-checking=yes,rtl.
32357 * config/microblaze/microblaze.c (microblaze_expand_shift):
32358 Replace GET_CODE test with CONST_INT_P and INTVAL test with
32359 test for const0_rtx.
32360 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
32361 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
32363 2017-03-09 Richard Biener <rguenther@suse.de>
32365 PR tree-optimization/79977
32366 * graphite-scop-detection.c (scop_detection::merge_sese):
32367 Handle the case of extra exits to blocks dominating the entry.
32369 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
32371 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
32374 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
32376 PR rtl-optimization/79949
32377 * lra-constraints.c (process_alt_operands): Check memory when
32378 trying to predict a cycle. Print about the overall increase.
32380 2017-03-09 Richard Biener <rguenther@suse.de>
32382 PR middle-end/79971
32383 * gimple-expr.c (useless_type_conversion_p): Preserve
32384 TYPE_SATURATING for fixed-point types.
32386 2017-03-09 Richard Biener <rguenther@suse.de>
32389 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
32390 alignment of BLKmode params.
32392 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32395 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
32396 (VALL_NO_V2Q): Likewise.
32398 * config/aarch64/aarch64-simd.md
32399 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
32401 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
32402 VALL_NO_V2Q mode iterator.
32403 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
32405 2017-03-09 Martin Liska <mliska@suse.cz>
32407 PR tree-optimization/79631
32408 * tree-chkp-opt.c (chkp_is_constant_addr): Call
32409 tree_int_cst_sign_bit just for INTEGER constants.
32411 2017-03-09 Martin Liska <mliska@suse.cz>
32415 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
32418 2017-03-09 Marek Polacek <polacek@redhat.com>
32421 * tree.c (inchash::add_expr): Handle TREE_VEC.
32423 2017-03-09 Martin Liska <mliska@suse.cz>
32426 (chkp_narrow_size_and_offset): New function.
32427 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
32428 (void chkp_parse_bit_field_ref): New function.
32429 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
32430 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
32432 2017-03-09 Martin Liska <mliska@suse.cz>
32435 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
32436 (chkp_find_bounds_1): Remove gcc_unreachable.
32438 2017-03-09 Jakub Jelinek <jakub@redhat.com>
32441 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
32442 BUILT_IN_SYNC*, determine the access type from the size suffix and
32443 always build a MEM_REF with that type. Handle forgotten
32444 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
32447 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
32448 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
32449 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
32450 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
32451 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
32452 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
32453 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
32454 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
32455 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
32456 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
32457 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
32458 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
32459 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
32460 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
32461 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
32462 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
32463 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
32464 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
32465 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
32466 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
32467 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
32468 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
32469 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
32470 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
32471 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
32472 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
32473 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
32474 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
32475 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
32476 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
32477 definitions outside of __OPTIMIZE__ guarded section.
32480 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
32481 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
32482 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
32483 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
32486 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32488 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
32489 ("vfenez<mode>"): Add missing constraints.
32491 2017-03-08 Martin Sebor <msebor@redhat.com>
32494 * config/nds32/nds32.c (nds32_option_override):
32495 Fix misspelled diagnostic.
32497 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32500 * gimplify.c (gimplify_omp_for): Replace index var in outer
32501 taskloop statement with an artificial variable and add
32502 OMP_CLAUSE_PRIVATE clause for it.
32504 2017-03-08 Richard Biener <rguenther@suse.de>
32506 PR tree-optimization/79955
32507 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
32508 for accesses that are completely outside of the variable.
32510 2017-03-08 Andrew Haley <aph@redhat.com>
32512 PR tree-optimization/79943
32513 * tree-ssa-loop-split.c (compute_new_first_bound): When
32514 calculating the new upper bound, (END-BEG) should be added, not
32517 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32519 * config/avr/avr.md (setmemhi): Make sure match_dup
32520 operand number comes before match_scratch.
32522 2017-03-08 Richard Biener <rguenther@suse.de>
32524 PR tree-optimization/79920
32525 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
32526 with ncopies == 1 to ...
32527 (vect_transform_slp_perm_load): ... here. Properly compute
32528 all element loads by iterating VF times over the group. Do
32529 not handle ncopies (computed in a broken way) in
32530 vect_create_mask_and_perm.
32532 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32535 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
32536 is a uniform vector, use uniform_vector_p return value instead of
32537 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
32539 2017-03-07 Marek Polacek <polacek@redhat.com>
32541 PR middle-end/79809
32542 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
32543 (alloca_call_type): Likewise.
32545 2017-03-07 Martin Liska <mliska@suse.cz>
32547 * gcov.c (process_args): Put comment to correct location.
32549 2017-03-07 Martin Liska <mliska@suse.cz>
32551 PR middle-end/68270
32552 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
32553 Use array_at_struct_end_p instead of DECL_CHAIN (field).
32554 (chkp_narrow_bounds_for_field): Likewise.
32555 (chkp_parse_array_and_component_ref): Pass one more argument to
32558 2017-03-07 Richard Biener <rguenther@suse.de>
32560 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
32563 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
32565 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
32566 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
32568 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32571 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
32572 to end of description.
32573 (PARAM_MAX_STORES_TO_MERGE): Likewise.
32575 2017-03-07 Jakub Jelinek <jakub@redhat.com>
32577 PR rtl-optimization/79901
32578 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
32580 (*avx512f_<code><mode>3<mask_name>): ... this.
32581 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
32582 iterator instead of VI8_AVX2_AVX512BW.
32584 PR rtl-optimization/79901
32585 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
32586 min/max expander, expand it using expand_vec_cond_expr.
32589 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
32592 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32595 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
32596 to void * for PCH reasons.
32597 * dwarf2out.c (output_loc_operands, output_die): Cast
32598 v.val_vec.array to unsigned char *.
32600 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
32603 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
32606 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
32608 PR rtl-optimization/79571
32609 * lra-constraints.c (process_alt_operands): Calculate static
32610 reject and subtract it from overall when only addresses will be
32613 2017-03-06 Julia Koval <julia.koval@intel.com>
32616 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
32617 incoming stack boundary to 128 for 64-bit targets.
32619 2017-03-06 Richard Biener <rguenther@suse.de>
32621 PR tree-optimization/79894
32622 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
32623 to NULL after folding it.
32625 2017-03-06 Richard Biener <rguenther@suse.de>
32627 PR tree-optimization/79824
32628 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
32629 check disabling peeling for gaps.
32631 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
32633 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
32634 attributes): Document gettimeofday.
32636 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32638 * config/s390/s390.c (s390_option_override_internal): Set
32639 PARAM_MIN_VECT_LOOP_BOUND
32641 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32643 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
32644 * config/s390/s390.md: Likewise.
32646 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32649 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
32650 (<avx2_avx512>_perm<mode>): Rename to ...
32651 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
32653 (<avx512>_perm<mode>_mask): Rename to ...
32654 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
32656 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
32657 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
32658 instead of VI8F_256_512.
32659 (avx512f_perm<mode>): New define_expand.
32660 (avx512f_perm<mode>_mask): Likewise.
32661 (avx512f_perm<mode>_1<mask_name>): New define_insn.
32662 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
32664 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32666 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
32667 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
32669 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
32671 2017-03-06 Martin Liska <mliska@suse.cz>
32674 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
32675 when having a SSA NAME w/o VAR_DECL assigned to it.
32677 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32679 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
32680 msa_dpsub_<su>_d): Fix MODE for vec_select.
32682 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32684 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
32686 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
32688 2017-03-06 Richard Biener <rguenther@suse.de>
32690 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
32691 * plugin.c (register_plugin_info): Likewise.
32692 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
32694 2017-03-05 Jakub Jelinek <jakub@redhat.com>
32696 * config/i386/sse.md (sse_storehps, sse_storelps,
32697 avx_<castmode><avxsizesuffix>_<castmode>,
32698 avx512f_<castmode><avxsizesuffix>_<castmode>,
32699 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
32700 in condition that at least one operand is not a MEM.
32702 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32704 PR middle-end/79805
32705 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
32706 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
32708 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
32709 gimple_call_nothrow_p flag based on whether original builtin can throw.
32710 If it can, emit following stmts on the fallthrough edge.
32711 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
32712 don't create new bb if inserting just debug stmts on the edge, try to
32713 insert them on the fallthru bb or just reset debug stmts.
32715 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
32718 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
32719 restore recog_data (including the operand rtxes inside it) around
32720 the call to get_insn_template.
32722 2017-03-03 Martin Sebor <msebor@redhat.com>
32724 PR tree-optimization/79699
32725 * context.c (context::~context): Free MPFR caches to avoid
32726 a memory leak on program exit.
32728 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32730 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
32731 Use wide_int::ulow () instead of .elt (0).
32733 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32735 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
32736 (*pushxf): Limit oF constraint to 32bit targets and add oC
32737 constraint for 64bit targets.
32738 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
32739 (*pushdf): Change rmF constraint to rmC.
32741 2017-03-03 Martin Liska <mliska@suse.cz>
32743 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
32744 Remove unused variable.
32746 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32749 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
32750 is a memory operand, increase num_memory.
32751 (ix86_expand_args_builtin): Likewise.
32753 2017-03-03 Jan Hubicka <jh@suse.cz>
32756 * ipa-devirt.c (maybe_record_node): Properly handle
32757 __cxa_pure_virtual visibility.
32759 2017-03-03 Martin Liska <mliska@suse.cz>
32761 PR tree-optimization/79803
32762 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
32764 (pass_loop_prefetch::execute): Disabled optimization if an
32765 assumption about L1 cache size is not met.
32767 2017-03-03 Martin Liska <mliska@suse.cz>
32769 PR rtl-optimization/79574
32770 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
32771 (hash_scan_set): Likewise.
32772 (dump_hash_table): Likewise.
32773 (hoist_code): Likewise.
32775 2017-03-03 Richard Biener <rguenther@suse.de>
32777 * fixed-value.c (fixed_from_string): Restore use of elt (1)
32778 in place of uhigh ().
32779 (fixed_convert_from_real): Likewise.
32781 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32784 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
32786 2017-03-03 Richard Biener <rguenther@suse.de>
32788 PR middle-end/79818
32789 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
32790 TYPE_OVERFLOW_UNDEFINED check.
32792 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32794 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
32796 (vector_ae_<mode>_p): Likewise.
32797 (vector_nez_<mode>_p): Likewise.
32798 (vector_ne_v2di_p): Likewise.
32799 (vector_ae_v2di_p): Likewise.
32800 (vector_ne_<mode>_p): Likewise.
32801 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
32803 (vsx_tsqrt<mode>2_fe): Likewise.
32805 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
32808 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
32810 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32812 PR rtl-optimization/79780
32813 * cprop.c (one_cprop_pass): When second and further conditional trap
32814 in a single basic block is turned into an unconditional trap, turn it
32815 into a deleted note to avoid RTL verification failures.
32817 2017-03-02 Richard Biener <rguenther@suse.de>
32819 * fold-const.c (const_binop): Use ulow () instead of elt (0).
32821 2017-03-02 Richard Biener <rguenther@suse.de>
32823 PR tree-optimization/79345
32825 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
32826 param and abort the walk, returning -1 if it is hit.
32827 (walk_aliased_vdefs): Take a limit param and pass it on.
32828 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
32829 defaulting to 0 and return a signed int.
32830 * tree-ssa-uninit.c (struct check_defs_data): New struct.
32831 (check_defs): New helper.
32832 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
32833 about uninitialized memory.
32834 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
32835 bogus uninitialized warning.
32836 (fixed_convert_from_real): Likewise.
32838 2017-03-02 Bin Cheng <bin.cheng@arm.com>
32840 PR tree-optimization/66768
32841 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
32842 iv_use if base object can't be determined.
32844 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32846 PR tree-optimization/79345
32847 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
32848 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
32849 (get_pattern_stats): Initialize it.
32850 * genemit.c (gen_expand): Verify match_scratch numbers come after
32851 match_operand/match_dup numbers.
32852 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
32853 match_scratch numbers.
32854 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
32856 * config/s390/s390.md (trunctdsd2): Likewise.
32858 2017-03-02 Richard Biener <rguenther@suse.de>
32860 * wide-int.h (wide_int_storage::operator=): Implement in terms
32863 2017-03-02 Richard Biener <rguenther@suse.de>
32865 PR tree-optimization/79777
32866 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
32867 the to insert expression to sth existing.
32869 2017-03-01 Martin Sebor <msebor@redhat.com>
32871 PR middle-end/79692
32872 * gimple-ssa-sprintf.c
32873 (directive::known_width_and_precision): New function.
32874 (format_integer): Use it.
32875 (get_mpfr_format_length): Consider the full range of precision
32876 when computing %g output with the # flag. Set the likely byte
32877 count to 3 rather than 1 when precision is indeterminate.
32878 (format_floating): Correct the lower bound of precision.
32880 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32882 * doc/invoke.texi: Document default code model for 64-bit Linux.
32884 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32887 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
32888 udiv rather than div since input pattern is unsigned.
32890 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
32892 * config/i386/i386.c (print_reg): Warn for values of
32893 unsupported size in integer register.
32895 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
32898 * config/rs6000/predicates.md (current_file_function_operand): Do
32899 not allow self calls to be local if the function is replaceable.
32901 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
32904 * config/rs6000/altivec.h (vec_ctz and others): Change the
32905 preprocessor macro that controls conditional compilation from
32906 _ARCH_PWR9 to __POWER9_VECTOR__.
32907 (vec_all_ne): Change parameterization of __altivec_scalar_pred
32908 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
32909 control (instead of _ARCH_PWR9 control) so that template
32910 definition uses power9-specific function.
32911 (vec_any_eq): Likewise.
32912 (vec_all_ne): Change macro definition to use a power9-specific
32913 expansion under #ifdef __POWER9_VECTOR__ control (instead of
32914 _ARCH_PWR9 control).
32915 (vec_any_eq) Likewise.
32916 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
32917 expansion for CMPNEF to remove support for xvcmpnesp instruction.
32918 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
32919 support for xvcmpnedp instruction.
32920 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
32921 macro expansion so that Power9 implementation of vec_all_ne does
32922 not use the AltiVec predicate framework.
32923 (VCMPNEH_P): Likewise.
32924 (VCMPNEW_P): Likewise.
32925 (VCMPNED_P): Likewise.
32926 (VCMPNEFP_P): Likewise.
32927 (VCMPNEDP_P): Likewise.
32928 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
32929 implementation of vec_any_eq to not use AltiVec predicate
32931 (VCMPAEH_P): Likewise.
32932 (VCMPAEW_P): Likewise.
32933 (VCMPAED_P): Likewise.
32934 (VCMPAEFP_P): Likewise.
32935 (VCMPAEDP_P): Likewise.
32936 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
32937 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
32938 not use the AltiVec predicate framework.
32939 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
32940 of vec_any_eq to not use AltiVec predicate framework.
32941 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32942 support for predefined __POWER9_VECTOR__ macro to indicate that
32943 Power9 instruction selection is enabled.
32944 (altivec_overloaded_builtins): Remove extraneous
32945 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
32946 function argument types RS6000_BTI_bool_V16QI and
32947 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
32948 entry for overloaded function argument types RS6000_BTI_bool_V4SI
32949 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
32950 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
32951 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
32952 Power9 for implementations of vec_cmpne. Change the signature for
32953 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
32954 (representing vec_all_ne) to remove the previously described first
32955 argument of type RS6000_BTI_INTSI, as this was an artifact of
32956 reliance on the AltiVec predicate framework, which is no longer
32957 used in the implementation of these functions. Add
32958 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
32959 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
32960 since, unlike the AltiVec predicate framework implementation, we
32961 do not share function descriptors between vec_alle and vec_anyeq.
32962 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
32963 set of modes that receive special treatment even when
32964 TARGET_P9_VECTOR is true. The special treatment emits code that
32965 does not depend on Power9 instructions.
32966 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
32967 define_expand to not rely on AltiVec predicate framework.
32968 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
32970 (vector_ne_v2di_p): Change this define_expand to not rely on
32971 AltiVec predicate framework.
32972 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
32974 (vector_ne_<mode>_p): Change this define_expand to not rely on
32975 AltiVec predicate framework.
32976 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
32978 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
32979 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
32980 define_insn pattern.
32981 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
32982 define_insn pattern because the xvcmpne<VSs>. instruction is not
32984 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
32985 instruction is not supported.
32987 2017-03-01 Jakub Jelinek <jakub@redhat.com>
32989 * config/nvptx/nvptx.c: Include intl.h.
32991 2017-03-01 Martin Jambor <mjambor@suse.cz>
32994 * ipa-prop.h (ipa_bits): Removed field known.
32995 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
32996 to pointers. Adjusted their comments to warn about their sharing.
32997 (ipcp_transformation_summary): Change bits to a vector of pointers.
32998 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
32999 (ipa_get_ipa_bits_for_value): Declare.
33000 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33001 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33002 (ipa_bits_hash_table): Likewise.
33003 (ipa_vr_ggc_hash_traits): Likewise.
33004 (ipa_vr_hash_table): Likewise.
33005 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33006 being pointers and vr_known being removed.
33007 (ipa_set_jf_unknown): Likewise.
33008 (ipa_get_ipa_bits_for_value): New function.
33009 (ipa_set_jfunc_bits): Likewise.
33010 (ipa_get_value_range): New overloaded functions.
33011 (ipa_set_jfunc_vr): Likewise.
33012 (ipa_compute_jump_functions_for_edge): Use the above functions to
33013 construct bits and vr parts of jump functions.
33014 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33015 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33017 (ipcp_grow_transformations_if_necessary): Also allocate
33018 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33020 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33021 them. Fix too long lines.
33022 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33023 vr_known being removed.
33024 (ipa_read_jump_function): Use new setter functions to construct bits
33025 and vr parts of jump functions or set them to NULL.
33026 (write_ipcp_transformation_info): Adjust for bits being pointers.
33027 (read_ipcp_transformation_info): Likewise.
33028 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33030 Include gt-ipa-prop.h.
33031 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33033 (ipcp_store_bits_results): Likewise.
33034 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33035 Do not write to existing jump functions but use a temporary instead.
33037 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33040 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33041 attempt to use its first operand as BIT_FIELD_REF base.
33043 2017-03-01 Richard Biener <rguenther@suse.de>
33045 PR middle-end/79721
33046 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33047 interpolating formula in wrapping arithmetic.
33048 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33050 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33052 PR tree-optimization/79734
33053 * tree-vect-generic.c (expand_vector_condition): Optimize
33054 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33055 Handle VEC_COND_EXPR where comparison has different inner width from
33056 type's inner width.
33058 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
33060 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
33061 markup, and similar issues. Remove @opindex entries for things
33062 that aren't options. Add missing -mmpy-option entries.
33064 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33066 PR tree-optimization/79737
33067 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
33068 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
33069 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
33070 instead of byte_size. Formatting fix.
33071 (shift_bytes_in_array_right): Formatting fix.
33073 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
33076 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
33077 condition on optimize for the leaf function test.
33079 2017-02-28 Martin Liska <mliska@suse.cz>
33082 * read-rtl-function.c (function_reader::handle_unknown_directive):
33083 Bail out when one uses -flto.
33085 2017-02-28 Martin Liska <mliska@suse.cz>
33087 * common.opt: Replace space with tabular for options of <number>
33089 * config/i386/i386.opt: Show <number> value for
33090 -mlarge-data-threshold.
33091 * opts.c (print_filtered_help): Do not display number in hexadecimal
33094 2017-02-28 Martin Liska <mliska@suse.cz>
33096 * common.opt: Fix --help=option -Q for options which are of
33099 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
33101 * config/i386/i386.c (print_reg): Error out for values
33102 of 8-bit size in invalid integer register.
33104 2017-02-28 Martin Sebor <msebor@redhat.com>
33106 PR tree-optimization/79691
33107 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
33109 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33112 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
33113 gcc_unreachable with output_operand_lossage.
33115 2017-02-28 Richard Biener <rguenther@suse.de>
33117 PR tree-optimization/79740
33118 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
33120 (visit_nary_op): Insert the nary into the hashtable if we
33121 pattern-matched sth.
33122 * tree-ssa-pre.c (eliminate_insert): Robustify.
33124 2017-02-28 Richard Biener <rguenther@suse.de>
33126 PR middle-end/79731
33127 * fold-const.c (decode_field_reference): Reject out-of-bound
33130 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33132 * config/i386/i386.c: Include intl.h.
33133 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
33134 instead of just cond ? "..." : "...".
33135 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
33136 * coverage.c (read_counts_file): Likewise.
33137 * omp-offload.c: Include intl.h.
33138 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
33139 of just cond ? "..." : "...".
33140 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
33141 of just cond ? "..." : "...".
33143 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
33146 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
33148 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
33149 'tune for' CPU name.
33150 * config/arm/arm-cpu-data.h: Regenerated.
33152 2017-02-28 Richard Biener <rguenther@suse.de>
33154 PR tree-optimization/79732
33155 * tree-inline.c (expand_call_inline): Do not shadow var.
33157 2017-02-28 Richard Biener <rguenther@suse.de>
33159 PR tree-optimization/79723
33160 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
33161 address-space properly.
33163 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
33165 * doc/optinfo.texi (Optimization groups): Fix option used for
33167 * doc/invoke.texi (-fopt-info): Document "omp".
33168 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
33169 (OPTGROUP_ALL): Add OPTGROUP_OMP.
33170 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
33171 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
33172 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
33174 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
33176 * dumpfile.c (optgroup_options): Instead of "openmp", associate
33177 OPTGROUP_OMP with "omp".
33179 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
33182 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
33183 for arithmetic shift of unsigned V2DI.
33185 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
33187 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
33188 arc/linux.h headers.
33189 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
33190 (LINK_SPEC): Likewise.
33191 (ARC_TLS_EXTRA_START_SPEC): Likewise.
33192 (EXTRA_SPECS): Likewise.
33193 (STARTFILE_SPEC): Likewise.
33194 (ENDFILE_SPEC): Likewise.
33195 (LIB_SPEC): Likewise.
33196 (TARGET_SDATA_DEFAULT): Likewise.
33197 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
33198 (MULTILIB_DEFAULTS): Likewise.
33199 (DWARF2_UNWIND_INFO): Likewise.
33200 * config/arc/big.h: New file.
33201 * config/arc/elf.h: Likewise.
33202 * config/arc/linux.h: Likewise.
33203 * config/arc/t-uClibc: Remove.
33205 2017-02-27 Bin Cheng <bin.cheng@arm.com>
33207 PR tree-optimization/77536
33208 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
33209 (tree_transform_and_unroll_loop): Use above function to compute the
33210 estimated niter of unrolled loop and use it when scaling profile.
33211 Also use count info rather than frequency if it's non-zero.
33212 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
33213 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
33214 (vect_transform_loop): Call above function.
33216 2017-02-27 Richard Biener <rguenther@suse.de>
33218 PR tree-optimization/45397
33219 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
33220 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
33221 (visit_nary_op): Add pattern matching for CSEing sign-changed
33222 or truncated operations with wider ones.
33224 2017-02-27 Richard Biener <rguenther@suse.de>
33226 PR tree-optimization/79690
33227 * tree-vect-stmts.c (vectorizable_store): Use vector type
33228 built from the DR with address-space.
33230 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
33232 * doc/invoke.texi (Optimize Options): Refine the description
33233 of asan-use-after-return.
33235 2017-02-25 Alan Modra <amodra@gmail.com>
33237 PR rtl-optimization/79584
33238 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
33239 base, not ad->base_term, the reg within base. Remove assertion
33240 that ad->base == ad->base_term. Replace gen_int_mode using
33241 bogus mode with const0_rtx.
33243 2017-02-25 Jakub Jelinek <jakub@redhat.com>
33245 PR middle-end/79396
33246 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
33247 FMA_EXPR like tcc_binary or tcc_unary.
33249 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
33252 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
33254 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
33255 (output_loc_operands): Handle DW_OP_call_ref and
33256 DW_OP_GNU_variable_value.
33257 (struct variable_value_struct): New type.
33258 (struct variable_value_hasher): Likewise.
33259 (variable_value_hash): New variable.
33260 (string_types): Remove.
33261 (copy_loc_descr): New function.
33262 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
33263 (prepend_loc_descr_to_each): New function.
33264 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
33265 instead of add_loc_descr_to_each if the first argument is single
33266 location list and the second has multiple.
33267 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
33268 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
33269 when looking for variable value which doesn't have other location info.
33270 (loc_list_from_tree): Formatting fix.
33271 (gen_array_type_die): Simplify DW_AT_string_length handling.
33272 (adjust_string_types): Remove.
33273 (gen_subprogram_die): Don't call adjust_string_types nor test/set
33274 string_types. Call resolve_variable_values.
33275 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
33276 (resolve_addr_in_expr): Likewise. Add A argument.
33277 (copy_deref_exprloc): Remove deref argument. Adjust for the
33278 original expression being DW_OP_GNU_variable_value with optionally
33279 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
33280 optionally after it.
33281 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
33282 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
33283 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
33284 (variable_value_hasher::hash, variable_value_hasher::equal): New
33286 (resolve_variable_value_in_expr, resolve_variable_value,
33287 resolve_variable_values, note_variable_value_in_expr,
33288 note_variable_value): New functions.
33289 (dwarf2out_early_finish): Call note_variable_value on all toplevel
33292 2017-02-24 Jakub Jelinek <jakub@redhat.com>
33295 * opts.h (handle_generated_option): Add GENERATED_P argument.
33296 * opts-common.c (handle_option): Adjust function comment.
33297 (handle_generated_option): Add GENERATED_P argument, pass it to
33299 (control_warning_option): Pass false to handle_generated_option
33301 * opts.c (maybe_default_option): Pass true to handle_generated_option
33303 * optc-gen.awk: Likewise.
33305 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33307 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
33308 a REG, look at the REG it is a SUBREG of.
33309 (splitter for cmpeqsi_t): Ditto.
33311 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33313 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
33314 the special USEs with the pattern of the insn, not the insn itself.
33316 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
33319 * doc/invoke.texi: Document -mload-store-pairs.
33321 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33322 Sandra Loosemore <sandra@codesourcery.com>
33324 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
33325 argument isn't a CONST_INT.
33326 (nios2_alternate_compare_const): Assert op is a CONST_INT.
33327 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
33328 (nios2_validate_compare): Bypass alternate compare logic if *op2
33329 is not a CONST_INT.
33330 (ldstwm_operation_p): Return false if first_base is not a REG or
33331 if first_offset is not a CONST_INT.
33333 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33335 * config/cris/cris.md: Use correct operand in a define_peephole2.
33337 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33339 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
33341 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33343 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
33344 this_insn if it is an INSN or JUMP_INSN.
33345 (force_offsettable): Look at base, not at addr.
33346 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
33347 on things that aren't necessarily CONST_INTs.
33349 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
33351 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
33352 -mfpmath=sse is the default also for x86-32 targets with SSE2
33353 instruction set when @option{-ffast-math} is enabled
33355 2017-02-24 Jeff Law <law@redhat.com>
33357 PR rtl-optimizatoin/79286
33358 * ira.c (update_equiv_regs): Drop may_trap_p exception to
33361 2017-02-24 Richard Biener <rguenther@suse.de>
33363 PR tree-optimization/79389
33364 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
33367 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
33369 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
33370 function comment to reflect reality.
33371 (loop_exits_before_overflow): Fix typo in function description.
33373 2017-02-24 Richard Biener <rguenther@suse.de>
33375 PR tree-optimization/79389
33376 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
33377 properly that a threading opportunity exists. Detect conditional
33378 copy/constant propagation opportunities.
33380 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
33382 * config/visium/visium.md (type): Add trap.
33383 (b): New mode attribute.
33384 (*btst): Rename into...
33385 (*btst<mode>): ...this and adjust.
33386 (*cbranchsi4_btst_insn): Rename into...
33387 (*cbranch<mode>4_btst_insn): ...this and adjust.
33388 (trap): New define_insn.
33390 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33392 PR tree-optimization/79389
33393 * ifcvt.c (struct noce_if_info): Add rev_cond field.
33394 (noce_reversed_cond_code): New function.
33395 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
33396 reversed_comparison_code. Formatting fix.
33397 (noce_try_store_flag): Test rev_cond != NULL in addition to
33398 reversed_comparison_code.
33399 (noce_try_store_flag_constants): Likewise.
33400 (noce_try_store_flag_mask): Likewise.
33401 (noce_try_addcc): Use rev_cond if non-NULL instead of
33402 reversed_comparison_code.
33403 (noce_try_cmove_arith): Likewise. Formatting fixes.
33404 (noce_try_minmax, noce_try_abs): Clear rev_cond.
33405 (noce_find_if_block): Initialize rev_cond.
33406 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
33407 instead of false as last argument never attempt to reverse it
33410 2017-02-23 Bin Cheng <bin.cheng@arm.com>
33412 PR tree-optimization/79663
33413 * tree-predcom.c (combine_chains): Process refs in reverse order
33414 only for ZERO length chains, and add explaining comment.
33416 2017-02-23 Jeff Law <law@redhat.com>
33418 PR tree-optimization/79578
33419 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
33420 in call to operand_equal_p.
33422 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
33425 * config/i386/cpuid.h: Fix another undefined behavior.
33427 2017-02-23 Richard Biener <rguenther@suse.de>
33429 PR tree-optimization/79683
33430 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
33431 vector types for data-refs.
33433 2017-02-23 Martin Liska <mliska@suse.cz>
33435 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
33437 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33439 PR middle-end/79665
33440 * internal-fn.c (get_range_pos_neg): Moved to ...
33441 * tree.c (get_range_pos_neg): ... here. No longer static.
33442 * tree.h (get_range_pos_neg): New prototype.
33443 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
33444 are known to be in between 0 and signed maximum inclusive, try to
33445 expand both unsigned and signed divmod and use the cheaper one from
33448 2017-02-22 Jeff Law <law@redhat.com>
33450 PR tree-optimization/79578
33451 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
33452 to compare base operands.
33454 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
33457 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
33458 gpc_reg_operand instead of fpr_reg_operand.
33460 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
33462 * config/mips/mips.c (mips_return_in_memory): Force FP
33463 vector types to be returned in memory for o32 ABI.
33465 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33467 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
33468 instead of DW_TAG_member for static data member declarations and don't
33469 set no_linkage_name for static inline data members.
33470 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
33473 2017-02-22 Martin Liska <mliska@suse.cz>
33475 * doc/invoke.texi: Replace inequality signs with square brackets
33478 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
33481 * lra-constraints.c (simplify_operand_subreg): Handle
33482 WORD_REGISTER_OPERATIONS targets.
33484 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33487 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
33488 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
33489 elimination by swapping fld*.
33491 2017-02-22 Richard Biener <rguenther@suse.de>
33493 PR tree-optimization/79673
33494 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
33495 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
33496 irrelevant address-space qualifiers and avoiding a
33497 ADDR_SPACE_CONVERT_EXPR from fold_convert.
33499 2017-02-22 Richard Biener <rguenther@suse.de>
33501 PR tree-optimization/79666
33502 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
33503 to not symbolically negate if that may introduce undefined
33506 2017-02-22 Martin Liska <mliska@suse.cz>
33509 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
33510 * data-streamer-out.c (streamer_write_gcov_count_stream):
33512 * value-prof.c (stream_out_histogram_value): Make assert more
33513 precise based on type of counter.
33515 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
33518 * config/i386/i386.md (standard_x87sse_constant_load splitter):
33519 Use nonimmediate_operand instead of memory_operand for operand 1.
33520 (float-extend standard_x87sse_constant_load splitter): Ditto.
33522 2017-02-21 Jeff Law <law@redhat.com>
33524 PR tree-optimization/79621
33525 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
33526 blocks with edges to themselves.
33528 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33531 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
33532 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
33533 Use gimple_call_builtin_p.
33536 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
33537 on temporarily removed DEBUG_INSNs.
33539 PR tree-optimization/79649
33540 * tree-loop-distribution.c (classify_partition): Give up on
33541 non-generic address space loads/stores.
33543 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
33545 * doc/loop.texi (Loop manipulation): Remove nonexistent
33546 tree_ssa_loop_version from the documentation.
33547 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
33549 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33552 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
33553 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
33554 * config/rs6000/rs6000.c: Include except.h.
33555 (rs6000_expand_split_stack_prologue): Call
33556 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
33558 2017-02-21 Martin Jambor <mjambor@suse.cz>
33561 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
33562 have been analyzed.
33564 2017-02-21 Martin Jambor <mjambor@suse.cz>
33566 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
33567 for backward compatibility only.
33568 * doc/invoke.texi (Option Summary): Remove all references to
33569 -fipa-cp-alignment.
33571 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
33575 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33577 * lra-constraints.c (curr_insn_transform): Handle
33578 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33580 2017-02-21 Martin Liska <mliska@suse.cz>
33582 * config/i386/i386.opt: Replace -masm-dialect with -masm.
33584 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
33586 PR translation/79638
33587 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
33589 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
33592 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
33593 (arm_function_ok_for_sibcall): Return false for an indirect call by
33594 descriptor if all the argument registers are used.
33595 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
33596 alignment of the function.
33598 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33600 PR tree-optimization/61441
33601 * simplify-rtx.c (simplify_const_unary_operation): For
33602 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
33603 the sNaN unmodified.
33605 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33607 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33608 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33609 instead of SYSTEM_HEADER_DIR.
33611 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
33612 Martin Liška <mliska@suse.cz>
33614 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
33615 Fix typos and grammar, use active voice, and clarify.
33617 2017-02-20 Marek Polacek <polacek@redhat.com>
33619 PR middle-end/79537
33620 * gimplify.c (gimplify_expr): Handle unused *&&L;.
33623 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
33625 2017-02-20 Jakub Jelinek <jakub@redhat.com>
33628 * config/i386/i386.c (ix86_expand_builtin): Handle
33629 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
33630 ix86_builtins_isa[fcode].isa as a requirement of those
33631 flags and any other flag in the bitmask.
33632 (ix86_init_mmx_sse_builtins): Use 0 instead of
33633 ~OPTION_MASK_ISA_64BIT as mask.
33634 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
33635 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
33636 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
33637 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
33639 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33642 * lra-constraints.c (split_reg): Check requested split mode
33643 is supported by the register.
33645 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33647 * lra-constraints.c (simplify_operand_subreg): Remove early
33650 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33653 * lra-constraints.c (curr_insn_transform): Tighten condition
33654 for converting SUBREG reloads from OP_OUT to OP_INOUT.
33656 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33659 * lra-constraints.c (curr_insn_transform): Handle
33660 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33662 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
33665 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
33667 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
33669 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
33672 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
33675 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33677 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
33680 2017-02-18 Jakub Jelinek <jakub@redhat.com>
33682 * final.c (last_columnnum, override_columnnum): New variables.
33683 (final_start_function): Set last_columnnum, pass it to begin_prologue
33684 hook and pass 0 to dwarf2out_begin_prologue.
33685 (final_scan_insn): Update override_columnnum. Pass last_columnnum
33686 to source_line debug hook.
33687 (notice_source_line): Compute last_columnnum and for debug_column_info
33688 return true on column changes.
33689 * debug.h (struct gcc_debug_hooks): Add column argument to
33690 source_line and begin_prologue hooks.
33691 (debug_nothing_int_charstar_int_bool): Remove prototype.
33692 (debug_nothing_int_int_charstar,
33693 debug_nothing_int_int_charstar_int_bool): New prototypes.
33694 (dwarf2out_begin_prologue): Add column argument.
33695 * debug.c (do_nothing_debug_hooks): Adjust source_line and
33696 begin_prologue hooks.
33697 (debug_nothing_int_charstar_int_bool): Remove.
33698 (debug_nothing_int_int_charstar,
33699 debug_nothing_int_int_charstar_int_bool): New functions.
33700 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
33701 through to dwarf2out_source_line.
33702 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
33703 (dwarf2out_source_line): Add column argument, emit it if requested.
33704 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
33706 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33707 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33708 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
33709 through to dwarf2out_begin_prologue.
33710 (vmsdbgout_source_line): Add column argument, pass it through to
33711 dwarf2out_source_line.
33712 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
33713 dbxout_source_line caller.
33714 (dbxout_source_line): Add column argument.
33716 * common.opt (gno-column-info, gcolumn-info): New options.
33717 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
33718 (check_die): Also test for multiple DW_AT_decl_column attributes.
33719 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
33720 DW_AT_decl_column if requested.
33721 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
33723 (gen_variable_die): Likewise.
33724 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
33725 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
33728 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
33729 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
33730 (ix86_handle_option): Handle OPT_m3dnowa.
33731 * doc/invoke.texi (-m3dnowa): Document.
33732 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
33733 -m3dnowa instead of -m3dnow -march=athlon.
33736 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
33737 instead of gcc_assert for K, r and R code checks. Formatting fixes.
33739 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33742 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
33743 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
33744 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
33745 generator for vsx_xxpermdi_<mode>_be.
33746 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
33747 force big-endian semantics.
33748 (vsx_xxpermdi_<mode>_be): New define_expand with same
33749 implementation as previous version of vsx_xxpermdi_<mode>.
33751 2017-02-17 Jakub Jelinek <jakub@redhat.com>
33753 PR tree-optimization/79327
33754 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
33755 variable, its initialization and use.
33757 2017-02-17 Julia Koval <julia.koval@intel.com>
33759 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
33760 (OPTION_MASK_ISA_PKU_UNSET): New.
33761 (ix86_handle_option): Handle -mrdpid.
33762 * config/i386/cpuid.h (bit_RDPID): New.
33763 * config/i386/driver-i386.c (host_detect_local_cpu):
33764 Detect RDPID feature.
33765 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
33766 * config/i386/i386-c.c (ix86_target_macros_internal):
33768 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
33769 (ix86_valid_target_attribute_inner_p): Add "rdpid".
33770 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
33771 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
33772 * config/i386/i386.md (define_insn "rdpid"): New.
33773 * config/i386/i386.opt Add -mrdpid.
33774 * config/i386/immintrin.h (_rdpid_u32): New.
33776 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
33778 PR rtl-optimization/79541
33779 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
33780 instead of transforming it into USE.
33782 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
33784 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
33785 If HONOR_SNANS (SFmode) force the input to a register.
33786 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
33787 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
33788 an frsp or similar insn.
33790 2017-02-17 Martin Liska <mliska@suse.cz>
33792 PR rtl-optimization/79577
33793 * params.def (selsched-max-sched-times): Increase minimum to 1.
33795 2017-02-17 Martin Liska <mliska@suse.cz>
33797 PR rtl-optimization/79574
33798 * gcse.c (want_to_gcse_p): Prevent integer overflow.
33800 2017-02-17 Martin Liska <mliska@suse.cz>
33802 PR tree-optimization/79529
33803 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
33804 ssa_defined_default_def_p to handle cases which are implicitly
33806 * tree-ssa.c (ssa_defined_default_def_p): New function.
33807 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
33808 which are implicitly defined.
33809 * tree-ssa.h (ssa_defined_default_def_p): Declare.
33811 2017-02-17 Richard Biener <rguenther@suse.de>
33813 PR middle-end/79576
33814 * params.def (max-ssa-name-query-depth): Limit to 10.
33816 2017-02-17 Richard Biener <rguenther@suse.de>
33818 PR tree-optimization/79552
33819 * tree-ssa-structalias.c (visit_loadstore): Properly verify
33822 2017-02-17 Richard Biener <rguenther@suse.de>
33825 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
33827 2017-02-17 Marek Polacek <polacek@redhat.com>
33829 PR middle-end/79536
33830 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
33831 (fold_negate_expr): New wrapper.
33833 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
33835 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
33836 Correct terminology and de-emphasize pre-standard behavior.
33838 2017-02-16 Alan Modra <amodra@gmail.com>
33840 PR rtl-optimization/79286
33841 * ira.c (def_dominates_uses): New function.
33842 (update_equiv_regs): Don't create an equivalence for insns that
33843 may trap where the register def does not dominate the use.
33845 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
33847 PR rtl-optimization/78127
33848 * lra.c (lra): Call lra_eliminate before finish the loop after
33851 2017-02-16 Richard Biener <rguenther@suse.de>
33853 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
33855 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
33856 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
33857 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
33858 (isl_val_int_from_wi): New function.
33859 (extract_affine_gmp): Rename to ...
33860 (extract_affine_wi): ... this, take a widest_int.
33861 (extract_affine_int): Just wrap extract_affine_wi.
33862 (add_param_constraints): Use isl_val_int_from_wi.
33863 (add_loop_constraints): Likewise, and extract_affine_wi.
33865 2017-02-15 Jeff Law <law@redhat.com>
33867 PR middle-end/79521
33868 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
33869 ira_init_register_move_cost_if_necessary.
33871 2017-02-15 Martin Sebor <msebor@redhat.com>
33873 PR middle-end/32003
33874 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
33875 removed in a prior commit.
33877 2017-02-15 Bin Cheng <bin.cheng@arm.com>
33879 PR tree-optimization/79347
33880 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
33881 counters during peeling.
33883 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
33885 * Makefile.in (site.exp): Remove "set ISLVER".
33887 2017-02-15 Jakub Jelinek <jakub@redhat.com>
33890 * real.c (real_from_integer): Call real_convert even for decimal.
33892 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
33895 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
33897 2017-02-14 Andrew Pinski <apinski@cavium.com>
33899 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
33900 cores and change the partno/implementer to be correct.
33901 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
33902 the 'B" as the implementer.
33903 * config/aarch64/aarch64-tune.md: Regenerate.
33905 2017-02-14 Carl Love <cel@us.ibm.com>
33907 * config/rs6000/rs6000.c: Add case statement entry to make the
33908 xvcvuxdsp built-in argument unsigned.
33909 * config/rs6000/vsx.md: Fix the source and return operand types so they
33910 match the instruction definitions from the ISA document. Fix typo
33911 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
33914 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
33917 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
33918 member early_clobber_alts.
33919 * lra-lives.c (reg_early_clobber_p): New.
33920 (process_bb_lives): Use it.
33921 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
33922 (debug_operand_data): Initialize early_clobber_alts.
33923 (setup_operand_alternative): Set up early_clobber_alts.
33924 (collect_non_operand_hard_regs): Ditto. Pass early clobber
33925 alternatives to new_insn_reg.
33926 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
33928 (lra_update_insn_regno_info): Pass the new arg.
33930 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33932 PR middle-end/79505
33933 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
33934 (new_oacc_loop_raw): Don't clear already cleared fields.
33937 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
33938 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
33939 _mm512_prefetch_i64gather_ps): New inline functions and macros.
33941 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
33944 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
33946 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
33949 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
33950 the extra instruction to the right place to store 128-bit constant
33953 2017-02-14 Martin Sebor <msebor@redhat.com>
33955 PR middle-end/79448
33956 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
33957 warning for strings of unknown length.
33959 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
33961 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
33963 2017-02-14 Jeff Law <law@redhat.com>
33966 * ira-costs.c (scan_one_insn): Initialize register move costs
33967 for pseudos seen in USE/CLOBBER insns.
33969 PR tree-optimization/79095
33970 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
33971 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
33972 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
33973 if the operands are known to be not equal, then the resulting range
33975 (intersect_ranges): If the new range is ~[0,0] and the old range is
33976 wide, then prefer ~[0,0].
33977 * tree-vrp.c (overflow_comparison_p_1): New function.
33978 (overflow_comparison_p): New function.
33979 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
33980 if NAME is used in an overflow test.
33981 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
33982 overflow check that can be expressed as an equality test, then adjust
33983 ops to be that equality test.
33985 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33987 * config/s390/s390-builtin-types.def: Remove flags argument.
33988 * config/s390/s390.c (s390_init_builtins): Likewise.
33990 2017-02-14 Martin Liska <mliska@suse.cz>
33992 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
33993 vector. Fix trailing white spaces.
33995 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
33997 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34000 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34002 PR rtl-optimization/68664
34003 * config/arm/arm.c (arm_sched_can_speculate_insn):
34004 New function. Declare prototype.
34005 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34007 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34009 PR rtl-optimization/68664
34010 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34012 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34014 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34016 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34017 max skip bytes for function, loop and jump.
34019 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34021 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34022 ABS_EXPR for gimple dump.
34024 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34027 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34029 PR tree-optimization/79408
34030 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34031 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34032 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34033 also if rhs1 is INTEGER_CST.
34035 2017-02-14 Richard Biener <rguenther@suse.de>
34037 PR middle-end/79432
34038 * tree-into-ssa.c (insert_phi_nodes): When the function can
34039 have abnormal edges rewrite SSA names with broken use-def
34040 dominance out of SSA and register them for PHI insertion.
34042 2017-02-13 Martin Sebor <msebor@redhat.com>
34044 PR middle-end/79496
34045 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34046 clearing info.nowrite flag when snprintf size argument is a range.
34048 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34050 * cprop.c (cprop_jump): Add missing space in string literal.
34051 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34052 (get_constraint_for_component_ref): Likewise.
34053 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34054 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34055 * lra-constraints.c (process_alt_operands): Likewise.
34056 * ipa-inline.c (inline_small_functions): Likewise.
34057 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34058 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
34059 * trans-mem.c (diagnose_tm_1_op): Likewise.
34060 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
34061 (grid_parallel_clauses_gridifiable): Likewise.
34063 * config/nvptx/mkoffload.c (process): Add space in between
34066 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
34067 "MOD4_SSE_REGS" and "ALL_REGS".
34069 * spellcheck.c (test_data): Add , in between "foo" and "food".
34071 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34074 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
34075 boundary crossing check and subsequent code generation agree.
34077 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34079 * config/aarch64/aarch64.c (has_memory_op): Delete.
34080 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
34083 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34085 PR rtl-optimization/79388
34086 PR rtl-optimization/79450
34087 * combine.c (distribute_notes): When removing TEM_INSN for which
34088 corresponding dest has last value recorded, invalidate that last
34091 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34093 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
34094 of explicit '@'. Add missing assembly comment marker on branch costs
34097 2017-02-13 Nathan Sidwell <nathan@acm.org>
34099 * gengtype-lex.l (<in_struct>): Add '/'.
34101 2017-02-13 Martin Liska <mliska@suse.cz>
34104 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
34106 2017-02-13 Richard Biener <rguenther@suse.de>
34108 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
34110 * configure: Re-generate.
34111 * config.in: Likewise.
34112 * graphite-dependences.c: Simplify as if
34113 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
34114 * graphite-isl-ast-to-gimple.c: Likewise.
34115 * graphite-optimize-isl.c: Likewise.
34116 * graphite-poly.c: Likewise.
34117 * graphite-sese-to-poly.c: Likewise.
34118 * graphite.h: Likewise.
34119 * toplev.c: Include isl/version.h and use isl_version () for
34120 printing the ISL version.
34121 * doc/install.texi: Update ISL requirement.
34123 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34125 * doc/standards.texi (Standards): Update reference to
34128 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34130 * doc/extend.texi (Named Address Spaces): sourceware.org now
34132 * doc/install.texi (Binaries): Ditto.
34135 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34137 * doc/cpp.texi: Replace "stringify"/"stringification" with C
34138 standard terminology "stringize"/"stringizing" throughout.
34139 * doc/cppinternals.texi: Likewise.
34141 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34143 * doc/extend.texi: Fix some spelling mistakes and typos.
34144 * doc/invoke.texi: Likewise.
34146 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34149 * params.def (inline-min-speedup) Change from 10 to 8.
34151 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34153 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
34156 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34159 * ipa-inline-analysis.c (get_minimal_bb): New function.
34160 (record_modified): Use it.
34161 (remap_edge_change_prob): Handle also ancestor functions.
34163 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
34165 * doc/contrib.texi (Contributors): Remove broken link into
34166 the Mauve CVS repository.
34168 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34170 PR middle-end/79454
34171 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
34172 result computation whenever lhs doesn't have vector mode, not
34173 just when it has BLKmode.
34175 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34177 * doc/makefile.texi (profiledbootstrap): Refer to the
34178 installation instructions only in textual form.
34180 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34183 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
34185 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34187 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
34188 (Specific): Update mingw-w64 reference.
34190 (Specific): Remove broken link to Renesas RX processor.
34192 2017-02-10 Richard Biener <rguenther@suse.de>
34194 * toplev.c (process_options): Do not mention obsolete graphite
34195 options when printing sorry message about missing graphite support.
34196 Mention -floop-nest-optimize.
34198 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
34200 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
34201 (vtst_p16): Likewise.
34202 (vtstq_p8): Likewise.
34203 (vtstq_p16): Likewise.
34205 (vtstq_p64): Likewise.
34206 * config/arm/arm_neon.h (vgetq_lane_p64): New.
34207 (vset_lane_p64): New.
34208 (vsetq_lane_p64): New.
34210 2017-02-10 Jakub Jelinek <jakub@redhat.com>
34212 PR tree-optimization/79411
34213 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
34214 stmt operands are SSA_NAMEs used in abnormal phis.
34215 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
34218 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34221 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
34224 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34226 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
34228 2017-02-09 Jakub Jelinek <jakub@redhat.com>
34230 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
34234 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
34235 not arbitrary TREE_CONSTANT.
34238 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
34239 "omp declare target link" attribute unless is_global_var.
34240 * omp-offload.c (find_link_var_op): Likewise.
34242 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
34243 Chung-Lin Tang <cltang@codesourcery.com>
34245 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
34247 (gimplify_adjust_omp_clauses): Don't delete TILE.
34248 (gimplify_omp_for): Deal with TILE.
34249 * internal-fn.c (expand_GOACC_TILE): New function.
34250 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
34252 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
34253 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
34255 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
34256 avoid DIV for outermost collapse var.
34257 (expand_oacc_for): Insert tile element loop as needed. Adjust.
34258 Remove out of date comments, fix whitespace.
34259 * omp-general.c (omp_extract_for_data): Deal with tiling.
34260 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
34261 adjust OLF_DIM_BASE value.
34262 (struct omp_for_data): Add tiling field.
34263 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
34264 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
34265 for auto loops. Remove default auto determining, moved to
34266 oacc_loop_fixed_partitions.
34267 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
34268 stmts, add e_mask field.
34269 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
34270 (oacc_thread_numbers): Use oacc_dim_call.
34271 (oacc_xform_tile): New.
34272 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
34273 (finish_oacc_loop): Adjust for ifns vector.
34274 (oacc_loop_discover_walk): Append loop abstraction sites to list,
34275 add case for GOACC_TILE fns.
34276 (oacc_loop_xform_loop): Delete.
34277 (oacc_loop_process): Iterate over call list directly, and add
34278 handling for GOACC_TILE fns.
34279 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
34281 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
34282 vector partitioning to outer loops. Assign 2 partitions to loops
34283 when available. Add TILE handling.
34284 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
34285 (execite_oacc_device_lower): Process GOACC_TILE fns,
34286 ignore unknown specs.
34287 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
34288 * tree.c (omp_clause_num_ops): Adjust TILE ops.
34289 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
34291 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
34293 * configure.ac (ACX_BUGURL): Update.
34294 * configure: Regenerate.
34296 2017-02-09 Richard Biener <rguenther@suse.de>
34298 PR tree-optimization/69823
34299 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
34300 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
34302 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34304 * config/arc/arc-c.def: Add __NPS400__ definition.
34305 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
34306 (TARGET_NPS400): Define.
34308 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34310 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
34312 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
34313 pointer, arch_info.
34314 (arc_cpu_types): Fill the arch_info field with a pointer into the
34315 arc_arch_types table.
34316 (arc_selected_cpu): Declare.
34317 * config/arc/arc.c (arc_selected_cpu): Make global.
34318 (arc_selected_arch): Delete.
34319 (arc_base_cpu): Delete.
34320 (arc_override_options): Remove references to deleted variables,
34321 update access to arch information.
34322 (ARC_OPT): Update access to arch information.
34323 (ARC_OPTX): Likewise.
34324 * config/arc/arc.h (arc_base_cpu): Remove declaration.
34325 (TARGET_ARC600): Update access to arch information.
34326 (TARGET_ARC601): Likewise.
34327 (TARGET_ARC700): Likewise.
34328 (TARGET_EM): Likewise.
34329 (TARGET_HS): Likewise.
34330 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
34333 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
34336 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
34337 condition/operands for integer GE/LE/GEU/LEU operations.
34339 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
34341 PR translation/79397
34342 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
34345 2017-02-08 Martin Jambor <mjambor@suse.cz>
34348 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
34349 whether allocation happened.
34350 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
34351 nothing was allocated.
34353 2017-02-08 Jakub Jelinek <jakub@redhat.com>
34355 PR tree-optimization/79408
34356 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
34357 constant, but SSA_NAME with a known integer range, use the minimum
34358 of that range instead of op1 to determine if modulo can be replaced
34359 with its first operand.
34361 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34363 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
34365 2017-02-08 Richard Biener <rguenther@suse.de>
34367 PR tree-optimization/71824
34368 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34369 Check all loops contained in the merged region.
34371 2017-02-07 Andrew Pinski <apinski@cavium.com>
34373 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
34375 2017-02-07 Andrew Pinski <apinski@cavium.com>
34377 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
34378 (thunderxt88): Likewise.
34379 (thunderxt81): Disable LSE and change v8.1 to v8.
34380 (thunderxt83): Likewise.
34382 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34383 Richard Biener <rguenther@suse.de>
34385 PR middle-end/79399
34386 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
34387 type from int to size_t.
34388 * ira-costs.c (struct_costs_size): Change type from int to size_t.
34390 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34392 PR rtl-optimization/79386
34393 * cprop.c (bypass_conditional_jumps): Initialize
34394 bypass_last_basic_block already before splitting bbs after
34395 unconditional traps...
34396 (bypass_conditional_jumps): ... rather than here.
34399 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
34400 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
34401 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
34402 fix -masm=intel patterns.
34404 2017-02-07 Richard Biener <rguenther@suse.de>
34406 PR tree-optimization/79256
34407 PR middle-end/79278
34408 * builtins.c (get_object_alignment_2): Use min_align_of_type
34409 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
34410 and ADJUST_FIELD_ALIGN.
34412 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
34414 * doc/tm.texi: Regenerate.
34415 * stor-layout.c (layout_decl): Adjust.
34416 (update_alignment_for_field): Likewise.
34417 (place_field): Likewise.
34418 (min_align_of_type): Likewise.
34419 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
34420 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
34421 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
34422 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
34423 * config/frv/frv.c (frv_adjust_field_align): Likewise.
34424 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
34425 * config/i386/i386.c (x86_field_alignment): Likewise.
34426 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
34427 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
34428 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
34429 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
34430 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
34431 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
34435 2017-01-30 Richard Biener <rguenther@suse.de>
34437 PR tree-optimization/79256
34438 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34439 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34442 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
34444 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
34445 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
34446 builtins to SImode and emit a zero-extend, if necessary.
34448 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34450 * docs/invoke.texi (RISC-V Options): Alphabetize.
34452 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34454 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
34457 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34459 * config/riscv/riscv.c: New file.
34460 * common/config/riscv/riscv-common.c: Likewise.
34461 * config.gcc: Likewise.
34462 * config/riscv/constraints.md: Likewise.
34463 * config/riscv/elf.h: Likewise.
34464 * config/riscv/generic.md: Likewise.
34465 * config/riscv/linux.h: Likewise.
34466 * config/riscv/multilib-generator: Likewise.
34467 * config/riscv/peephole.md: Likewise.
34468 * config/riscv/pic.md: Likewise.
34469 * config/riscv/predicates.md: Likewise.
34470 * config/riscv/riscv-builtins.c: Likewise.
34471 * config/riscv/riscv-c.c: Likewise.
34472 * config/riscv/riscv-ftypes.def: Likewise.
34473 * config/riscv/riscv-modes.def: Likewise.
34474 * config/riscv/riscv-opts.h: Likewise.
34475 * config/riscv/riscv-protos.h: Likewise.
34476 * config/riscv/riscv.h: Likewise.
34477 * config/riscv/riscv.md: Likewise.
34478 * config/riscv/riscv.opt: Likewise.
34479 * config/riscv/sync.md: Likewise.
34480 * config/riscv/t-elf-multilib: Likewise.
34481 * config/riscv/t-linux: Likewise.
34482 * config/riscv/t-linux-multilib: Likewise.
34483 * config/riscv/t-riscv: Likewise.
34484 * configure.ac: Likewise.
34485 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
34486 Waterman as RISC-V maintainers.
34487 * doc/install.texi: Add RISC-V entries.
34488 * doc/invoke.texi: Add RISC-V options section.
34489 * doc/md.texi: Add RISC-V constraints section.
34490 * configure: Regenerated.
34492 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
34495 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
34496 false values to be constant vectors with all 0 or all 1 bits set.
34497 (vcondu<mode><mode>): Likewise.
34498 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
34500 (fpmask_comparison_operator): Update comment.
34501 (vecint_comparison_operator): New predicate.
34502 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
34503 vector conditionals when the true and false values are constant
34504 vectors with all 0 bits or all 1 bits set.
34506 2017-02-06 Martin Sebor <msebor@redhat.com>
34508 PR tree-optimization/79376
34509 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
34511 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
34513 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
34514 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
34515 to simplify split condition.
34517 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34519 * omp-expand.c (oxpand_omp_atomic_fetch_op,
34520 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
34523 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
34525 PR rtl-optimization/68664
34526 * target.def (can_speculate_insn): New hook.
34527 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
34528 * doc/tm.texi: Regenerate.
34529 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
34530 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
34531 (rs6000_sched_can_speculate_insn): New function.
34533 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34535 PR tree-optimization/79284
34536 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
34537 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
34538 vectorizable_mask_load_store, vectorizable_operation,
34539 vect_is_simple_cond, get_same_sized_vectype): Use it instead
34540 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
34541 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
34542 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
34543 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
34544 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
34545 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
34546 is_gimple_assign (stmt). Replace another such test with
34547 is_gimple_assign (stmt).
34549 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34552 * config/avr/avr.c (rtl-iter.h): Include it.
34553 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
34554 (avr_legitimate_combined_insn): ...and implementation.
34556 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
34558 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
34559 * config/s390/s390.c (s390_const_operand_ok)
34560 (s390_canonicalize_comparison, s390_extract_part)
34561 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
34562 (s390_contiguous_bitmask_p, s390_rtx_costs)
34563 (legitimize_pic_address): Likewise.
34564 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
34565 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
34566 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
34567 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
34568 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
34570 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34572 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
34573 REGNO($0) == REGNO($1).
34575 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34577 * config/s390/linux.h(SIZE_TYPE): Add comment.
34579 2017-02-06 Julian Brown <julian@codesourcery.com>
34580 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34581 Virendra Pathak <virendra.pathak@broadcom.com>
34583 * config/aarch64/aarch64-cores.def: Change the scheduler
34585 * config/aarch64/aarch64.md: Include thunderx2t99.md.
34586 * config/aarch64/thunderx2t99.md: New file.
34588 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34590 * doc/standards.texi (Go Language): Update link to language
34593 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
34595 * tree-eh.c (lower_resx): Sanitize profile.
34596 (cleanup_empty_eh_move_lp): Likewise.
34598 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
34601 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
34603 * cfgloopmanip.h (loop_version): Update prototype.
34604 * modulo-sched.c (sms_schedule): Update call of loop_version.
34605 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
34606 * tree-parloops.c (gen_parallel_loop): Likewise.
34607 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
34608 * tree-ssa-loop-split.c (split_loop): Likewise.
34609 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
34610 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
34612 2017-02-05 Martin Liska <mliska@suse.cz>
34615 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
34617 (print_operand_address): Initialize a struct to zero.
34619 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34621 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
34622 garbage collector only in textual form.
34624 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34626 * doc/extend.texi (x86 specific memory model extensions for
34627 transactional memory): Simplify a phrase.
34629 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
34632 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
34633 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
34634 (atomic_storedi_1): Likewise.
34636 2017-02-04 Jakub Jelinek <jakub@redhat.com>
34638 PR tree-optimization/79338
34639 * tree-parloops.c (gather_scalar_reductions): Don't call
34640 vect_analyze_loop_form for loop->inner before destroying loop's
34643 2017-02-03 Martin Sebor <msebor@redhat.com>
34645 PR tree-optimization/79327
34646 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
34647 when precision has resulted in leading zeros.
34648 (format_integer): Adjust the likely counter to assume an unknown
34649 argument that may be zero is non-zero.
34651 2017-02-03 Jason Merrill <jason@redhat.com>
34654 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
34655 avoid copying non-taken branch.
34657 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34659 PR tree-optimization/79340
34660 * tree-vect-loop.c (vectorizable_reduction): Release
34661 vec_defs elements after safe_splicing them into other vectors.
34664 PR tree-optimization/79327
34665 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
34666 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
34668 (format_integer): Use wide_int_to_tree instead of build_int_cst
34669 + to_?hwi. If argmin is NULL, just set argmin and argmax to
34670 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
34671 of shortest and longest sequence.
34673 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
34675 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
34676 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
34678 2017-02-03 Walter Lee <walt@tilera.com>
34681 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
34682 after initial stackframe link reg save.
34683 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
34685 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34688 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
34689 wu for stxssp alternative.
34691 2017-02-03 Martin Sebor <msebor@redhat.com>
34693 PR tree-optimization/79352
34694 * gimple-fold.c (get_range_strlen): Add argument.
34695 (get_range_strlen): Change return type to bool.
34696 (get_maxval_strlen): Pass in a dummy argument.
34697 * gimple-fold.h (get_range_strlen): Change return type to bool.
34698 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
34699 * tree.h (array_at_struct_end_p): Add argument.
34700 * tree.c (array_at_struct_end_p): Handle it.
34702 2017-02-03 Martin Liska <mliska@suse.cz>
34705 * multiple_target.c (create_dispatcher_calls): Redirect edge
34706 from a caller of a dispatcher.
34707 (expand_target_clones): Make the clones local.
34708 (ipa_target_clone): Do both target clones and resolvers.
34709 (ipa_dispatcher_calls): Remove the pass.
34710 (pass_dispatcher_calls::gate): Likewise.
34711 (make_pass_dispatcher_calls): Likewise.
34712 * passes.def (pass_target_clone): Put as very first IPA early
34715 2017-02-03 Martin Liska <mliska@suse.cz>
34717 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
34718 in case of a function with ifunc attribute.
34720 2017-02-03 Martin Liska <mliska@suse.cz>
34722 * cgraph.c (cgraph_node::dump): Dump function version info.
34723 * symtab.c (symtab_node::dump_base): Add missing new line.
34725 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34727 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
34728 (ifcombine_ifandif): Use it.
34730 2017-02-03 Martin Liska <mliska@suse.cz>
34732 * doc/invoke.texi: Document default value for
34733 use-after-scope-direct-emission-threshold.
34735 2017-02-03 Martin Liska <mliska@suse.cz>
34737 PR tree-optimization/79339
34738 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
34739 (format_floating): Likewise.
34741 2017-02-03 Martin Liska <mliska@suse.cz>
34744 * ipa-prop.c (ipa_node_params_t::insert): Remove current
34746 (ipa_node_params_t::remove): Likewise.
34747 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
34748 initialization from removed ipa_node_params_t::insert.
34749 (ipa_node_params::~ipa_node_params): Move from removed
34750 ipa_node_params_t::release.
34751 * symbol-summary.h (symbol_summary::m_released): New member.
34752 Do not release a summary twice. Do not allow to call finalizer
34753 for types of a summary that live in GGC memory.
34755 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34757 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
34760 2017-02-02 Martin Sebor <msebor@redhat.com>
34762 PR middle-end/79275
34763 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
34764 (format_string): Tighten up the range of output for non-constant
34765 strings and correct the expected range for wide non-constant strings.
34767 2017-02-02 Martin Sebor <msebor@redhat.com>
34769 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
34771 PR middle-end/32003
34772 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
34774 (-fdump-tree-@var): Add to index and document how to come up
34775 with pass-specific option and dump file names.
34776 (-fdump-passes): Clarify where to look for output.
34778 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34780 PR middle-end/77445
34781 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
34782 statistics of the analyzed path; allow threading for speed when
34783 any of BBs along the path are optimized for speed.
34785 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
34787 PR middle-end/78468
34788 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
34789 settings of the virtual registers.
34792 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
34794 * explow.c (get_dynamic_stack_size): Take known alignment of stack
34795 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
34798 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34800 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
34801 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
34803 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34805 * config/s390/s390.md: Add missing comments with the expanded
34807 * config/s390/vector.md: Likewise.
34808 * config/s390/vx-builtins.md: Likewise.
34810 2017-02-02 Jakub Jelinek <jakub@redhat.com>
34813 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
34814 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
34815 conditions on a single line.
34817 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34819 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34820 __S390_VX__ to __VX__.
34822 2017-02-01 Andrew Pinski <apinski@cavium.com>
34824 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
34825 stmt_info to record_stmt_cost.
34826 (vect_get_known_peeling_cost): Pass stmt_info if known to
34828 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
34829 cpu_vector_cost field into
34830 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
34831 field into vec_int_stmt_cost and vec_fp_stmt_cost.
34832 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
34833 splitting of scalar_stmt_cost and vec_stmt_cost.
34834 (thunderx_vector_cost): Likewise.
34835 (cortexa57_vector_cost): LIkewise.
34836 (exynosm1_vector_cost): Likewise.
34837 (xgene1_vector_cost): Likewise.
34838 (thunderx2t99_vector_cost): Improve after the splitting of the two
34840 (aarch64_builtin_vectorization_cost): Update for the splitting of
34841 scalar_stmt_cost and vec_stmt_cost.
34843 2017-02-01 Torvald Riegel <triegel@redhat.com>
34844 Richard Henderson <rth@redhat.com>
34846 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
34847 conditional on existance of a fast atomic load.
34848 * optabs-query.c (can_atomic_load_p): New function.
34849 * optabs-query.h (can_atomic_load_p): Declare it.
34850 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
34851 no fast atomic load is available for the particular size of access.
34852 (expand_atomic_compare_and_swap): Likewise.
34853 (expand_atomic_load): Likewise.
34854 (expand_atomic_store): Likewise.
34855 (expand_atomic_fetch_op): Likewise.
34856 * testsuite/lib/target-supports.exp
34857 (check_effective_target_sync_int_128): Remove x86 because it provides
34858 no fast atomic load.
34859 (check_effective_target_sync_int_128_runtime): Likewise.
34861 2017-02-01 Richard Biener <rguenther@suse.de>
34863 * graphite.c: Include tree-vectorizer.h for find_loop_location.
34864 (graphite_transform_loops): Provide opt-info for optimized nests.
34865 * tree-parloop.c (parallelize_loops): Provide opt-info for
34866 parallelized loops.
34868 2017-02-01 Richard Biener <rguenther@suse.de>
34870 PR middle-end/79315
34871 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
34872 was not set before.
34874 2017-02-01 Richard Biener <rguenther@suse.de>
34876 PR tree-optimization/71824
34877 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34878 Verify the loops are valid in the merged SESE region.
34879 (scop_detection::can_represent_loop_1): Check analyzing the
34880 evolution of the number of iterations in the region succeeds.
34882 2017-01-31 Ian Lance Taylor <iant@golang.org>
34884 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
34885 REG_ARGS_SIZE note to 32-bit push insns and call insn.
34887 2017-01-31 David Malcolm <dmalcolm@redhat.com>
34889 PR preprocessor/79210
34890 * input.c (get_substring_ranges_for_loc): Replace line_width
34891 assertion with error-handling.
34893 2017-01-31 Richard Biener <rguenther@suse.de>
34895 PR tree-optimization/77318
34896 * graphite-sese-to-poly.c (extract_affine): Fix assert.
34897 (create_pw_aff_from_tree): Take loop parameter.
34898 (add_condition_to_pbb): Pass loop of the condition to
34899 create_pw_aff_from_tree.
34901 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34903 * config/s390/s390.c (s390_asan_shadow_offset): New function.
34904 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
34906 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
34910 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
34912 (convert_int_to_float128): Likewise.
34913 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
34914 (convert_int_to_float128): Likewise.
34915 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
34916 (UNSPEC_IEEE128_CONVERT): Likewise.
34917 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
34918 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
34919 Use local variables for IBM extended format.
34920 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
34921 (fix_trunc<mode>si2_fprs): Likewise.
34922 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
34923 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
34924 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
34925 to know that we can now have integers of all sizes in vector
34927 (fix<uns>_<mode>di2_hw): Likewise.
34928 (float<uns>_<mode>si2_hw): Likewise.
34929 (fix_<mode>si2_hw): Likewise.
34930 (fixuns_<mode>si2_hw): Likewise.
34931 (float<uns>_<mode>di2_hw): Likewise.
34932 (float_<mode>di2_hw): Likewise.
34933 (float_<mode>si2_hw): Likewise.
34934 (floatuns_<mode>di2_hw): Likewise.
34935 (floatuns_<mode>si2_hw): Likewise.
34936 (xscvqp<su>wz_<mode>): Delete, no longer used.
34937 (xscvqp<su>dz_<mode>): Likewise.
34938 (xscv<su>dqp_<mode>): Likewise.
34939 (ieee128_mfvsrd_64bit): Likewise.
34940 (ieee128_mfvsrd_32bit): Likewise.
34941 (ieee128_mfvsrwz): Likewise.
34942 (ieee128_mtvsrw): Likewise.
34943 (ieee128_mtvsrd_64bit): Likewise.
34944 (ieee128_mtvsrd_32bit): Likewise.
34946 2017-01-31 Martin Liska <mliska@suse.cz>
34949 * ipa-prop.c (ipa_free_all_node_params): Call release method
34950 instead of ~sumbol_summary to not to trigger double times
34953 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
34955 PR tree-optimization/71691
34956 * bitmap.h (class auto_bitmap): New.
34957 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
34958 is_maybe_undefined instead of ssa_undefined_value_p.
34960 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34962 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34963 __S390_ARCH_LEVEL__ to __ARCH__.
34965 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34967 PR tree-optimization/79267
34968 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
34969 if should_remove_lhs_p is true.
34971 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
34974 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
34975 (add_alignment_attribute): New.
34976 (base_type_die): Add alignment attribute.
34977 (subrange_type_die): Likewise.
34978 (modified_type_die): Likewise.
34979 (gen_array_type_die): Likewise.
34980 (gen_descr_array_type_die: Likewise.
34981 (gen_enumeration_type_die): Likewise.
34982 (gen_subprogram_die): Likewise.
34983 (gen_variable_die): Likewise.
34984 (gen_field_die): Likewise.
34985 (gen_ptr_to_mbr_type_die): Likewise.
34986 (gen_struct_or_union_type_die): Likewise.
34987 (gen_subroutine_type_die): Likewise.
34988 (gen_typedef_die): Likewise.
34989 (base_type_cmp): Compare alignment attribute.
34991 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34994 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
34995 (setb_unsigned) New pattern for setb with CCUNS.
34996 * config/rs6000/rs6000.c (expand_block_compare): Use a different
34997 subfc./subfe sequence to avoid overflow problems. Generate a
34998 shorter sequence with cmpld/setb for power9.
34999 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35000 for generating subfc. instruction.
35001 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35002 now uses this instruction.
35004 2017-01-30 Ian Lance Taylor <iant@google.com>
35007 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35008 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35010 2017-01-30 Martin Sebor <msebor@redhat.com>
35012 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35013 Move constant to the right of a relational operator.
35014 (get_mpfr_format_length, format_character, format_string): Ditto.
35015 (should_warn_p, maybe_warn): Same.
35017 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35019 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35022 * asan.c (get_translation_unit_decl): Remove function.
35023 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35025 2017-01-30 Martin Liska <mliska@suse.cz>
35027 PR gcov-profile/79259
35028 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35029 -fprofile-generate.
35031 2017-01-30 Martin Liska <mliska@suse.cz>
35034 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35035 Initialize variables with NULL value.
35037 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35040 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35042 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35044 2017-01-30 Richard Biener <rguenther@suse.de>
35046 PR tree-optimization/79276
35047 * tree-vrp.c (process_assert_insertions): Properly adjust common
35048 when removing a duplicate.
35050 2017-01-30 Richard Biener <rguenther@suse.de>
35052 PR tree-optimization/79256
35053 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35054 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35056 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35058 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
35061 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
35062 ("*r<noxa>sbg_<mode>_sll_bitmask")
35063 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
35064 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
35065 Use contiguous_bitmask_nowrap_operand.
35067 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35070 * config/rs6000/altivec.h (vec_xl): Revise #define.
35071 (vec_xst): Likewise.
35073 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
35075 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
35077 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
35079 PR rtl-optimization/79194
35080 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
35081 traps before call to bypass_conditional_jumps.
35083 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35085 PR tree-optimization/71374
35086 * lra-constraints.c (check_conflict_input_operands): New.
35087 (match_reload): Use it.
35089 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35092 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
35093 account to calculate conflict_set.
35095 2017-01-27 Bin Cheng <bin.cheng@arm.com>
35097 PR rtl-optimization/78559
35098 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
35099 other_insn in combine.
35101 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
35103 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
35104 uint16_type_node for BT_UINT16.
35106 2017-01-27 David Malcolm <dmalcolm@redhat.com>
35108 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
35109 "RTL Tests" to menu.
35110 (GIMPLE Tests): New node.
35111 (RTL Tests): New node.
35113 2017-01-27 Richard Biener <rguenther@suse.de>
35115 PR tree-optimization/79245
35116 * tree-loop-distribution.c (distribute_loop): Apply cost
35117 modeling also to detected patterns.
35119 2017-01-27 Richard Biener <rguenther@suse.de>
35121 PR tree-optimization/71433
35122 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
35123 (compare_assert_loc): New function.
35124 (process_assert_insertions): Sort and optimize assert locations
35125 to remove duplicates and push down identical assertions on
35126 edges to their destination block.
35128 2017-01-27 Richard Biener <rguenther@suse.de>
35130 PR tree-optimization/79244
35131 * tree-vrp.c (remove_range_assertions): Forcefully propagate
35132 out SSA names even if abnormal.
35134 2017-01-27 Jakub Jelinek <jakub@redhat.com>
35136 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
35137 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
35138 instead of MPFR_RNDN.
35140 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
35143 * arm.c (arm_option_override): Don't call build_target_option_node
35144 until after doing all option overrides.
35145 (arm_valid_target_attribute_tree): Likewise.
35147 2017-01-27 Martin Liska <mliska@suse.cz>
35149 * doc/invoke.texi (-fprofile-arcs): Document profiling support
35150 for {cd}tors and C++ {cd}tors.
35152 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
35154 * config/s390/s390.md ("*setmem_long_and")
35155 ("*setmem_long_and_31z"): Use zero_extend instead of and.
35157 2017-01-26 Martin Sebor <msebor@redhat.com>
35159 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
35162 2017-01-26 Martin Sebor <msebor@redhat.com>
35164 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
35165 HAVE_DFmode before using XFmode or DFmode.
35166 (parse_directive): Avoid using the z length modifier to avoid
35167 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
35169 PR middle-end/78703
35170 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
35171 to accept adjustment as an array.
35172 (get_int_range): New function.
35173 (struct directive): Make width and prec arrays.
35174 (directive::set_width, directive::set_precision): Call get_int_range.
35175 (format_integer, format_floating): Handle width and precision ranges.
35176 (format_string, parse_directive): Same.
35178 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35181 * dwarf2out.c (generate_skeleton_bottom_up): For children with
35182 comdat_type_p set, just clone them, but keep the children in the
35186 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
35187 which have direct callers with -fvar-tracking-assignments enabled
35189 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
35190 inside of type units.
35192 2017-01-26 Martin Sebor <msebor@redhat.com>
35194 PR middle-end/78703
35195 * gimple-ssa-sprintf.c (struct result_range): Add likely and
35197 (struct format_result): Replace number_chars, number_chars_min,
35198 and number_chars_max with a single member of struct result_range.
35200 (format_result::operator+=): Adjust.
35201 (struct fmtresult): Remove bounded. Handle likely and unlikely
35203 (fmtresult::adjust_for_width_or_precision): New function.
35204 (fmtresult:type_max_digits): New function.
35205 (bytes_remaining): Handle likely and unlikely counters.
35206 (min_bytes_remaining): Remove.
35207 (format_percent): Simplify.
35208 (format_integer, format_floating): Set likely and unlikely counters.
35209 (get_string_length, format_character, format_string): Same.
35210 (format_plain, should_warn_p): New function.
35211 (maybe_warn): Call should_warn_p. Update diagnostic messages
35212 and handle those for all directives, including plain strings.
35213 (format_directive): Handle likely and unlikely counters.
35214 Remove unnecessary quoting from diagnostics. Add an informational
35216 (add_bytes): Remove.
35217 (pass_sprintf_length::compute_format_length): Simplify.
35218 (try_substitute_return_value): Handle likely and unlikely counters.
35220 2017-01-26 Carl Love <cel@us.ibm.com>
35222 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
35223 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
35225 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
35228 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
35229 endianess for subregs into account.
35230 * lra-constraints.c (lra_constraints): Do risky transformations
35231 always on the first iteration.
35232 * lra-lives.c (check_pseudos_live_through_calls): Add arg
35233 last_call_used_reg_set.
35234 (process_bb_lives): Define and use last_call_used_reg_set.
35235 * lra.c (lra): Always continue after lra_constraints on the first
35238 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
35240 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
35242 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
35244 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35246 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
35247 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
35248 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
35249 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
35250 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
35251 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
35252 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
35253 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
35254 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
35256 2017-01-26 Marek Polacek <polacek@redhat.com>
35259 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
35260 for the third operand.
35262 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35264 PR middle-end/79236
35265 * omp-low.c (struct omp_context): Add simt_stmt field.
35266 (scan_omp_for): Return omp_context *.
35267 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
35268 context to the _simt_ SIMD stmt.
35269 (lower_omp_for): For combined SIMD with sibling _simt_
35270 SIMD, make sure to use the same decls in _looptemp_
35271 clauses as in the sibling.
35273 2017-01-26 David Sherwood <david.sherwood@arm.com>
35275 PR middle-end/79212
35276 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
35279 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35282 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
35283 emit fld b; fld a; if possible.
35285 * brig-builtins.def: Update copyright years.
35286 * config/arm/arm_acle_builtins.def: Update copyright years.
35288 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
35291 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
35292 constraint instead of o for the stxsd instruction.
35294 2017-01-25 Carl Love <cel@us.ibm.com>
35296 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
35297 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
35299 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
35301 * doc/invoke.texi (C++ Dialect Options): Fix typo.
35303 2017-01-25 Richard Biener <rguenther@suse.de>
35305 PR tree-optimization/69264
35306 * target.def (vector_alignment_reachable): Improve documentation.
35307 * doc/tm.texi: Regenerate.
35308 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
35310 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
35311 earlier changes with respect to TYPE_USER_ALIGN.
35312 (vector_alignment_reachable_p): Likewise. Improve dumping.
35314 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35317 * config/arm/arm.md (xordi3): Force constant operand into a register
35320 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35322 * doc/invoke.texi (-fstore-merging): Correct default optimization
35323 levels at which it is enabled.
35324 (-O): Move -fstore-merging from list to...
35327 2017-01-25 Richard Biener <rguenther@suse.de>
35330 * omp-expand.c: Include debug.h.
35331 (expand_omp_taskreg): Make sure to generate early debug before
35332 outlining anything from a function.
35333 (expand_omp_target): Likewise.
35334 (grid_expand_target_grid_body): Likewise.
35336 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
35339 * asan.c (get_translation_unit_decl): New function.
35340 (asan_add_global): Extract modules file name from globals
35341 TRANSLATION_UNIT_DECL name.
35343 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
35346 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
35347 for long calls with APCS frame and VFP.
35349 2017-01-24 David Malcolm <dmalcolm@redhat.com>
35351 * cfg.c (original_copy_tables_initialized_p): New function.
35352 * cfg.h (original_copy_tables_initialized_p): New decl.
35353 * cfgrtl.c (relink_block_chain): Guard the call to
35354 free_original_copy_tables with a call to
35355 original_copy_tables_initialized_p.
35356 * cgraph.h (symtab_node::native_rtl_p): New decl.
35357 * cgraphunit.c (symtab_node::native_rtl_p): New function.
35358 (symtab_node::needed_p): Don't assert for early assembly output
35359 for __RTL functions.
35360 (cgraph_node::finalize_function): Set "force_output" for __RTL
35362 (cgraph_node::analyze): Bail out early for __RTL functions.
35363 (analyze_functions): Update assertion to support __RTL functions.
35364 (cgraph_node::expand): Bail out early for __RTL functions.
35365 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
35367 * function.h (struct function): Update comment for field
35369 * gimple-expr.c: Include "tree-pass.h".
35370 (gimple_has_body_p): Return false for __RTL functions.
35371 * Makefile.in (OBJS): Add run-rtl-passes.o.
35372 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
35374 (gcc::pass_manager::get_clean_slate): New accessor.
35375 * passes.c: Include "insn-addr.h".
35376 (should_skip_pass_p): Add logging. Update logic for running
35377 "expand" to be compatible with both __GIMPLE and __RTL. Guard
35378 property-provider override so it is only done for gimple passes.
35380 (skip_pass): New function.
35381 (execute_one_pass): Call skip_pass when skipping passes.
35382 * read-md.c (md_reader::read_char): Support filtering
35383 the input to a subset of line numbers.
35384 (md_reader::md_reader): Initialize fields
35385 m_first_line and m_last_line.
35386 (md_reader::read_file_fragment): New function.
35387 * read-md.h (md_reader::read_file_fragment): New decl.
35388 (md_reader::m_first_line): New field.
35389 (md_reader::m_last_line): New field.
35390 * read-rtl-function.c (function_reader::create_function): Only
35391 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
35392 curr_properties. Set DECL_INITIAL to a dummy block.
35393 (read_rtl_function_body_from_file_range): New function.
35394 * read-rtl-function.h (read_rtl_function_body_from_file_range):
35396 * run-rtl-passes.c: New file.
35397 * run-rtl-passes.h: New file.
35399 2017-01-24 Jeff Law <law@redhat.com>
35401 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
35404 2017-01-24 Bin Cheng <bin.cheng@arm.com>
35406 PR tree-optimization/79159
35407 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
35408 (record_nonwrapping_iv): Improve boundary using above function if no
35409 value range information.
35411 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
35412 Martin Jambor <mjambor@suse.cz>
35414 * brig-builtins.def: New file.
35415 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
35416 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
35417 (DEF_HSAIL_SAT_BUILTIN): Likewise.
35418 (DEF_HSAIL_INTR_BUILTIN): Likewise.
35419 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
35420 * builtin-types.def (BT_INT8): New.
35421 (BT_INT16): Likewise.
35422 (BT_UINT8): Likewise.
35423 (BT_UINT16): Likewise.
35424 (BT_FN_ULONG): Likewise.
35425 (BT_FN_UINT_INT): Likewise.
35426 (BT_FN_UINT_ULONG): Likewise.
35427 (BT_FN_UINT_LONG): Likewise.
35428 (BT_FN_UINT_PTR): Likewise.
35429 (BT_FN_ULONG_PTR): Likewise.
35430 (BT_FN_INT8_FLOAT): Likewise.
35431 (BT_FN_INT16_FLOAT): Likewise.
35432 (BT_FN_UINT32_FLOAT): Likewise.
35433 (BT_FN_UINT16_FLOAT): Likewise.
35434 (BT_FN_UINT8_FLOAT): Likewise.
35435 (BT_FN_UINT64_FLOAT): Likewise.
35436 (BT_FN_UINT16_UINT32): Likewise.
35437 (BT_FN_UINT32_UINT16): Likewise.
35438 (BT_FN_UINT16_UINT16_UINT16): Likewise.
35439 (BT_FN_INT_PTR_INT): Likewise.
35440 (BT_FN_UINT_PTR_UINT): Likewise.
35441 (BT_FN_LONG_PTR_LONG): Likewise.
35442 (BT_FN_ULONG_PTR_ULONG): Likewise.
35443 (BT_FN_VOID_UINT64_UINT64): Likewise.
35444 (BT_FN_UINT8_UINT8_UINT8): Likewise.
35445 (BT_FN_INT8_INT8_INT8): Likewise.
35446 (BT_FN_INT16_INT16_INT16): Likewise.
35447 (BT_FN_INT_INT_INT): Likewise.
35448 (BT_FN_UINT_FLOAT_UINT): Likewise.
35449 (BT_FN_FLOAT_UINT_UINT): Likewise.
35450 (BT_FN_ULONG_UINT_UINT): Likewise.
35451 (BT_FN_ULONG_UINT_PTR): Likewise.
35452 (BT_FN_ULONG_ULONG_ULONG): Likewise.
35453 (BT_FN_UINT_UINT_UINT): Likewise.
35454 (BT_FN_VOID_UINT_PTR): Likewise.
35455 (BT_FN_UINT_UINT_PTR: Likewise.
35456 (BT_FN_UINT32_UINT64_PTR): Likewise.
35457 (BT_FN_INT_INT_UINT_UINT): Likewise.
35458 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
35459 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
35460 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
35461 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
35462 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
35463 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
35464 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
35465 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
35466 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
35467 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
35468 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
35469 * doc/frontends.texi: List BRIG FE.
35470 * doc/install.texi (Testing): Add BRIG tesring requirements.
35471 * doc/invoke.texi (Overall Options): Mention BRIG.
35472 * doc/standards.texi (Standards): Doucment BRIG HSA version.
35474 2017-01-24 Richard Biener <rguenther@suse.de>
35476 PR translation/79208
35477 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
35479 2017-01-24 Martin Jambor <mjambor@suse.cz>
35482 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
35483 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
35484 and known_contexts.
35486 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
35488 PR middle-end/79123
35489 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
35490 casts from signed to unsigned really don't have a range.
35492 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
35494 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
35495 GMP_RNDx for compatiblity.
35497 2017-01-24 Martin Liska <mliska@suse.cz>
35500 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
35501 that would prevent us to call alloca with -1 as argument.
35503 2017-01-24 Jakub Jelinek <jakub@redhat.com>
35505 * dwarf2out.c (output_compilation_unit_header, output_file_names):
35506 Avoid -Wformat-security warning.
35508 2017-01-23 Andrew Pinski <apinski@cavium.com>
35510 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
35513 2017-01-23 Martin Sebor <msebor@redhat.com>
35515 PR middle-end/78703
35516 * gimple-ssa-sprintf.c (warn_level): New global.
35517 (format_integer): Use it here and throughout the rest of the file.
35518 Use the same switch to compute sign as base.
35519 (maybe_warn): New function.
35520 (format_directive): Factor out warnings into maybe_warn.
35521 Add debugging output. Use warn_level.
35522 (add_bytes): Use warn_level.
35523 (pass_sprintf_length::compute_format_length): Add debugging output.
35524 (try_substitute_return_value): Same.
35525 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
35527 PR middle-end/78703
35528 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
35529 (struct fmtresult, format_integer, format_floating): Adjust.
35530 (fmtresult::fmtresult): Set max correctly in two argument ctor.
35531 (get_string_length, format_string,format_directive): Same.
35532 (pass_sprintf_length::compute_format_length): Same.
35533 (try_substitute_return_value): Simplify slightly.
35535 PR middle-end/78703
35536 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
35537 (fmtresult::operator+=): Outlined.
35538 (struct fmtresult): Add ctors.
35539 (struct conversion_spec): Rename...
35540 (struct directive): ...to this. Add and remove data members.
35541 (directive::set_width, directive::set_precision): New functions.
35542 (format_percent): Use fmtresult ctor.
35543 (get_width_and_precision): Remove.
35544 (format_integer): Make naming changes. Avoid computing width and
35546 (format_floating): Same. Adjust indentation.
35547 (format_character, format_none): New functions.
35548 (format_string): Moved character handling to format_character.
35549 (format_directive): Remove arguments, change return type.
35550 (parse_directive): New function.
35551 (pass_sprintf_length::compute_format_length): Move directive
35552 parsing to parse_directive.
35554 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35556 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
35557 (assign_assembler_name_if_needed): ... this.
35558 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
35559 (assign_assembler_name_if_needed): ... this.
35560 (free_lang_data_in_cgraph): Adjust callers.
35561 * cgraphunit.c (cgraph_node::analyze): Likewise.
35562 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
35565 2017-01-23 Richard Biener <rguenther@suse.de>
35567 PR tree-optimization/79088
35568 PR tree-optimization/79188
35569 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
35570 resetting loop bounds after last path deletion. Reset loop
35571 bounds of the target loop, make code match the comments.
35572 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
35573 Make sure loops need no fixups.
35575 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35577 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
35578 exponent support with double type for first argument.
35579 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
35580 type returned by __builtin_vec_extract_sig,
35581 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
35582 functions from "vector int" to "vector unsigned int" or from
35583 "vector long long int" to "vector unsigned long long int".
35584 Changed type returned by __builtin_vec_extract_exp,
35585 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
35586 functions from "vector int" to "vector unsigned int" or from
35587 "vector long long int" to "vector unsigned long long int".
35588 Changed return type of __builtin_vec_test_data_class,
35589 __builtin_vec_test_data_class_sp, and
35590 __builtin_vec_test_data_class_dp from "vector int" to
35591 "vector bool int" or from "vector long long int" to "vector bool
35592 long long int" and changed second argument type from "unsigned
35593 int" to "int". Added new overloaded function forms "vector float
35594 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
35595 "vector float __builtin_vec_insert_exp_sp (vector float, vector
35596 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
35597 double, vector unsigned long long int)" and "vector double
35598 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
35599 long int)". Changed return type of
35600 __builtin_scalar_test_data_class and
35601 __builtin_scalar_test_data_class_sp and
35602 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
35603 int" and changed second argument from "unsigned int" to "int".
35604 Changed type returned by __builtin_scalar_test_neg,
35605 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
35606 from "int" to "bool int". Added new overloaded function form
35607 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
35608 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
35609 exponent double-precision with floating point first argument.
35610 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
35611 documentation of scalar_test_data_class, scalar_test_neg,
35612 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
35613 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
35614 vec_test_data_class built-in functions to reflect refinements in
35615 their type signatures.
35617 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
35619 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
35621 (aarch64_elf_asm_destructor): Likewise.
35623 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
35625 PR rtl-optimization/78634
35626 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
35627 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
35628 * ifcvt.c (noce_try_cmove): Add missing cost check.
35630 PR rtl-optimization/71724
35631 * combine.c (if_then_else_cond): Look for situations where it is
35632 beneficial to undo the work of one of the recursive calls.
35634 2017-01-23 Bin Cheng <bin.cheng@arm.com>
35636 PR tree-optimization/70754
35637 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
35638 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
35639 combined stmt before it if not NULL.
35640 (combine_chains): Process refs reversely and compute dominance point
35643 2017-01-23 Martin Liska <mliska@suse.cz>
35645 PR tree-optimization/79196
35646 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
35647 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
35649 (strlen_optimize_stmt): Call the renamed function.
35651 2017-01-23 Michael Matz <matz@suse.de>
35653 PR tree-optimization/78384
35654 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
35656 2017-01-23 Richard Biener <rguenther@suse.de>
35658 PR tree-optimization/79186
35659 * tree-vrp.c (register_new_assert_for): Make sure we've seen
35660 both incoming edges before moving an assert.
35662 2017-01-23 Martin Jambor <mjambor@suse.cz>
35664 * ipa-prop.c (load_from_param_1): Removed.
35665 (load_from_unmodified_param): Bits from load_from_param_1 put back
35667 (load_from_param): Removed.
35668 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
35669 with stmt. Reverted back to use of load_from_unmodified_param.
35671 2017-01-23 Martin Jambor <mjambor@suse.cz>
35674 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
35675 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
35676 field a pointer to garbage collected vector, mark lattices and
35677 ipcp_orig_node with GTY((skip)).
35678 (ipa_get_param_count): Adjust to descriptors being a pointer.
35679 (ipa_get_param): Likewise.
35680 (ipa_get_type): Likewise.
35681 (ipa_get_param_move_cost): Likewise.
35682 (ipa_set_param_used): Likewise.
35683 (ipa_get_controlled_uses): Likewise.
35684 (ipa_set_controlled_uses): Likewise.
35685 (ipa_is_param_used): Likewise.
35686 (ipa_node_params_t): Move into garbage collector. New methods insert
35688 (ipa_node_params_sum): Annotate wth GTY(()).
35689 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
35691 (ipa_load_from_parm_agg): Adjust declaration.
35692 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
35693 * ipa-profile.c (ipa_profile): Likewise.
35694 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
35695 (ipa_populate_param_decls): Make descriptors parameter garbage
35697 (ipa_dump_param): Adjust to descriptors being a pointer.
35698 (ipa_alloc_node_params): Likewise.
35699 (ipa_initialize_node_params): Likewise.
35700 (load_from_param_1): Make descriptors parameter garbage collected.
35701 (load_from_unmodified_param): Likewise.
35702 (load_from_param): Likewise.
35703 (ipa_load_from_parm_agg): Likewise.
35704 (ipa_node_params::~ipa_node_params): Removed.
35705 (ipa_free_all_node_params): Remove call to delete operator.
35706 (ipa_node_params_t::insert): New.
35707 (ipa_node_params_t::remove): Likewise.
35708 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
35709 copy known_csts and known_contexts vectors.
35710 (ipa_read_node_info): Adjust to descriptors being a pointer.
35711 (ipcp_modif_dom_walker): Make m_descriptors field garbage
35713 (ipcp_transform_function): Make descriptors variable garbage
35716 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
35718 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
35719 * config/i386/avx512dqintrin.h: Ditto.
35720 * config/i386/avx512fintrin.h: Ditto.
35721 * config/i386/i386.c: Handle new builtins.
35722 * config/i386/i386-builtin.def: Add new builtins.
35723 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
35724 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
35726 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35727 Martin Liska <mliska@suse.cz>
35729 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
35730 * asan.c (asan_expand_poison_ifn): Support stores and use
35731 appropriate ASAN report function.
35732 * internal-fn.c (expand_ASAN_POISON_USE): New function.
35733 * internal-fn.def (ASAN_POISON_USE): Declare.
35734 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
35735 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
35736 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
35737 ASAN_POISON calls w/o LHS.
35738 * tree-ssa.c (execute_update_addresses_taken): Create clobber
35739 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
35740 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
35741 * gimplify.c (asan_poison_variables): Add attribute
35742 use_after_scope_memory to variables that really needs to live
35744 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
35745 having the attribute.
35747 2017-01-23 Martin Liska <mliska@suse.cz>
35749 * asan.c (create_asan_shadow_var): New function.
35750 (asan_expand_poison_ifn): Likewise.
35751 * asan.h (asan_expand_poison_ifn): New declaration.
35752 * internal-fn.c (expand_ASAN_POISON): Likewise.
35753 * internal-fn.def (ASAN_POISON): New builtin.
35754 * sanopt.c (pass_sanopt::execute): Expand
35755 asan_expand_poison_ifn.
35756 * tree-inline.c (copy_decl_for_dup_finish): Make function
35758 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
35759 * tree-ssa.c (is_asan_mark_p): New function.
35760 (execute_update_addresses_taken): Rewrite local variables
35761 (identified just by use-after-scope as addressable) into SSA.
35763 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35765 * doc/install.texi (Specific): opensource.apple.com uses https
35766 now. Remove trailing slash.
35768 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35770 * README.Portability: Remove note on an Irix compatibility issue.
35772 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
35774 * gcov.c (INCLUDE_ALGORITHM): Define.
35775 (INCLUDE_VECTOR): Define.
35776 No longer include <vector> and <algorithm> directly.
35778 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
35780 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
35782 * doc/invoke.texi (Code Gen Options): Ditto.
35784 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
35787 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
35789 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
35791 rtl-optimization/79125
35792 * cprop.c (local_cprop_pass): Handle cases where we make an
35793 unconditional trap.
35795 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
35799 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
35800 read from, for big endian.
35802 2017-01-20 Jiong Wang <jiong.wang@arm.com>
35804 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
35805 register pauth builtins for LP64 only.
35807 2017-01-20 Marek Polacek <polacek@redhat.com>
35810 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
35813 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35815 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
35817 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
35818 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
35819 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
35821 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35824 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
35825 in big-endian mode when they are not a single duplicated value.
35827 2017-01-20 Richard Biener <rguenther@suse.de>
35829 * BASE-VER: Bump to 7.0.1.
35831 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35833 * omp-low.c (omplow_simd_context): New struct. Use it...
35834 (lower_rec_simd_input_clauses): ...here and...
35835 (lower_rec_input_clauses): ...here to hold common data. Adjust all
35836 references to idx, lane, max_vf, is_simt.
35838 2017-01-20 Graham Markall <graham.markall@embecosm.com>
35840 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
35843 2017-01-20 Martin Jambor <mjambor@suse.cz>
35845 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
35846 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
35848 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
35849 (GTFILES): Rename hsa.c to hsa-common.c.
35850 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
35851 * hsa-dump.c: Likewise.
35852 * hsa-gen.c: Likewise.
35853 * hsa-regalloc.c: Likewise.
35854 * ipa-hsa.c: Likewise.
35855 * omp-expand.c: Likewise.
35856 * omp-low.c: Likewise.
35857 * toplev.c: Likewise.
35859 2017-01-20 Marek Polacek <polacek@redhat.com>
35862 * doc/invoke.texi: Document -Wduplicated-branches.
35863 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
35864 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
35865 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
35866 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
35867 return 0 only when not OEP_LEXICOGRAPHIC.
35868 (fold_build_cleanup_point_expr): Use the expression
35869 location when building CLEANUP_POINT_EXPR.
35870 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
35871 * tree.c (add_expr): Handle error_mark_node.
35873 2017-01-20 Martin Liska <mliska@suse.cz>
35876 * tree-profile.c (init_ic_make_global_vars): Do not call
35878 (gimple_init_gcov_profiler): Likewise.
35880 2017-01-20 Martin Liska <mliska@suse.cz>
35883 * cgraph.h (maybe_create_reference): Remove argument and
35885 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
35887 * ipa-cp.c (create_specialized_node): Likewise.
35888 * symtab.c (symtab_node::maybe_create_reference): Handle
35889 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
35891 2017-01-20 Martin Liska <mliska@suse.cz>
35893 * read-rtl-function.c (function_reader::create_function): Use
35894 build_decl instread of build_decl_stat.
35896 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
35898 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
35899 * config/i386/avx512dqintrin.h: Ditto.
35900 * config/i386/avx512fintrin.h: Ditto.
35901 * config/i386/i386-builtin-types.def: Add new types.
35902 * config/i386/i386.c: Handle new types.
35903 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
35904 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
35905 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
35906 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
35907 (__builtin_ia32_kshiftridi): New.
35908 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
35910 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
35914 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
35915 define to rs6000_init_stack_protect_guard.
35916 (rs6000_init_stack_protect_guard): New function.
35918 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35919 Yunqiang Su <yunqiang.su@imgtec.com>
35921 * config.gcc (supported_defaults): Add madd4.
35922 (with_madd4): Add validation.
35923 (all_defaults): Add madd4.
35924 * config/mips/mips.opt (mmadd4): New option.
35925 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35927 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35929 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
35930 (ISA_HAS_FUSED_MADD4): Likewise.
35931 * doc/invoke.texi (-mmadd4): Document the new option.
35932 * doc/install.texi (--with-madd4): Document the new option.
35934 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35936 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
35937 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
35938 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
35939 (aarch64_init_pauth_hint_builtins): New.
35940 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
35941 (aarch64_expand_builtin): Expand new builtins.
35943 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35945 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
35946 * combine-stack-adj.c (no_unhandled_cfa): Handle
35947 REG_CFA_TOGGLE_RA_MANGLE.
35948 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
35949 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
35950 info for return address signing.
35951 (aarch64_expand_epilogue): Likewise.
35953 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35955 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
35956 * config/aarch64/aarch64-protos.h
35957 (aarch64_return_address_signing_enabled): New declaration.
35958 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
35960 (aarch64_expand_prologue): Sign return address before it's pushed onto
35962 (aarch64_expand_epilogue): Authenticate return address fetched from
35964 (aarch64_override_options): Sanity check for ILP32 and ISA level.
35965 (aarch64_attributes): New function attributes for "sign-return-address".
35966 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
35967 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
35968 ("*do_return"): Generate combined instructions according to key index.
35969 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
35970 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
35972 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
35973 * config/aarch64/aarch64.opt (msign-return-address=): New.
35974 * doc/extend.texi (AArch64 Function Attributes): Documents
35975 "sign-return-address=".
35976 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
35978 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35980 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
35981 overall option summary.
35983 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35985 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
35986 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
35987 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
35988 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
35990 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
35992 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
35993 -mpower9-minmax by default for -mcpu=power9.
35994 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
35995 128-bit floating point.
35997 2017-01-20 Alan Modra <amodra@gmail.com>
35999 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36000 optimizing for size.
36002 2017-01-20 Alan Modra <amodra@gmail.com>
36005 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36006 for strcmp and strncmp from corresponding builtin decl.
36008 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36010 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36011 instead of i386/rtems-64.h.
36012 * config/i386/rtems-64.h: Remove.
36014 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36018 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36020 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36022 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36024 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36025 Change int to HOST_WIDE_INT.
36026 * config/aarch64/aarch64-protos.h
36027 (aarch64_simd_gen_const_vector_dup): Likewise.
36028 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36030 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36032 * langhooks-def.h (lhd_type_for_size): New decl.
36033 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36034 * langhooks.c (lhd_type_for_size): New function, taken from
36037 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36039 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36040 define_bypass for CR latency.
36041 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36042 (power9-alu2): Add define_bypass for CR latency.
36044 (power9-mul): Update insn latency.
36045 (power9-mul-compare): Update insn latency, bypass latency and remove
36048 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36050 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36052 * config/aarch64/aarch64.md
36053 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36054 aarch64_nopcrelative_literal_loads.
36055 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36057 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
36059 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
36060 TARGET_LOONGSON_3A.
36061 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
36063 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
36066 * config.gcc (supported_defaults): Add lxc1-sxc1.
36067 (with_lxc1_sxc1): Add validation.
36068 (all_defaults): Add lxc1-sxc1.
36069 * config/mips/mips.opt (mlxc1-sxc1): New option.
36070 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36072 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36073 __mips_no_lxc1_sxc1.
36074 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
36075 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
36076 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
36078 2017-01-19 Richard Biener <rguenther@suse.de>
36080 PR tree-optimization/72488
36081 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
36082 sure to restore SSA info.
36083 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
36085 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
36087 PR rtl-optimization/79121
36088 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
36089 of the inner type when shifting an extended value.
36091 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36094 * symtab.c (symtab_node::equal_address_to): Fix comparing of
36095 interposable aliases.
36097 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
36100 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
36101 Use the evmergelohi instruction.
36102 (mov_si<mode>_e500_subreg4_2_le): Likewise.
36103 (mov_sitf_e500_subreg8_2_be): Likewise.
36104 (mov_sitf_e500_subreg12_2_le): Likewise.
36105 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
36106 (mov_si<mode>_e500_subreg4_2_be): Likewise.
36107 (mov_sitf_e500_subreg8_2_le): Likewise.
36108 (mov_sitf_e500_subreg12_2_be): Likewise.
36110 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36112 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
36113 attribute from vecsimple to vecperm.
36114 (altivec_vbpermq2): Likewise.
36116 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36119 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
36121 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36122 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
36123 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
36124 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
36125 case where N arg is SIZE_MAX.
36126 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
36127 (cmpstrsi): Add pattern.
36129 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
36131 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
36132 __builtin_vec_revb builtins.
36133 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
36134 built-in functions to support generation of the ISA 3.0 XXBR<x>
36135 vector byte reverse instructions.
36136 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
36137 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
36138 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
36139 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
36140 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
36141 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
36142 (P9V_BUILTIN_VEC_REVB): Likewise.
36143 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
36144 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
36145 (p9_xxbrq_v16qi): Likewise.
36146 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
36147 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
36148 (p9_xxbrh_v8hi): Likewise.
36149 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
36150 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
36151 vec_revb built-in functions.
36153 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
36155 PR rtl-optimization/78952
36156 * config/i386/i386.md (any_extract): New code iterator.
36157 (*insvqi_2): Use any_extract for source operand.
36158 (*insvqi_3): Use any_shiftrt for source operand.
36160 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
36162 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
36164 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
36166 2017-01-18 Matthias Klose <doko@ubuntu.com>
36168 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
36170 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36172 * config/rs6000/altivec.h (vec_bperm): Change #define.
36173 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
36174 (altivec_vbpermq2): New define_insn.
36175 (altivec_vbpermd): Likewise.
36176 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
36177 function interface.
36178 (VBPERMD): Likewise.
36179 (VBPERM): New polymorphic function interface.
36180 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
36181 Add entries for P9V_BUILTIN_VEC_VBPERM.
36182 * doc/extend.texi: Add interfaces for vec_bperm.
36184 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36186 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
36187 first letter of error messages.
36188 (s390_resolve_overloaded_builtin): Likewise.
36189 * config/s390/s390.c (s390_expand_builtin): Likewise.
36190 (s390_invalid_arg_for_unprototyped_fn): Likewise.
36191 (s390_valid_target_attribute_inner_p): Likewise.
36192 * config/s390/s390.md ("tabort"): Likewise.
36194 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
36196 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
36197 (ISA_AVOID_DIV_HILO): New macro.
36198 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
36199 (ISA_HAS_DDIV): Likewise.
36201 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36203 * doc/invoke.texi (fabi-version): Correct number of occurrences.
36205 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36207 * doc/invoke.texi (fabi-version): Spelling fix.
36209 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36212 * doc/invoke.texi (fabi-version): Mention mangling fix for
36215 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36218 * doc/invoke.texi (fabi-version): Document discriminator mangling.
36220 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
36223 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
36224 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
36226 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
36228 (stack_protect_test): Ditto.
36229 * config/rs6000/rs6000.opt (mstack-protector-guard=,
36230 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
36232 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
36233 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
36234 -mstack-protector-guard-offset=.
36235 (RS/6000 and PowerPC Options): Ditto.
36237 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36239 * config/i386/i386.h (MASK_CLASS_P): New define.
36240 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
36241 there are no registers from different register sets also when
36242 mask registers are used. Update function comment.
36243 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
36244 to (*k/*r) and (*k/*km) alternatives.
36246 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
36248 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
36249 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
36250 (EH_RETURN_HANDLER_RTX): New define.
36251 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
36252 Force frame pointer in EH return functions.
36253 (aarch64_expand_epilogue): Add barrier for eh_return.
36254 (aarch64_final_eh_return_addr): Remove.
36255 (aarch64_eh_return_handler_rtx): New function.
36256 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
36258 (aarch64_eh_return_handler_rtx): New prototype.
36260 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36262 * config/rs6000/altivec.h (vec_rlmi): New #define.
36263 (vec_vrlnm): Likewise.
36264 (vec_rlnm): Likewise.
36265 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
36266 (UNSPEC_VRLNM): Likewise.
36267 (VIlong): New mode iterator.
36268 (altivec_vrl<VI_char>mi): New define_insn.
36269 (altivec_vrl<VI_char>nm): Likewise.
36270 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
36272 (VRLDNM): Likewise.
36273 (RLNM): New polymorphic function entry.
36274 (VRLWMI): New monomorphic function entry.
36275 (VRLDMI): Likewise.
36276 (RLMI): New polymorphic function entry.
36277 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
36278 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
36279 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
36282 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36285 * dwarf2out.c (field_byte_offset): Restore the
36286 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
36287 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
36288 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
36291 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36294 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
36296 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36299 * dwarf2out.c (add_data_member_location_attribute): For constant
36300 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
36301 instead of DW_AT_data_member_location, DW_AT_bit_offset and
36302 DW_AT_byte_size attributes.
36304 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36306 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
36307 after forcing to constant memory when the code model is medium.
36309 2017-01-17 Julia Koval <julia.koval@intel.com>
36312 * config/i386/avx512fintrin.h
36313 (_mm512_i32gather_ps): Change __addr type to void const*.
36314 (_mm512_mask_i32gather_ps): Ditto.
36315 (_mm512_i32gather_pd): Ditto.
36316 (_mm512_mask_i32gather_pd): Ditto.
36317 (_mm512_i64gather_ps): Ditto.
36318 (_mm512_mask_i64gather_ps): Ditto.
36319 (_mm512_i64gather_pd): Ditto.
36320 (_mm512_mask_i64gather_pd): Ditto.
36321 (_mm512_i32gather_epi32): Ditto.
36322 (_mm512_mask_i32gather_epi32): Ditto.
36323 (_mm512_i32gather_epi64): Ditto.
36324 (_mm512_mask_i32gather_epi64): Ditto.
36325 (_mm512_i64gather_epi32): Ditto.
36326 (_mm512_mask_i64gather_epi32): Ditto.
36327 (_mm512_i64gather_epi64): Ditto.
36328 (_mm512_mask_i64gather_epi64): Ditto.
36329 (_mm512_i32scatter_ps): Change __addr type to void*.
36330 (_mm512_mask_i32scatter_ps): Ditto.
36331 (_mm512_i32scatter_pd): Ditto.
36332 (_mm512_mask_i32scatter_pd): Ditto.
36333 (_mm512_i64scatter_ps): Ditto.
36334 (_mm512_mask_i64scatter_ps): Ditto.
36335 (_mm512_i64scatter_pd): Ditto.
36336 (_mm512_mask_i64scatter_pd): Ditto.
36337 (_mm512_i32scatter_epi32): Ditto.
36338 (_mm512_mask_i32scatter_epi32): Ditto.
36339 (_mm512_i32scatter_epi64): Ditto.
36340 (_mm512_mask_i32scatter_epi64): Ditto.
36341 (_mm512_i64scatter_epi32): Ditto.
36342 (_mm512_mask_i64scatter_epi32): Ditto.
36343 (_mm512_i64scatter_epi64): Ditto.
36344 (_mm512_mask_i64scatter_epi64): Ditto.
36345 * config/i386/avx512pfintrin.h
36346 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
36347 (_mm512_mask_prefetch_i32gather_ps): Ditto.
36348 (_mm512_mask_prefetch_i64gather_pd): Ditto.
36349 (_mm512_mask_prefetch_i64gather_ps): Ditto.
36350 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
36351 (_mm512_prefetch_i32scatter_ps): Ditto.
36352 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
36353 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
36354 (_mm512_prefetch_i64scatter_pd): Ditto.
36355 (_mm512_prefetch_i64scatter_ps): Ditto.
36356 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
36357 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
36358 * config/i386/avx512vlintrin.h
36359 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
36360 (_mm_mmask_i32gather_ps): Ditto.
36361 (_mm256_mmask_i32gather_pd): Ditto.
36362 (_mm_mmask_i32gather_pd): Ditto.
36363 (_mm256_mmask_i64gather_ps): Ditto.
36364 (_mm_mmask_i64gather_ps): Ditto.
36365 (_mm256_mmask_i64gather_pd): Ditto.
36366 (_mm_mmask_i64gather_pd): Ditto.
36367 (_mm256_mmask_i32gather_epi32): Ditto.
36368 (_mm_mmask_i32gather_epi32): Ditto.
36369 (_mm256_mmask_i32gather_epi64): Ditto.
36370 (_mm_mmask_i32gather_epi64): Ditto.
36371 (_mm256_mmask_i64gather_epi32): Ditto.
36372 (_mm_mmask_i64gather_epi32): Ditto.
36373 (_mm256_mmask_i64gather_epi64): Ditto.
36374 (_mm_mmask_i64gather_epi64): Ditto.
36375 (_mm256_i32scatter_ps): Change __addr type to void*.
36376 (_mm256_mask_i32scatter_ps): Ditto.
36377 (_mm_i32scatter_ps): Ditto.
36378 (_mm_mask_i32scatter_ps): Ditto.
36379 (_mm256_i32scatter_pd): Ditto.
36380 (_mm256_mask_i32scatter_pd): Ditto.
36381 (_mm_i32scatter_pd): Ditto.
36382 (_mm_mask_i32scatter_pd): Ditto.
36383 (_mm256_i64scatter_ps): Ditto.
36384 (_mm256_mask_i64scatter_ps): Ditto.
36385 (_mm_i64scatter_ps): Ditto.
36386 (_mm_mask_i64scatter_ps): Ditto.
36387 (_mm256_i64scatter_pd): Ditto.
36388 (_mm256_mask_i64scatter_pd): Ditto.
36389 (_mm_i64scatter_pd): Ditto.
36390 (_mm_mask_i64scatter_pd): Ditto.
36391 (_mm256_i32scatter_epi32): Ditto.
36392 (_mm256_mask_i32scatter_epi32): Ditto.
36393 (_mm_i32scatter_epi32): Ditto.
36394 (_mm_mask_i32scatter_epi32): Ditto.
36395 (_mm256_i32scatter_epi64): Ditto.
36396 (_mm256_mask_i32scatter_epi64): Ditto.
36397 (_mm_i32scatter_epi64): Ditto.
36398 (_mm_mask_i32scatter_epi64): Ditto.
36399 (_mm256_i64scatter_epi32): Ditto.
36400 (_mm256_mask_i64scatter_epi32): Ditto.
36401 (_mm_i64scatter_epi32): Ditto.
36402 (_mm_mask_i64scatter_epi32): Ditto.
36403 (_mm256_i64scatter_epi64): Ditto.
36404 (_mm256_mask_i64scatter_epi64): Ditto.
36405 (_mm_i64scatter_epi64): Ditto.
36406 (_mm_mask_i64scatter_epi64): Ditto.
36407 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
36408 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
36409 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
36410 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
36411 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
36412 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
36413 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
36414 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
36415 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
36416 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
36417 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
36418 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
36419 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
36420 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
36421 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
36422 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
36423 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
36424 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
36425 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
36426 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
36427 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
36428 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
36429 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
36430 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
36431 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
36432 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
36433 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
36434 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
36435 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
36436 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
36437 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
36438 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
36439 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
36440 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
36441 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
36442 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
36443 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
36444 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
36445 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
36446 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
36447 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
36448 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
36449 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
36450 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
36451 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
36452 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
36453 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
36454 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
36455 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
36456 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
36457 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
36458 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
36459 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
36460 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
36461 definitions accordingly.
36463 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
36464 Kuan-Lin Chen <kuanlinchentw@gmail.com>
36467 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
36470 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
36473 * ira-conflicts.c (ira_build_conflicts): Update total conflict
36474 hard regs for inner regno.
36476 2017-01-17 Martin Liska <mliska@suse.cz>
36479 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
36480 assumption and add comment.
36482 2017-01-17 Nathan Sidwell <nathan@acm.org>
36484 * ipa-visibility.c (localize_node): New function, broken out of ...
36485 (function_and_variable_visibility): ... here. Call it.
36487 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36489 PR middle-end/77445
36490 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
36491 correctly set frequency of oudgoing edge.
36492 (duplicate_thread_path): Fix profile updating.
36494 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36497 * configure.ac: Add GCC_BASE_VER.
36498 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
36499 version from BASE-VER file.
36500 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
36501 (gcc.o): Depend on $(BASEVER).
36502 * common.opt (dumpfullversion): New option.
36503 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
36504 * doc/invoke.texi: Document -dumpfullversion.
36505 * doc/install.texi: Document --with-gcc-major-version-only.
36506 * configure: Regenerated.
36508 2017-01-17 Richard Biener <rguenther@suse.de>
36510 PR tree-optimization/71433
36511 * tree-vrp.c (register_new_assert_for): Merge same asserts
36512 on all incoming edges.
36513 (process_assert_insertions_for): Handle insertions at the
36516 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
36518 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
36519 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
36521 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
36524 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
36527 2017-01-17 Alan Modra <amodra@gmail.com>
36530 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
36531 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
36532 symbolic stack limit when pic.
36534 2017-01-16 Martin Sebor <msebor@redhat.com>
36536 PR tree-optimization/78608
36537 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
36539 2017-01-16 Jeff Law <law@redhat.com>
36542 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
36543 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
36544 for several include directories that may be relative to sysroot.
36545 * config/i386/x-mingw32 (gplus_includedir): Define.
36546 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
36547 (native_system_includedir): Likewise.
36548 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
36549 override if TARGET_SYSTEM_ROOT is defined.
36550 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
36552 PR tree-optimization/79090
36553 PR tree-optimization/33562
36554 PR tree-optimization/61912
36555 PR tree-optimization/77485
36556 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
36557 and computed trims into the dump file.
36559 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36561 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
36563 2017-01-16 Jakub Jelinek <jakub@redhat.com>
36566 * gimplify.c (gimplify_init_constructor): If want_value and
36567 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
36571 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
36572 sequence. Formatting fixes.
36573 (doloop_optimize): Formatting fixes.
36576 * gcc.c (debug_level_greater_than_spec_func): New function.
36577 (static_spec_functions): Add debug-level-gt spec function.
36578 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
36580 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
36581 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
36582 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
36583 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
36584 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
36585 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
36587 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
36589 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
36590 QImode fixups to general and mask registers only.
36592 2017-01-16 Carl Love <cel@us.ibm.com>
36594 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
36595 for built-in functions
36596 vector signed char vec_nabs (vector signed char)
36597 vector signed short vec_nabs (vector signed short)
36598 vector signed int vec_nabs (vector signed int)
36599 vector signed long long vec_nabs (vector signed long long)
36600 vector float vec_nabs (vector float)
36601 vector double vec_nabs (vector double)
36602 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
36604 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
36605 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
36606 * doc/extend.texi: Update the documentation file for the new built-in
36609 2017-01-16 Martin Sebor <msebor@redhat.com>
36611 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
36614 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36616 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
36617 UNSPEC_VSX__XXSPLTD to require special splat handling.
36619 2017-01-16 David Malcolm <dmalcolm@redhat.com>
36622 * system.h: Poison strndup.
36624 2017-01-16 Alan Modra <amodra@gmail.com>
36627 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
36630 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
36632 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
36634 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
36636 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
36637 call recog here. Assert that INSN_CODE (insn) is non-negative.
36639 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36642 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
36644 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
36645 in the currently scheduled RTL fragment.
36647 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36649 PR rtl-optimization/78751
36650 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
36653 2017-01-14 Jeff Law <law@redhat.com>
36655 PR tree-optimization/79090
36656 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
36657 variable length stores.
36658 (compute_trims): Delete dead assignment to *trim_tail.
36659 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
36662 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
36664 PR rtl-optimization/78626
36665 PR rtl-optimization/78727
36666 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
36667 of a block, and split such blocks after everything else is finished.
36669 2017-01-14 Alan Modra <amodra@gmail.com>
36672 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
36673 target legitimate_combined_insn.
36674 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
36675 (rs6000_legitimate_combined_insn): New function.
36676 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
36678 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
36679 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
36680 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
36682 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
36684 * doc/frontends.texi (G++ and GCC): Remove references to Java.
36686 2017-01-13 Jeff Law <law@redhat.com>
36688 PR tree-optimization/33562
36689 PR tree-optimization/61912
36690 PR tree-optimization/77485
36691 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
36693 (delete_dead_assignment): Likewise.
36694 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
36695 statement to delete_dead_call and delete_dead_assignment.
36697 2017-01-13 David Malcolm <dmalcolm@redhat.com>
36700 * substring-locations.c (format_warning_va): Strengthen case 1 so
36701 that both endpoints of the substring must be within the format
36702 range for just the substring to be printed.
36704 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
36706 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
36707 * config/i386/i386.c (ix86_target_string): Add missing options
36708 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
36709 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
36710 flags_other and ix86_target_other to flags2_other. Display unknown
36712 (ix86_valid_target_attribute_inner_p): Add missing options and
36713 reorder options by implied ISAs, as in ix86_target_string.
36715 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36717 * hash-table.h (hash_table::too_empty_p): New function.
36718 (hash_table::expand): Use it.
36719 (hash_table::traverse): Likewise.
36720 (hash_table::empty_slot): Use sizeof (value_type) instead of
36721 sizeof (PTR) to convert bytes to elements. Shrink the table
36722 if the current size is excessive for the current number of
36725 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36727 * ira-costs.c (record_reg_classes): Break from the inner loop
36728 early once alt_fail is known to be true. Update outer loop
36729 handling accordingly.
36731 2017-01-13 Jeff Law <law@redhat.com>
36733 * tree-ssa-dse.c (decrement_count): New function.
36734 (increment_start_addr, maybe_trim_memstar_call): Likewise.
36735 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
36736 when we know the partially dead statement is a mem* function.
36738 PR tree-optimization/61912
36739 PR tree-optimization/77485
36740 * tree-ssa-dse.c: Include expr.h.
36741 (maybe_trim_constructor_store): New function.
36742 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
36744 PR tree-optimization/33562
36745 PR tree-optimization/61912
36746 PR tree-optimization/77485
36747 * doc/invoke.texi: Document new dse-max-object-size param.
36748 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
36749 * tree-ssa-dse.c: Include params.h.
36750 (dse_store_status): New enum.
36751 (initialize_ao_ref_for_dse): New, partially extracted from
36753 (valid_ao_ref_for_dse, normalize_ref): New.
36754 (setup_live_bytes_from_ref, compute_trims): Likewise.
36755 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
36756 (maybe_trim_partially_dead_store): Likewise.
36757 (maybe_trim_complex_store): Likewise.
36758 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
36759 Track what bytes live from the original store. Return tri-state
36760 for dead, partially dead or live.
36761 (dse_dom_walker): Add constructor, destructor and new private members.
36762 (delete_dead_call, delete_dead_assignment): New extracted from
36764 (dse_optimize_stmt): Make a member of dse_dom_walker.
36765 Use initialize_ao_ref_for_dse.
36767 PR tree-optimization/33562
36768 PR tree-optimization/61912
36769 PR tree-optimization/77485
36770 * sbitmap.h (bitmap_count_bits): Prototype.
36771 (bitmap_clear_range, bitmap_set_range): Likewise.
36772 * sbitmap.c (bitmap_clear_range): New function.
36773 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
36775 2017-01-13 Martin Liska <mliska@suse.cz>
36778 * function.c (set_cfun): Add new argument force.
36779 * function.h (set_cfun): Likewise.
36780 * ipa-inline-transform.c (inline_call): Use the function when
36781 strict alising from is dropped for function we inline to.
36783 2017-01-13 Richard Biener <rguenther@suse.de>
36785 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
36786 for dumping GIMPLE INTEGER_CSTs.
36788 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
36790 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
36791 to 201112L since C++17.
36793 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
36796 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
36797 if -fsanitize=kernel-address is present.
36799 2017-01-13 Richard Biener <rguenther@suse.de>
36801 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
36802 as _Literal ( type ) number in case usual suffixes do not
36803 preserve all information.
36805 2017-01-13 Richard Biener <rguenther@suse.de>
36807 PR tree-optimization/77283
36808 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
36809 and ssa-iterators.h.
36810 (is_feasible_trace): Implement a cost model based on joiner
36813 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
36816 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
36817 char or short to __float128/_Float128 directly.
36819 2017-01-12 Martin Sebor <msebor@redhat.com>
36821 to -Wformat-overflow.
36822 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
36823 (min_bytes_remaining): Same.
36824 (get_string_length): Same.
36825 (format_string): Same.
36826 (format_directive): Same.
36828 (pass_sprintf_length::handle_gimple_call): Same.
36830 2017-01-12 Jakub Jelinek <jakub@redhat.com>
36832 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
36833 info.nowrite calls with no lhs that can't throw. Return bool
36834 whether gsi_remove has been called or not.
36835 (pass_sprintf_length::handle_gimple_call): Return bool whether
36836 try_substitute_return_value called gsi_remove. Formatting fix.
36837 (pass_sprintf_length::execute): Don't use gsi_remove if
36838 handle_gimple_call returned true.
36841 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
36842 be removed due to side-effects, don't remove following barrier nor
36843 turn the successor edge into fallthru edge.
36845 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36848 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
36849 element-reversing loads and stores as not swappable.
36851 2017-01-12 Nathan Sidwell <nathan@acm.org>
36852 Nicolai Stange <nicstange@gmail.com>
36854 * combine.c (try_combine): Don't ignore result of overlap checking
36855 loop. Combine overlap & asm check into single loop.
36857 2017-01-12 Richard Biener <rguenther@suse.de>
36859 * tree-pretty-print.c (dump_generic_node): Provide -gimple
36860 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
36862 2017-01-12 Richard Biener <rguenther@suse.de>
36864 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
36865 and TS_TARGET_OPTION directly derive from TS_BASE.
36866 * tree-core.h (tree_optimization_option): Derive from tree_base.
36867 (tree_target_option): Likewise.
36869 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
36871 * config/i386/i386.c (memory_address_length): Increase len
36872 only when rip_relative_addr_p returns false.
36874 2017-01-11 Julia Koval <julia.koval@intel.com>
36876 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
36877 (OPTION_MASK_ISA_SGX_SET): New.
36878 (ix86_handle_option): Handle OPT_msgx.
36879 * config.gcc: Added sgxintrin.h.
36880 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
36881 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
36882 * config/i386/i386.c (ix86_target_string): Add -msgx.
36884 (ix86_option_override_internal): Handle new options.
36885 (ix86_valid_target_attribute_inner_p): Add sgx.
36886 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
36887 * config/i386/i386.opt: Add msgx.
36888 * config/i386/sgxintrin.h: New file.
36889 * config/i386/x86intrin.h: Add sgxintrin.h.
36891 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36894 * fold-const.c (maybe_nonzero_address): Return 1 for function
36896 (tree_single_nonzero_warnv_p): Don't handle function local objects
36900 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
36903 2017-01-11 David Malcolm <dmalcolm@redhat.com>
36906 * opts.c: Include "spellcheck.h"
36907 (struct string_fragment): New struct.
36908 (struct edit_distance_traits<const string_fragment &>): New
36910 (get_closest_sanitizer_option): New function.
36911 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
36913 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36915 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
36917 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
36918 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
36919 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
36920 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
36921 for initial die_offset if dwarf_split_debug_info.
36922 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
36923 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
36925 (output_skeleton_debug_sections): Formatting fix. Use
36926 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
36927 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
36929 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
36931 * config/arm/cortex-a53.md: Add bypasses for
36932 cortex_a53_r2f_cvt.
36933 (cortex_a53_r2f): Only use for transfers.
36934 (cortex_a53_f2r): Likewise.
36935 (cortex_a53_r2f_cvt): Add reservation for conversions.
36936 (cortex_a53_f2r_cvt): Likewise.
36938 2017-01-11 Tamar Christina <tamar.christina@arm.com>
36940 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
36941 to all inlined functions, change static to extern.
36943 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
36946 * config/arm/arm.c (legitimize_pic_address): Handle reference to
36948 (arm_assemble_integer): Likewise.
36950 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
36952 * config.gcc: Use new awk script to check CPU, FPU and architecture
36953 parameters for --with-... options.
36954 * config/arm/parsecpu.awk: New file
36955 * config/arm/arm-cpus.in: New file.
36956 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
36958 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
36960 * config/arm/t-arm: Update dependency rules.
36961 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
36962 of processing .def files.
36963 * config/arm/genopt.sh: Deleted.
36964 * config/arm/gentune.sh: Deleted.
36965 * config/arm/arm-cores.def: Deleted.
36966 * config/arm/arm-arches.def: Deleted.
36967 * config/arm/arm-fpus.def: Deleted.
36968 * config/arm/arm-tune.md: Regenerated.
36969 * config/arm/arm-tables.opt: Regenerated.
36970 * config/arm/arm-cpu.h: New generated file.
36971 * config/arm/arm-cpu-data.h: New generated file.
36972 * config/arm/arm-cpu-cdata.h: New generated file.
36974 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
36977 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
36979 (input_varpool_node): Unpack dynamically_initialized bit.
36981 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
36983 PR rtl-optimization/79032
36984 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
36985 the alignment of the adjusted memory reference against that of MODE,
36986 instead of the alignment of the original memory reference.
36988 2017-01-11 Martin Jambor <mjambor@suse.cz>
36990 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
36992 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
36993 decorated functions.
36995 2017-01-11 Richard Biener <rguenther@suse.de>
36997 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
36998 set range/nonnull info for PHI results. Do not set it on
36999 stmts marked for removal.
37001 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37003 * expr.c (store_field): In the bitfield case, fetch the return value
37004 from the registers before applying a single big-endian adjustment.
37005 Always do a final load for a BLKmode value not larger than a word.
37007 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37010 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37011 that we correctly handle column numbers greater than
37012 LINE_MAP_MAX_COLUMN_NUMBER.
37014 2017-01-10 Martin Sebor <msebor@redhat.com>
37016 PR middle-end/78245
37017 * gimple-ssa-sprintf.c (get_destination_size): Call
37018 {init,fini}object_sizes.
37019 * tree-object-size.c (addr_object_size): Adjust.
37020 (pass_through_call): Adjust.
37021 (pass_object_sizes::execute): Adjust.
37022 * tree-object-size.h (fini_object_sizes): Declare.
37024 2017-01-10 Martin Sebor <msebor@redhat.com>
37026 PR tree-optimization/78775
37027 * builtins.c (get_size_range): Move...
37028 * calls.c: ...to here.
37029 (alloc_max_size): Accept zero argument.
37030 (operand_signed_p): Remove.
37031 (maybe_warn_alloc_args_overflow): Call get_size_range.
37032 * calls.h (get_size_range): Declare.
37034 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37036 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37037 from TI's devices.csv file as of September 2016.
37038 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37040 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37042 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37043 * doc/invoke.texi: Likewise.
37044 * doc/md.texi: Likewise.
37045 * doc/objc.texi: Likewise.
37047 2017-01-10 Joshua Conner <joshconner@google.com>
37049 * config/arm/fuchsia-elf.h: New file.
37050 * config/fuchsia.h: New file.
37051 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37052 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37054 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37056 2016-01-10 Richard Biener <rguenther@suse.de>
37058 PR tree-optimization/79034
37059 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
37060 Propagate out degenerate PHIs in the joiner.
37062 2017-01-10 Martin Liska <mliska@suse.cz>
37064 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
37065 (sort_congruence_classes_by_decl_uid): Likewise.
37066 (sort_congruence_class_groups_by_decl_uid): Likewise.
37067 (sem_item_optimizer::merge_classes): Sort class, groups in these
37068 classes and members in the groups by DECL_UID of declarations.
37069 This would make merge operations stable.
37071 2017-01-10 Martin Liska <mliska@suse.cz>
37073 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
37074 usage of m_classes_vec.
37075 (sem_item_optimizer::~sem_item_optimizer): Likewise.
37076 (sem_item_optimizer::get_group_by_hash): Likewise.
37077 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
37078 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
37079 (sem_item_optimizer::verify_classes): Likewise.
37080 (sem_item_optimizer::process_cong_reduction): Likewise.
37081 (sem_item_optimizer::dump_cong_classes): Likewise.
37082 (sem_item_optimizer::merge_classes): Likewise.
37083 * ipa-icf.h (congruence_class_hash): Rename from
37084 congruence_class_group_hash. Remove declaration of m_classes_vec.
37086 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
37088 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
37089 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
37090 * config.gcc: Add avx512vpopcntdqintrin.h.
37091 * config/i386/avx512vpopcntdqintrin.h: New.
37092 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
37093 * config/i386/i386-builtin-types.def: Add new types.
37094 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
37095 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
37096 __builtin_ia32_vpopcountq_v8di_mask): New.
37097 * config/i386/i386-c.c (ix86_target_macros_internal): Define
37098 __AVX512VPOPCNTDQ__.
37099 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
37100 (PTA_AVX512VPOPCNTDQ): Define.
37101 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
37102 TARGET_AVX512VPOPCNTDQ_P): Define.
37103 * config/i386/i386.opt: Add mavx512vpopcntdq.
37104 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
37105 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
37107 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37109 PR middle-end/77484
37110 * predict.def (PRED_CALL): Set to 67.
37112 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
37114 * expr.c (store_field): In the bitfield case, if the value comes from
37115 a function call and is of an aggregate type returned in registers, do
37116 not modify the field mode; extract the value in all cases if the mode
37117 is BLKmode and the size is not larger than a word.
37119 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
37122 * config/i386/cpuid.h: Fix undefined behavior.
37124 2017-01-04 Jeff Law <law@redhat.com>
37126 PR tree-optimization/79007
37127 PR tree-optimization/67955
37128 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
37129 conservative for pt.null when flag_non_call_exceptions is on.
37131 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37133 PR translation/79019
37134 PR translation/79020
37135 * params.def (PARAM_INLINE_MIN_SPEEDUP,
37136 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
37137 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
37139 * config/avr/avr.opt (maccumulate-args): Likewise.
37140 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
37141 * common.opt (freport-bug): Likewise.
37142 * cif-code.def (CIF_FINAL_ERROR): Likewise.
37143 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
37144 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
37145 translatable string.
37146 * config/i386/i386.c (function_value_32): Likewise.
37147 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
37148 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
37150 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
37151 * common/config/msp430/msp430-common.c (msp430_handle_option):
37153 * symtab.c (symtab_node::verify_base): Likewise.
37154 * opts.c (set_debug_level): Likewise.
37155 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
37156 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
37157 missing whitespace to translatable strings.
37158 * config/avr/avr.md (bswapsi2): Fix typo in comment.
37159 * config/sh/superh.h: Likewise.
37160 * config/i386/xopintrin.h: Likewise.
37161 * config/i386/znver1.md: Likewise.
37162 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
37163 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
37164 * double-int.h (struct double_int): Likewise.
37165 * double-int.c (div_and_round_double): Likewise.
37166 * wide-int.cc: Likewise.
37167 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
37168 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
37169 * cfgcleanup.c (crossjumps_occured): Renamed to ...
37170 (crossjumps_occurred): ... this.
37171 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
37174 PR tree-optimization/78899
37175 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
37176 returning bool return struct loop *, NULL for failure and the new
37178 (versionable_outer_loop_p): Don't version outer loop if it has
37179 dont_vectorized bit set.
37180 (tree_if_conversion): When versioning outer loop, ensure
37181 tree_if_conversion is performed also on the inner loop of the
37182 non-vectorizable outer loop copy.
37183 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
37184 LOOP_VECTORIZED in inner loop of the scalar outer loop and
37185 prevent vectorization of it.
37186 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
37187 the outer loop vectorization of the non-scalar version is attempted
37188 before vectorization of the inner loop in scalar version. If
37189 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
37190 vectorization of its inner loop.
37191 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
37192 has 2 inner loops, rename also on edges from bb whose single pred
37193 is outer_loop->header. Fix typo in function comment.
37195 2017-01-09 Martin Sebor <msebor@redhat.com>
37198 * asan.c (asan_emit_stack_protection): Increase local buffer size
37199 to avoid snprintf truncation warning.
37201 2017-01-09 Andrew Pinski <apinski@cavium.com>
37203 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
37204 to reference thunderx2t99 for the tuning structure
37205 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
37207 (thunderx2t99_extra_costs): This.
37208 * config/aarch64/aarch64-tune.md: Regenerate.
37209 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
37210 (vulcan_addrcost_table): This.
37211 (vulcan_regmove_cost): Rename to ...
37212 (thunderx2t99_regmove_cost): This.
37213 (vulcan_vector_cost): Rename to ...
37214 (thunderx2t99_vector_cost): this.
37215 (vulcan_branch_cost): Rename to ...
37216 (thunderx2t99_branch_cost): This.
37217 (vulcan_tunings): Rename to ...
37218 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
37219 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
37221 2017-01-09 Martin Jambor <mjambor@suse.cz>
37225 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
37226 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
37227 (propagate_vr_accross_jump_function): Use the above function for all
37228 value range computations for pass-through jump functions and type
37229 converasion from explicit value range values.
37230 (ipcp_propagate_stage): Do not attempt to deduce types of formal
37231 parameters from TYPE_ARG_TYPES.
37232 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
37233 (ipa_write_node_info): Stream type of the actual argument.
37234 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
37236 2017-01-09 Martin Liska <mliska@suse.cz>
37239 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
37240 (lookup_compiler): Do not show error message with have_E.
37242 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37244 PR tree-optimization/78938
37245 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
37246 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
37247 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
37248 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
37251 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37253 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
37256 2017-01-09 Richard Biener <rguenther@suse.de>
37258 PR tree-optimization/78997
37259 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
37260 name condition properly.
37262 2017-01-09 Richard Biener <rguenther@suse.de>
37265 * dwarf2out.c (is_cxx): New overload with context.
37266 (is_naming_typedef_decl): Use it.
37268 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37270 * invoke.texi (Option Summary): Correct spacing in option lists
37271 and add line breaks to fix over-long lines.
37273 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37275 PR middle-end/17660
37277 * extend.texi (Common Variable Attributes): Add xref to GCC
37278 Internals manual to explain mode attribute keywords.
37280 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37283 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
37284 and Preprocessor Options.
37285 (Options for Linking): Document -pthread here....
37286 (RS/6000 and PowerPC Options): ...not here.
37287 (Solaris 2 Options): ...or here.
37288 * doc/cppopts.texi: Document -pthread.
37290 2017-01-08 Martin Sebor <msebor@redhat.com>
37292 PR middle-end/77708
37293 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
37294 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
37295 New member functions.
37296 (format_directive): Used them.
37298 (pass_sprintf_length::handle_gimple_call): Same.
37299 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
37300 to avoid truncation for any argument.
37301 (extract_affine_mul): Same.
37302 * tree.c (get_file_function_name): Same.
37304 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37306 PR middle-end/77484
37307 * predict.def (PRED_INDIR_CALL): Set to 86.
37309 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37311 PR preprocessor/54124
37312 * doc/cppopts.texi: Reformat -d subtable to list the full name
37313 of the options. Add cross-reference to the docs for the general
37314 compiler -d options.
37315 * doc/invoke.texi (Developer Options): Add cross-reference to the
37316 preprocessor-specific -d option documentation.
37318 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37320 PR preprocessor/13498
37321 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
37322 redudant material, and reflect new command-line options.
37323 (System Headers): Likewise.
37325 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37327 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
37328 -isystem, and -idirafter. Copy-edit.
37329 * doc/cppopts.texi: Copy-edit. Remove contradiction about
37330 default for -ftrack-macro-expansion. Delete obsolete and
37331 badly-formatted implementation details about -fdebug-cpp output.
37332 * doc/cppwarnopts.texi: Copy-edit.
37334 2017-01-07 David Malcolm <dmalcolm@redhat.com>
37337 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37338 that the transition from a max line width >= 1<<10 to narrower
37339 lines works correctly.
37341 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
37343 * doc/options.texi (PerFunction): New.
37344 * opt-functions.awk (switch_flags): Map both Optimization and
37345 PerFunction to CL_OPTIMIZATION.
37346 * opth-gen.awk: Test for PerFunction flag along with
37348 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
37349 it only when the latter is present. Skip those that don't in
37350 the hash function generator.
37351 * common.opt (fvar-tracking): Mark as PerFunction instead of
37353 (fvar-tracking-assignments): Likewise.
37354 (fvar-tracking-assignments-toggle): Likewise.
37355 (fvar-tracking-uninit): Likewise.
37357 2017-01-07 Jakub Jelinek <jakub@redhat.com>
37359 PR translation/79018
37360 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
37363 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
37366 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
37367 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
37368 TARGET_LONG_JUMP_TABLE_OFFSETS.
37369 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
37370 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
37371 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
37372 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
37373 * config/m68k/m68k.md (tablejump expander): Likewise.
37374 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
37375 TARGET_LONG_JUMP_TABLE_OFFSETS.
37376 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
37377 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
37379 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
37380 David Holsgrove <david.holsgrove@xilinx.com>
37382 * common/config/microblaze/microblaze-common.c
37383 (TARGET_EXCEPT_UNWIND_INFO): Remove.
37384 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
37386 * config/microblaze/microblaze.c (microblaze_must_save_register)
37387 (microblaze_expand_epilogue, microblaze_return_addr): Handle
37389 (microblaze_eh_return): New function.
37390 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
37391 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
37392 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
37393 * config/microblaze/microblaze.md (eh_return): New pattern.
37395 2017-01-06 Jakub Jelinek <jakub@redhat.com>
37397 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
37398 GCC_DIAGNOSTIC_STRINGIFY): Define.
37400 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
37402 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37404 * config/arm/arm.md (<mcrr>): New.
37406 * config/arm/arm.c (arm_arch5te): New.
37407 (arm_option_override): Set arm_arch5te.
37408 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
37410 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
37411 (arm_mcrr_qualifiers): ... this. New.
37412 (MRRC_QUALIFIERS): Define to...
37413 (arm_mrrc_qualifiers): ... this. New.
37414 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
37416 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
37417 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
37418 (MRRCI, mrrc, MRRC): New.
37419 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
37420 VUNSPEC_MRRC2): New.
37422 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37424 * config/arm/arm.md (<mcr>): New.
37426 * config/arm/arm.c (arm_coproc_builtin_available): Add
37427 support for mcr, mrc, mcr2 and mrc2.
37428 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
37429 (arm_mcr_qualifiers): ... this. New.
37430 (MRC_QUALIFIERS): Define to ...
37431 (arm_mrc_qualifiers): ... this. New.
37432 (MCR_QUALIFIERS): Define to ...
37433 (arm_mcr_qualifiers): ... this. New.
37434 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
37436 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
37437 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
37438 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
37439 VUNSPEC_MRC2): New.
37441 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37443 * config/arm/arm.md (*ldc): New.
37447 * config/arm/arm.c (arm_coproc_builtin_available): Add
37448 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
37449 (arm_coproc_ldc_stc_legitimate_address): New.
37450 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
37451 'qualifier_const_pointer'.
37452 (LDC_QUALIFIERS): Define to...
37453 (arm_ldc_qualifiers): ... this. New.
37454 (STC_QUALIFIERS): Define to...
37455 (arm_stc_qualifiers): ... this. New.
37456 * config/arm/arm-protos.h
37457 (arm_coproc_ldc_stc_legitimate_address): New.
37458 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
37459 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
37460 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
37461 stc2, stcl, stc2l): New.
37462 * config/arm/constraints.md (Uz): New.
37463 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
37464 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
37465 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
37466 VUNSPEC_STC2L): New.
37468 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37470 * config/arm/arm.md (<cdp>): New.
37471 * config/arm/arm.c (neon_const_bounds): Rename this ...
37472 (arm_const_bounds): ... this.
37473 (arm_coproc_builtin_available): New.
37474 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
37475 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
37476 (CDP_QUALIFIERS): Define to...
37477 (arm_cdp_qualifiers): ... this. New.
37479 (arm_expand_builtin_args): Add case for 6 arguments.
37480 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
37481 (arm_const_bounds): ... this.
37482 (arm_coproc_builtin_available): New.
37483 * config/arm/arm_acle.h (__arm_cdp): New.
37485 * config/arm/arm_acle_builtins.def (cdp): New.
37487 * config/arm/iterators.md (CDPI,CDP,cdp): New.
37488 * config/arm/neon.md: Rename all 'neon_const_bounds' to
37489 'arm_const_bounds'.
37490 * config/arm/types.md (coproc): New.
37491 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
37492 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
37493 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
37494 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
37496 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37498 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
37499 (UBINOP_QUALIFIERS): New.
37501 (acle_builtin_data): New. Change comment.
37502 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
37503 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
37504 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
37505 arm_acle_builtins.def.
37506 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
37507 (arm_init_acle_builtins): New.
37508 (CRC32_BUILTIN): Remove.
37509 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
37510 crc32cb, crc32ch and crc32cw.
37511 (arm_init_crc32_builtins): Remove.
37512 (arm_init_builtins): Use arm_init_acle_builtins rather
37513 than arm_init_crc32_builtins.
37514 (arm_expand_acle_builtin): New.
37515 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
37516 * config/arm/arm_acle_builtins.def: New.
37518 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37520 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
37521 (arm_builtin_datum): ... this.
37522 (arm_init_neon_builtin): Rename to ...
37523 (arm_init_builtin): ... this. Add a new parameters PREFIX
37524 and USE_SIG_IN_NAME.
37525 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
37526 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
37527 'arm_builtin_datum'.
37528 (arm_init_vfp_builtins): Likewise.
37529 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
37530 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
37531 (arm_expand_neon_args): Rename to ...
37532 (arm_expand_builtin_args): ... this. Rename builtin_arg
37533 enum values and differentiate between ARG_BUILTIN_MEMORY
37534 and ARG_BUILTIN_NEON_MEMORY.
37535 (arm_expand_neon_builtin_1): Rename to ...
37536 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
37537 values, arm_expand_builtin_args and add bool parameter NEON.
37538 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
37539 (arm_expand_vfp_builtin): Likewise.
37540 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
37542 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37544 PR middle-end/77484
37545 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
37546 * predict.c (tree_estimate_probability_bb): Reverse direction of
37547 polymorphic call predictor.
37549 2017-01-06 David Malcolm <dmalcolm@redhat.com>
37551 * passes.c (execute_one_pass): Split out pass-skipping logic into...
37552 (determine_pass_name_match): ...this new function and...
37553 (should_skip_pass_p): ...this new function.
37555 2017-01-06 Nathan Sidwell <nathan@acm.org>
37557 * ipa-visibility.c (function_and_variable_visibility): Reformat
37558 comments and long lines. Remove extrneous if.
37559 * symtab.c (symtab_node::make_decl_local): Fix code format.
37560 (symtab_node::set_section_for_node): Fix comment typo.
37562 2017-01-06 Martin Liska <mliska@suse.cz>
37565 * lra-constraints.c: Rename invariant to lra_invariant.
37566 * predict.c (set_even_probabilities): Initialize e to NULL.
37568 2017-01-05 Martin Sebor <msebor@redhat.com>
37570 PR tree-optimization/78910
37571 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
37572 (format_integer): Correct off-by-one error in the handling
37573 of precision with negative numbers in signed conversions..
37575 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
37577 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
37579 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37581 PR tree-optimization/71016
37582 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
37583 factor_out_conditional_conversion. Formatting fix.
37584 (factor_out_conditional_conversion): Add cond_stmt argument.
37585 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
37586 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
37589 2017-01-05 David Malcolm <dmalcolm@redhat.com>
37591 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
37592 read-rtl-function.o, and selftest-rtl.o.
37593 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
37594 (selftest::aarch64_test_loading_full_dump): New function.
37595 (selftest::aarch64_run_selftests): New function.
37596 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
37597 selftest::aarch64_run_selftests.
37598 * config/i386/i386.c
37599 (selftest::ix86_test_loading_dump_fragment_1): New function.
37600 (selftest::ix86_test_loading_call_insn): New function.
37601 (selftest::ix86_test_loading_full_dump): New function.
37602 (selftest::ix86_test_loading_unspec): New function.
37603 (selftest::ix86_run_selftests): Call the new functions.
37604 * emit-rtl.c (maybe_set_max_label_num): New function.
37605 * emit-rtl.h (maybe_set_max_label_num): New decl.
37606 * function.c (instantiate_decls): Guard call to
37607 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
37608 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
37610 * gensupport.c (gen_reader::gen_reader): Pass "false"
37611 for new "compact" param of rtx_reader.
37612 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
37613 rather than an empty string for NULL strings.
37614 * read-md.c: Potentially include config.h rather than bconfig.h.
37615 Wrap include of errors.h with #ifdef GENERATOR_FILE.
37616 (have_error): New global, copied from errors.c.
37617 (md_reader::read_name): Rename to...
37618 (md_reader::read_name_1): ...this, adding "out_loc" param,
37619 and converting "missing name or number" to returning false, rather
37621 (md_reader::read_name): Reimplement in terms of read_name_1.
37622 (md_reader::read_name_or_nil): New function.
37623 (md_reader::read_string): Handle "(nil)" by returning NULL.
37624 (md_reader::md_reader): Add new param "compact".
37625 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
37626 (md_reader::read_file): New method.
37627 * read-md.h (md_reader::md_reader): Add new param "compact".
37628 (md_reader::read_file): New method.
37629 (md_reader::is_compact): New accessor.
37630 (md_reader::read_name): Convert return type from void to file_location.
37631 (md_reader::read_name_or_nil): New decl.
37632 (md_reader::read_name_1): New decl.
37633 (md_reader::m_compact): New field.
37634 (noop_reader::noop_reader): Pass "false" for new "compact" param
37636 (rtx_reader::rtx_reader): Add new "compact" param.
37637 (rtx_reader::read_rtx_operand): Make virtual and convert return
37638 type from void to rtx.
37639 (rtx_reader::read_until): New decl.
37640 (rtx_reader::handle_any_trailing_information): New virtual function.
37641 (rtx_reader::postprocess): New virtual function.
37642 (rtx_reader::finalize_string): New virtual function.
37643 (rtx_reader::m_in_call_function_usage): New field.
37644 (rtx_reader::m_reuse_rtx_by_id): New field.
37645 * read-rtl-function.c: New file.
37646 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
37647 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
37648 (selftest::verify_three_block_rtl_cfg): New decl.
37649 * read-rtl-function.h: New file.
37650 * read-rtl.c: Potentially include config.h rather than bconfig.h.
37651 For host, include function.h, memmodel.h, and emit-rtl.h.
37652 (one_time_initialization): New function.
37653 (struct compact_insn_name): New struct.
37654 (compact_insn_names): New array.
37655 (find_code): Handle insn codes in compact dumps.
37656 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
37657 (bind_subst_iter_and_attr): Likewise.
37658 (add_condition_to_string): Likewise.
37659 (add_condition_to_rtx): Likewise.
37660 (apply_attribute_uses): Likewise.
37661 (add_current_iterators): Likewise.
37662 (apply_iterators): Likewise.
37663 (initialize_iterators): Guard usage of apply_subst_iterator with
37664 #ifdef GENERATOR_FILE.
37665 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
37666 (md_reader::read_mapping): Likewise.
37667 (add_define_attr_for_define_subst): Likewise.
37668 (add_define_subst_attr): Likewise.
37669 (read_subst_mapping): Likewise.
37670 (check_code_iterator): Likewise.
37671 (rtx_reader::read_rtx): Likewise. Move one-time initialization
37673 (one_time_initialization): New function.
37674 (rtx_reader::read_until): New method.
37675 (read_flags): New function.
37676 (parse_reg_note_name): New function.
37677 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
37678 Handle reuse_rtx ids.
37679 Wrap iterator lookup within #ifdef GENERATOR_FILE.
37680 Add parsing support for RTL dumps, mirroring the special-cases in
37681 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
37682 values, and calling handle_any_trailing_information.
37683 (rtx_reader::read_rtx_operand): Convert return type from void
37684 to rtx, returning return_rtx. Handle case 'e'. Call
37685 finalize_string on XSTR and XTMPL fields.
37686 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
37687 "(nil)" values were omitted. Call the postprocess vfunc on the
37689 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
37690 class ctor. Initialize m_in_call_function_usage. Call
37691 one_time_initialization.
37692 * rtl-tests.c (selftest::test_uncond_jump): Call
37693 set_new_first_and_last_insn.
37694 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
37695 * selftest-rtl.c: New file.
37696 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
37697 (selftest::get_insn_by_uid): New decl.
37698 * selftest-run-tests.c (selftest::run_tests): Call
37699 read_rtl_function_c_tests.
37700 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
37701 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
37704 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
37706 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
37707 operands in a special way. Assert that pos+len <= mode precision.
37709 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37711 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
37712 3 argument Alias with unlimited for the negative form.
37713 (fno-vect-cost-model): Removed.
37715 2017-01-05 Martin Liska <mliska@suse.cz>
37717 * hsa-gen.c (gen_hsa_divmod): New function.
37718 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
37720 2017-01-05 Martin Liska <mliska@suse.cz>
37723 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
37726 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37728 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
37729 small constant length operands.
37731 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37733 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
37734 between loop iterations.
37736 2017-01-05 Martin Liska <mliska@suse.cz>
37739 * gimplify.c (gimplify_decl_expr): Compare to
37740 asan_poisoned_variables instread of checking flags.
37741 (gimplify_target_expr): Likewise.
37742 (gimplify_expr): Likewise.
37743 (gimplify_function_tree): Conditionally initialize
37744 asan_poisoned_variables.
37746 2017-01-04 Jeff Law <law@redhat.com>
37748 PR tree-optimizatin/78812
37749 * rtl.h (contains_mem_rtx_p): Prototype.
37750 * ifcvt.c (containts_mem_rtx_p): Move from here to...
37751 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
37752 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
37753 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
37754 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
37756 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37758 * input.c (assert_char_at_range): Default-initialize actual_range.
37760 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37762 * df-scan.c (df_ref_create_structure): Make regno unsigned,
37763 to match the caller.
37765 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37767 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
37768 insns after final jump in test to emit dummy move.
37770 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37772 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
37773 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
37775 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37777 * multiple_target.c (create_dispatcher_calls): Init e_next.
37778 * tree-ssa-loop-split.c (split_loop): Init border.
37779 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
37782 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
37787 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
37788 (altivec_register_operand): Do not return true if the operand
37789 contains a SUBREG mixing SImode and SFmode.
37790 (vsx_register_operand): Likewise.
37791 (vsx_reg_sfsubreg_ok): New predicate.
37792 (vfloat_operand): Do not return true if the operand contains a
37793 SUBREG mixing SImode and SFmode.
37794 (vint_operand): Likewise.
37795 (vlogical_operand): Likewise.
37796 (gpc_reg_operand): Likewise.
37797 (int_reg_operand): Likewise.
37798 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
37799 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
37800 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
37802 (rs6000_emit_move_si_sf_subreg): New helper function.
37803 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
37804 fixup SUBREGs involving SImode and SFmode.
37805 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
37806 numbers for the new peephole2 optimization.
37807 (peephole2 for SFmode unions): New peephole2 to optimize cases in
37808 the GLIBC math library that do AND/IOR/XOR operations on single
37809 precision floating point.
37810 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
37811 target macros to say whether we need to avoid SUBREGs mixing
37813 (TARGET_ALLOW_SF_SUBREG): Likewise.
37814 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
37815 (UNSPEC_SI_FROM_SF): Likewise.
37816 (iorxor): Change spacing.
37817 (and_ior_xor): New iterator for AND, IOR, and XOR.
37818 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
37819 (movdi_from_sf_zero_ext): Likewise.
37820 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
37821 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
37822 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
37823 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
37824 (fms<mode>4): Likewise.
37825 (fnma<mode>4): Likewise.
37826 (fnms<mode>4): Likewise.
37827 (nfma<mode>4): Likewise.
37828 (nfms<mode>4): Likewise.
37830 2017-01-04 Marek Polacek <polacek@redhat.com>
37833 * doc/invoke.texi: Document -Wpointer-compare.
37835 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37837 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
37840 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
37841 descriptions for -gdwarf-5 and emit them as uleb128 instead of
37844 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
37847 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
37848 documentation of the powerpc_popcntb_ok attribute.
37849 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
37850 code to issue warning messages if a requested CPU configuration is
37851 not supported by the binary (assembler and loader) toolchain.
37852 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
37853 made to define a built-in function that has been disabled.
37854 (paired_init_builtins): Add assertion to prevent ICE if attempt is
37855 made to define a built-in function that has been disabled.
37856 (altivec_init_builtins): Add comment explaining why definition
37857 of the DST built-in functions is not preceded by an assertion
37858 check. Add assertions to prevent ICE if attempts are made to
37859 define an altivec predicate or an abs* built-in function that has
37861 (htm_init_builtins): Add comment explaining why definition of the
37862 htm built-in functions is not preceded by an assertion check.
37864 2017-01-04 Jeff Law <law@redhat.com>
37866 PR tree-optimizatin/67955
37867 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
37868 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
37869 the points-to solution does not include pt_null. Use DECL_PT_UID
37872 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
37874 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
37875 Use gen_int_mode instead of gen_lopwart for const_int operands.
37877 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37879 PR tree-optimization/71563
37880 * match.pd: Simplify X << Y into X if Y is known to be 0 or
37881 out of range value - has low bits known to be zero.
37883 2017-01-04 Alan Modra <amodra@gmail.com>
37885 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
37886 * configure: Regenerate.
37887 * config.in: Regenerate.
37889 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37892 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
37893 a substring of the message, but strcmp with the whole message. Ifdef
37894 ENABLE_NLS, translate the message first using dgettext.
37896 2017-01-03 Jeff Law <law@redhat.com>
37898 PR tree-optimizatin/78856
37899 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
37900 (mark_threaded_blocks): Remove code to truncate thread paths that
37901 cross multiple loop headers. Instead invalidate the cached loop
37902 iteration information and handle case of a thread path walking
37903 into an irreducible region.
37905 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
37908 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
37909 assertions. Add support for doing the signbit if the IEEE 128-bit
37910 floating point value is in a GPR.
37911 * config/rs6000/rs6000.md (Fsignbit): Delete.
37912 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
37913 Update the length attribute if the value is in a GPR.
37914 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
37915 the sign or zero extension instruction, since the value is always 0/1.
37916 (signbit<mode>2_dm2): Delete using <Fsignbit>.
37919 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
37920 extracting SImode to a GPR register so that we can generate a
37921 store, limit the vector to be in a traditional Altivec register
37922 for the vextuwrx instruction.
37924 2017-01-03 Ian Lance Taylor <iant@google.com>
37926 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
37928 2017-01-03 Martin Sebor <msebor@redhat.com>
37930 PR tree-optimization/78696
37931 * gimple-ssa-sprintf.c (format_floating): Correct handling of
37932 precision. Use MPFR for %f for greater fidelity. Correct handling
37934 (pass_sprintf_length::compute_format_length): Set width and precision
37935 specified by asrerisk to void_node for vararg functions.
37936 (try_substitute_return_value): Adjust dump output.
37938 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
37940 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
37942 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
37944 * doc/invoke.texi (SPARC options): Document -mlra as the default.
37945 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
37946 -mlra/-mno-lra was passed to the compiler.
37948 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
37950 PR rtl-optimization/65618
37951 * emit-rtl.c (try_split): Move initialization of "before" and
37952 "after" to just before the call to emit_insn_after_setloc.
37954 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
37956 * doc/md.texi (Standard Names): Remove reference to Java frontend.
37958 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
37960 * dwarf2out.c (gen_enumeration_type_die): When
37961 -gno-strict-dwarf, add a DW_AT_encoding attribute.
37963 2017-01-03 Jakub Jelinek <jakub@redhat.com>
37965 PR tree-optimization/78965
37966 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
37967 Change first argument from const call_info & to call_info &. For %n
37968 set info.nowrite to false.
37970 PR middle-end/78901
37971 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
37972 possibly throwing calls.
37974 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
37975 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
37976 and fns handling, rather than in a separate case SSA_NAME.
37978 2017-01-02 Jeff Law <law@redhat.com>
37980 * config/darwin-driver.c (darwin_driver_init): Const-correctness
37981 fixes for first_period and second_period variables.
37983 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
37986 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
37987 (*insvqi_1): New insn pattern.
37988 (*insvqi_1_mem_rex64): Ditto.
37989 (*insvqi_2): Ditto.
37990 (*insvqi_3): Rename from *insvqi.
37992 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
37994 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
37996 * doc/cfg.texi (Edges): Remove reference to Java.
37997 (Maintaining the CFG): Ditto.
37999 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38001 PR middle-end/77674
38002 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38003 transparent aliases.
38005 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38007 PR middle-end/77484
38008 * predict.def (PRED_CALL): Update hitrate.
38009 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38010 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38011 into direct/indirect/polymorphic variants.
38013 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38015 Update copyright years.
38017 * gcc.c (process_command): Update copyright notice dates.
38018 * gcov-dump.c (print_version): Ditto.
38019 * gcov.c (print_version): Ditto.
38020 * gcov-tool.c (print_version): Ditto.
38021 * gengtype.c (create_file): Ditto.
38022 * doc/cpp.texi: Bump @copying's copyright year.
38023 * doc/cppinternals.texi: Ditto.
38024 * doc/gcc.texi: Ditto.
38025 * doc/gccint.texi: Ditto.
38026 * doc/gcov.texi: Ditto.
38027 * doc/install.texi: Ditto.
38028 * doc/invoke.texi: Ditto.
38030 Copyright (C) 2017 Free Software Foundation, Inc.
38032 Copying and distribution of this file, with or without modification,
38033 are permitted in any medium without royalty provided the copyright
38034 notice and this notice are preserved.