1 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3 * tree.c (build_vector): Delete.
4 * tree.h (build_vector): Make static and move into the self-testing
7 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
9 * vector-builder.h (vector_builder::derived): New const overload.
10 (vector_builder::elt): New function.
11 * tree-vector-builder.h (tree_vector_builder::type): New function.
12 (tree_vector_builder::apply_step): Declare.
13 * tree-vector-builder.c (tree_vector_builder::apply_step): New
15 * gimple-fold.h (tree_vector_builder): Declare.
16 (gimple_build_vector): Take a tree_vector_builder instead of a
17 type and vector of elements.
18 * gimple-fold.c (gimple_build_vector): Likewise.
19 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
21 (get_initial_defs_for_reduction): Likewise.
22 (vectorizable_induction): Likewise.
24 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
26 * tree-vector-builder.h
27 (tree_vector_builder::new_binary_operation): Declare.
28 * tree-vector-builder.c
29 (tree_vector_builder::new_binary_operation): New function.
30 * fold-const.c (fold_relational_const): Use it.
31 (const_binop): Likewise. Check that both input vectors have
32 the same number of elements, thus excluding things like WIDEN_SUM.
33 Check whether it is possible to operate directly on the encodings
36 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
38 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
39 new_unary_operation, operating only on the encoded elements.
40 (const_unop): Likewise.
41 (exact_inverse): Likewise.
42 (distributes_over_addition_p): New function.
43 (const_binop): Use tree_vector_builder and new_unary_operation
44 for combinations of VECTOR_CST and INTEGER_CST. Operate only
45 on the encoded elements unless the encoding is strided and the
46 operation does not distribute over addition.
47 (fold_convert_const): Use tree_vector_builder and
48 new_unary_operation. Operate only on the encoded elements
49 for truncating integer conversions, or for non-stepped encodings.
51 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
53 * config/sparc/sparc.c: Include tree-vector-builder.h.
54 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
55 * expmed.c: Include tree-vector-builder.h.
56 (make_tree): Use tree_vector_builder instead of build_vector.
57 * fold-const.c: Include tree-vector-builder.h.
58 (const_binop): Use tree_vector_builder instead of build_vector.
59 (const_unop): Likewise.
60 (native_interpret_vector): Likewise.
61 (fold_vec_perm): Likewise.
62 (fold_ternary_loc): Likewise.
63 * gimple-fold.c: Include tree-vector-builder.h.
64 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
66 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
67 (simplify_vector_constructor): Use tree_vector_builder instead
69 * tree-vect-generic.c: Include tree-vector-builder.h.
70 (add_rshift): Use tree_vector_builder instead of build_vector.
71 (expand_vector_divmod): Likewise.
72 (optimize_vector_constructor): Likewise.
73 * tree-vect-loop.c: Include tree-vector-builder.h.
74 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
75 of build_vector. Explicitly use a stepped encoding for
77 * tree-vect-slp.c: Include tree-vector-builder.h.
78 (vect_get_constant_vectors): Use tree_vector_builder instead
80 (vect_transform_slp_perm_load): Likewise.
81 (vect_schedule_slp_instance): Likewise.
82 * tree-vect-stmts.c: Include tree-vector-builder.h.
83 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
84 (vect_gen_perm_mask_any): Likewise.
85 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
86 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
88 (build_vector_from_val): Likewise. Explicitly use a duplicate
91 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
93 * doc/generic.texi (VECTOR_CST): Describe new representation of
95 * vector-builder.h: New file.
96 * tree-vector-builder.h: Likewise.
97 * tree-vector-builder.c: Likewise.
98 * Makefile.in (OBJS): Add tree-vector-builder.o.
99 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
100 * tree-core.h (tree_base): Add a vector_cst field to the u union.
101 (tree_vector): Change the number of elements to
102 vector_cst_encoded_nelts.
103 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
104 (VECTOR_CST_ELTS): Delete.
105 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
106 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
107 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
108 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
109 (VECTOR_CST_ENCODED_ELT): Likewise.
110 (vector_cst_encoded_nelts): New function.
111 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
112 VECTOR_CST_NELTS_PER_PATTERN as arguments.
113 (vector_cst_int_elt, vector_cst_elt): Declare.
114 * tree.c: Include tree-vector-builder.h.
115 (tree_code_size): Abort if passed VECTOR_CST.
116 (tree_size): Update for new VECTOR_CST layout.
117 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
118 VECTOR_CST_NELTS_PER_PATTERN as arguments.
119 (build_vector): Use tree_vector_builder.
120 (vector_cst_int_elt, vector_cst_elt): New functions.
121 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
122 encoded elements and then create the vector in the canonical form.
123 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
124 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
125 (tree_c_tests): Call test_vector_cst_patterns.
126 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
128 (hash_tree): Likewise.
129 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
130 (streamer_write_tree_header): Likewise.
131 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
132 (streamer_alloc_tree): Likewise. Update call to make_vector.
133 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
135 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
137 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
138 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
139 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
141 2017-12-07 Bin Cheng <bin.cheng@arm.com>
142 Richard Biener <rguenther@suse.de>
144 PR tree-optimization/81303
145 * Makefile.in (gimple-loop-interchange.o): New object file.
146 * common.opt (floop-interchange): Reuse the option from graphite.
147 * doc/invoke.texi (-floop-interchange): Ditto. New document for
148 -floop-interchange and mention it for -O3.
149 * opts.c (default_options_table): Enable -floop-interchange at -O3.
150 * gimple-loop-interchange.cc: New file.
151 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
152 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
153 * passes.def (pass_linterchange): New pass.
154 * timevar.def (TV_LINTERCHANGE): New time var.
155 * tree-pass.h (make_pass_linterchange): New declaration.
156 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
157 interchange. Record IV before/after increment in new parameters.
158 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
159 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
161 (check_reduction_path): ...New function here.
162 * tree-vectorizer.h (check_reduction_path): New declaration.
164 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
167 PR rtl-optimization/80818
168 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
169 always early clobbered.
170 * lra-lives.c (process_bb_lives): Check input hard regs for early
171 clobbered non-operand hard reg.
173 2017-12-07 Jakub Jelinek <jakub@redhat.com>
176 * tree-cfg.c (verify_gimple_assign_binary): Don't require
177 types_compatible_p, just that TYPE_MODE is the same.
179 2017-12-07 Martin Sebor <msebor@redhat.com>
182 * attribs.c (empty_attribute_table): Initialize new member of
183 struct attribute_spec.
184 (decl_attributes): Add argument. Handle mutually exclusive
185 combinations of attributes.
186 (selftests::test_attribute_exclusions): New function.
187 (selftests::attribute_c_tests): Ditto.
188 * attribs.h (decl_attributes): Add default argument.
189 * selftest.h (attribute_c_tests): Declare.
190 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
191 * tree-core.h (attribute_spec::exclusions, exclude): New type and
193 * doc/extend.texi (Common Function Attributes): Update const and pure.
194 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
195 of struct attribute_spec.
196 * config/arc/arc.c (arc_attribute_table): Same.
197 * config/arm/arm.c (arm_attribute_table): Same.
198 * config/avr/avr.c ( avr_attribute_table): Same.
199 * config/bfin/bfin.c (bfin_attribute_table): Same.
200 * config/cr16/cr16.c (cr16_attribute_table): Same.
201 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
202 * config/h8300/h8300.c (h8300_attribute_table): Same.
203 * config/i386/i386.c (ix86_attribute_table): Same.
204 * config/ia64/ia64.c (ia64_attribute_table): Same.
205 * config/m32c/m32c.c (m32c_attribute_table): Same.
206 * config/m32r/m32r.c (m32r_attribute_table): Same.
207 * config/m68k/m68k.c (m68k_attribute_table): Same.
208 * config/mcore/mcore.c (mcore_attribute_table): Same.
209 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
210 * config/mips/mips.c (mips_attribute_table): Same.
211 * config/msp430/msp430.c (msp430_attribute_table): Same.
212 * config/nds32/nds32.c (nds32_attribute_table): Same.
213 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
214 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
215 * config/rl78/rl78.c (rl78__attribute_table): Same.
216 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
217 * onfig/rx/rx.c (rx_attribute_table): Same.
218 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
219 * config/sh/sh.c (sh_attribute_table): Same.
220 * config/sparc/sparc.c (sparc_attribute_table): Same.
221 * config/spu/spu.c (spu_attribute_table): Same.
222 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
223 * config/v850/v850.c (v850_attribute_table): Same.
224 * config/visium/visium.c (visium_attribute_table): Same.
226 2017-12-07 Tamar Christina <tamar.christina@arm.com>
229 * config/arm/arm.c (INCLUDE_STRING): Define.
230 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
231 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
234 2017-12-07 Michael Matz <matz@suse.de>
236 Add unroll and jam pass
238 * gimple-loop-jam.c: New file.
239 * Makefile.in (OBJS): Add gimple-loop-jam.o.
240 * common.opt (funroll-and-jam): New option.
241 * opts.c (default_options_table): Add unroll-and-jam at -O3.
242 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
243 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
244 * passes.def: Add pass_loop_jam.
245 * timevar.def (TV_LOOP_JAM): Add.
246 * tree-pass.h (make_pass_loop_jam): Declare.
247 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
248 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
249 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
250 to flow_loop_tree_node_add.
251 (duplicate_subloops, copy_loops_to): Append to sibling list.
252 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
253 * doc/invoke.texi (-funroll-and-jam): Document new option.
254 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
256 2017-12-07 Richard Biener <rguenther@suse.de>
258 PR tree-optimization/83296
259 PR tree-optimization/67769
260 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
261 flow sensitive info in an unrelated BB.
262 (value_replacement): Use reset_flow_sensitive_info.
263 (minmax_replacement): Reset flow sensitive info on the def
264 we move. Do not reset flow sensitive info in the whole BB
266 (abs_replacement): Likewise.
268 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
271 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
272 rs6000_cpu to the given -mcpu=, or to the default processor.
274 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
276 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
277 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
278 instead of rs6000_cpu_attr.
279 (rs6000_adjust_cost): Ditto.
280 (is_microcoded_insn): Ditto.
281 (rs6000_adjust_priority): Ditto.
282 (rs6000_issue_rate): Ditto.
283 (rs6000_use_sched_lookahead): Ditto.
284 (rs6000_use_sched_lookahead_guard): Ditto.
285 (rs6000_sched_reorder): Ditto.
286 (force_new_group): Ditto.
287 * config/rs6000/rs6000.md (cpu attribute): Ditto.
288 (group_ending_nop): Ditto.
290 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
292 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
293 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
294 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
295 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
297 (rs6000_reassociation_width): Ditto.
298 (rs6000_emit_epilogue): Ditto.
299 (rs6000_adjust_cost): Ditto.
300 (is_microcoded_insn): Ditto.
301 (is_cracked_insn): Ditto.
302 (rs6000_adjust_priority): Ditto.
303 (rs6000_sched_reorder): Ditto.
304 (rs6000_sched_reorder2): Ditto.
305 (insn_must_be_first_in_group): Ditto.
306 (insn_must_be_last_in_group): Ditto.
307 (rs6000_register_move_cost): Ditto.
308 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
311 2017-12-07 Julia Koval <julia.koval@intel.com>
313 * config.gcc: Add vaesintrin.h.
314 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI): New type.
315 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
316 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi): New builtins.
317 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
318 * config/i386/immintrin.h: Include vaesintrin.h.
319 * config/i386/sse.md (vaesdec_<mode>): New pattern.
320 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
321 _mm_aesdec_epi128): New intrinsics.
323 2017-12-06 David Malcolm <dmalcolm@redhat.com>
325 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
326 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
327 c-family/c-spellcheck.cc.
328 (best_macro_match::best_macro_match): Likewise.
330 (struct edit_distance_traits<cpp_hashnode *>): Move to
331 c-family/c-spellcheck.h.
332 (class best_macro_match): Likewise.
334 2017-12-06 Jakub Jelinek <jakub@redhat.com>
336 PR tree-optimization/83293
337 * gimple-ssa-strength-reduction.c (insert_initializers): Use
338 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
339 might insert into empty bb.
342 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
343 simplify for plus with :c added, and pointer_plus without that.
344 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
345 with undefined overflow and the conversion is not widening,
346 perform negation in utype and only convert to type afterwards.
347 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
348 simplify for plus with :c added, and pointer_plus without that.
349 If type is integral with undefined overflow and the conversion is
350 not widening, perform minus in utype and only convert to type
351 afterwards. Move the last pointer_diff_expr simplify into the
354 2017-12-06 Martin Sebor <msebor@redhat.com>
356 PR tree-optimization/82646
357 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
360 2017-12-06 Martin Sebor <msebor@redhat.com>
362 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
365 PR tree-optimization/83075
366 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
367 strncat/strncpy don't change length of source string.
369 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
372 2017-11-29 Martin Aberg <maberg@gaisler.com>
374 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
375 to prevent b2bst errata sequence.
376 (sqrtdf2_fix): Likewise.
378 2017-12-06 Jakub Jelinek <jakub@redhat.com>
380 PR tree-optimization/81945
381 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
382 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
383 to dest_cfun has orig_loop_num set, either remap it to the new
384 loop number if the loop got moved too, or clear it.
386 2017-12-05 Steve Ellcey <sellcey@cavium.com>
388 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
390 (thunderx2t99_nothing): New insn reservation.
391 (thunderx2t99_mrs): New insn reservation.
392 (thunderx2t99_multiple): New insn reservation.
393 (thunderx2t99_alu_basi): Add bfx to reservation.
394 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
396 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
399 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
401 2017-12-05 Bin Cheng <bin.cheng@arm.com>
403 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
404 tree-ssa-pre.c::remove_dead_inserted_code.
405 * tree-ssa-dce.h: New file.
406 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
407 (remove_dead_inserted_code): Move and rename to function
408 tree-ssa-dce.c::simple_dce_from_worklist.
409 (pass_pre::execute): Update use.
411 2017-12-05 Richard Biener <rguenther@suse.de>
413 PR tree-optimization/83277
414 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
415 to code-gen liveout vars.
417 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
419 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
420 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
421 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
424 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
426 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
428 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
429 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
432 2017-12-05 Richard Biener <rguenther@suse.de>
434 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL):
436 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
437 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
438 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
440 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
442 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
444 2017-12-05 Martin Liska <mliska@suse.cz>
445 Jakub Jelinek <jakub@redhat.com>
447 * doc/invoke.texi: Document the options.
448 * flag-types.h (enum sanitize_code): Add
449 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
450 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
451 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
452 * opts.c: Define new sanitizer options.
453 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
454 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
456 2017-12-05 Julia Koval <julia.koval@intel.com>
458 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
459 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
460 (ix86_handle_option): Handle -mavx512vnni.
461 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
462 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
463 * config/i386/i386-c (__AVX512VNNI__): New.
464 * config/i386/i386.c (ix86_target_string): Handle new option.
465 (ix86_valid_target_attribute_inner_p): Handle new option.
466 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
467 * config/i386/i386.opt (mavx512vnni): New option.
469 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
472 * athlon.md: Disable for generic.
473 * haswell.md: Enable for generic.
474 * i386.c (ix86_sched_init_global): Add core hooks for generic.
475 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
477 (ix86_adjust_cost): Move generic to haswell path.
479 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
481 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
482 instead of MEM_P in a couple more places. Fix formatting issues.
484 2017-12-04 Jim Wilson <jimw@sifive.com>
486 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
487 instead of GP_REG_LAST-1.
488 (riscv_adjust_libcall_cfi_prologue): Likewise.
489 (riscv_adjust_libcall_cri_epilogue): Likewise.
490 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
493 2017-12-04 Luis Machado <luis.machado@linaro.org>
495 * ipa-pure-const.c (check_decl): Add missing newline.
496 (state_from_flags): Likewise.
498 2017-12-04 Jeff Law <law@redhat.com>
500 PR tree-optimizatin/78496
501 * gimple-ssa-evrp-analyze.h
502 (evrp_range_analyzer::get_vr_values): Simplify.
503 * gimple-ssa-evrp-analyze.c: Corresponding changes.
504 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
505 and gimple-ssa-evrp-analyze.h.
506 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
507 (simplify_stmt_for_jump_threading): Copy a blob of code from
508 tree-vrp.c to use ranges to simplify statements.
509 (dom_opt_dom_walker::before_dom_children): Call
510 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
511 (dom_opt_dom_walker::after_dom_children): Similarly for
512 evrp_range_analyzer::leave.
513 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
516 * gimple-ssa-evrp-analyze.c
517 (evrp_range_analyzer::extract_range_from_stmt): Always use
518 vr_values::update_value_range so preexisting range info is
519 medged with new range info, even if the new range is VR_VARYING.
521 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
523 * combine.c: Adjust comment.
524 (use_crosses_set_p): Delete.
525 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
526 (try_combine): Ditto.
528 2017-12-04 Richard Biener <rguenther@suse.de>
530 PR tree-optimization/83255
531 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
532 Re-add zero-iteration check.
534 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
536 PR rtl-optimization/83245
537 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
538 hard registers as earlyclobber, also if not in an asm.
540 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
544 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
547 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
548 rs6000_cpu based on cpu_index, not tune_index.
550 2017-12-04 Richard Biener <rguenther@suse.de>
552 PR tree-optimization/83238
553 * graphite-scop-detection.c (scop_detection::merge_sese): Make
554 code match comment, rejecting invalid SESE regions.
556 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
558 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
559 require base operand is a REG_POINTER prior to reload on targets
560 with non-equivalent space registers.
562 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
564 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
565 (update_specialized_profile): Fix updating of counts.
566 (perhaps_add_new_callers): Likewise.
568 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
571 * x86-tune.def: Remove obsolette FIXMEs.
572 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
573 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
574 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
576 (X86_TUNE_PAD_RETURNS): Disable for generic.
578 2017-12-02 Jakub Jelinek <jakub@redhat.com>
580 PR tree-optimization/83170
581 PR tree-optimization/83241
582 * gimple-ssa-store-merging.c
583 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
584 gimple_vuse (ins_stmt) in case it has changed.
585 (imm_store_chain_info::output_merged_store): Likewise.
587 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
591 * tree-cfg.c (pass_warn_function_return::execute): Handle
592 __builtin_ubsan_handle_missing_return like __builtin_unreachable
593 with BUILTINS_LOCATION.
597 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
598 is BLKmode for vector field with vector raw mode, use TYPE_MODE
599 instead of DECL_MODE.
601 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
602 last argument to rtx pointer.
603 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
604 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
605 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
606 depending on the chosen ISAs.
607 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
608 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
609 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
611 * config/i386/sse.md (mov<mode>_internal): Likewise.
612 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
614 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
616 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
618 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
620 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
621 parameters from prototype.
622 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
623 parameters. Don't print a comment.
624 (emit_fusion_gpr_load): Adjust.
625 (emit_fusion_load_store): Adjust.
626 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
627 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
628 comment on the second line.
630 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
633 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
634 rs6000_cpu based on cpu_index, not tune_index.
636 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
638 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
639 which_alternative instead of which_alternative + 1.
640 (output_asm_insn): Print an extra tab if the template is short.
642 2017-12-01 Jim Wilson <jimw@sifive.com>
644 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
646 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
647 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
649 * doc/tm.texi: Regenerate.
651 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
654 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
655 whether we can allocate pseudos before trying to fix an address.
656 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
657 memory address is indexed or indirect.
658 (floatuns_<mode>si2_hw2): Likewise.
660 2017-12-01 Jason Merrill <jason@redhat.com>
662 * Makefile.in (TAGS): Add c-family/*.cc.
664 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
666 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
667 (call_value_insn): Likewise.
668 (sibcall_insn): Likewise.
669 (sibcall_value_insn): Likewise.
670 (movsi_aarch64): Likewise.
671 (movdi_aarch64): Likewise.
672 (add_losym_): Likewise.
673 (ldr_got_small_): Likewise.
674 (ldr_got_small_sidi): Likewise.
675 (ldr_got_small_28k_): Likewise.
676 (ldr_got_small_28k_sidi): Likewise.
677 * config/aarch64/aarch64.c (aarch64_print_address_internal):
678 Move output_addr_const to symbolic case. Add error check.
680 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
682 * config/s390/predicates.md (plus16_Q_operand): New predicate.
683 * config/s390/s390.md: Disable MVC merging peephole if it would
684 disable operand forwarding.
685 (new peephole2): Split MVCs if it would turn them into up to 2
688 2017-12-01 Richard Biener <rguenther@suse.de>
690 PR tree-optimization/83232
691 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
692 detection of same access. Instead of breaking the group here
693 do not consider the duplicate. Add comment explaining real fix.
695 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
697 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
699 2017-12-01 Sudakshina Das <sudi.das@arm.com>
701 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
703 2017-12-01 Jakub Jelinek <jakub@redhat.com>
705 * function.h (struct function): Remove cilk_frame_decl,
706 is_cilk_function and calls_cilk_spawn fields.
707 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
709 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
710 cilk_elemental field.
711 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
712 * target.def: Adjust comment.
713 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
714 Don't test cilk_elemental.
716 PR tree-optimization/83233
717 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
718 bswap_stat name for the struct.
721 * tree.h (SWITCH_BREAK_LABEL_P): Define.
722 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
723 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
724 SWITCH_BREAK_LABEL_P set on the label.
725 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
726 added for default case if it was missing and not all cases covered.
727 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
728 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
730 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
731 to avoid -Wimplicit-fallthrough warning.
732 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
733 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
735 PR tree-optimization/83221
736 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
738 (init_reassoc): Formatting fix.
741 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
742 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
746 * tree-cfg.c: Include asan.h.
747 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
749 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
751 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
753 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
755 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
756 vpcmpeqd instruction.
758 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
760 * config/i386/i386.c (standard_sse_constant_opcode): Fix
761 registers type for 128bit mode.
763 2017-11-30 Jakub Jelinek <jakub@redhat.com>
765 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
767 * gengtype-state.c (read_state_pair): Likewise.
768 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
769 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
770 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
771 * ifcvt.c (noce_try_store_flag_constants): Likewise.
772 * tree-ssa-ccp.c (ccp_finalize): Likewise.
773 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
774 * builtins.c (fold_builtin_3): Likewise.
775 * graphite-scop-detection.c
776 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
777 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
780 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
783 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
784 parameter. Return true if ignore_decl_rtl_set_p is true and other
785 conditions are satisfied.
786 * asan.h (asan_protect_global): Add new parameter.
787 * varasm.c (categorize_decl_for_section): Pass true as second parameter
788 to asan_protect_global calls.
790 2017-11-30 Jim Wilson <jimw@sifive.com>
792 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
793 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
794 -mstrict-align, add info on default value. Delete redundant lines for
795 -mabi. Add missing -mexplicit-relocs docs.
797 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
799 * config/arc/arc.md (trap): New pattern.
801 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
803 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
804 instruction to end into a delay slot.
805 * config/arc/arc.md (cond_delay_insn): Check if the instruction
806 can be placed into a delay slot against reg_note.
808 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
810 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
811 labels number of usages.
813 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
815 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
817 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
819 2017-11-30 Jakub Jelinek <jakub@redhat.com>
822 * internal-fn.c (expand_mul_overflow): Optimize unsigned
823 multiplication by power of 2 constant into two shifts + comparison.
825 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
828 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
830 2017-11-30 Richard Biener <rguenther@suse.de>
832 PR tree-optimization/83202
833 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
834 allow_peel argument and guard peeling.
835 (canonicalize_loop_induction_variables): Likewise.
836 (canonicalize_induction_variables): Pass false.
837 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
838 peeling from cunrolli.
840 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
842 * combine.c (try_combine): Print a message to dump file whenever
843 I0, I1, or I2 cannot be combined into I3.
845 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
847 PR rtl-optimization/83156
848 PR rtl-optimization/82621
849 * combine.c (try_combine): Don't split an I2 if one of the dests is
850 set again before I3. Allow unused dests.
852 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
854 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
856 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
858 PR rtl-optimization/80818
859 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
860 recursively. Use insn code for clobber.
861 (lra_set_insn_recog_data): Pass the new arg to
862 collect_non_operand_hard_regs.
863 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
865 (lra_update_insn_regno_info): Pass insn to
866 add_regs_to_insn_regno_info.
868 2017-11-29 Jim Wilson <jimw@sifive.com>
869 Andrew Waterman <andrew@sifive.com>
871 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
872 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
873 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
874 use SINGLE_SHIFT_COST.
875 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
876 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
878 2017-11-29 Julia Koval <julia.koval@intel.com>
880 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
881 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
882 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
883 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
884 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
885 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
886 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
887 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
888 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
889 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
890 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
891 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
892 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
893 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
894 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
895 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
896 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
897 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
898 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
899 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
900 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
901 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
902 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
903 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
904 __builtin_ia32_vpshldv_v2di_mask,
905 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
906 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
907 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
909 2017-11-29 Julia Koval <julia.koval@intel.com>
911 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
912 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
913 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
914 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
915 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
916 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
917 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
918 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
919 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
920 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
921 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
922 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
923 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
924 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
925 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
926 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
927 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
928 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
929 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
930 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
931 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
933 2017-11-29 Daniel Cederman <cederman@gaisler.com>
935 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
936 movsi_pic_gotdata_op instruction as a load for the UT699 errata
939 2017-11-29 Martin Aberg <maberg@gaisler.com>
941 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
942 to prevent b2bst errata sequence.
943 (sqrtdf2_fix): Likewise.
945 2017-11-29 Daniel Cederman <cederman@gaisler.com>
947 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
948 (div_sqrt_insn_p): New function.
949 (sparc_do_work_around_errata): Insert NOP instructions to
950 prevent sequences that could trigger the TN-0013 errata for
951 certain LEON3 processors.
952 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
953 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
954 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
955 (in_branch_delay): Prevent div and sqrt in delay slot if
957 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
959 2017-11-29 Daniel Cederman <cederman@gaisler.com>
961 * config/sparc/sparc.c (atomic_insn_p): New function.
962 (sparc_do_work_around_errata): Insert NOP instructions to
963 prevent sequences that could trigger the TN-0010 errata for
965 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
966 instruction referable in atomic_insns_p.
968 2017-11-29 Daniel Cederman <cederman@gaisler.com>
970 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
971 (atomic_compare_and_swap_leon3_1): Likewise.
974 2017-11-29 Daniel Cederman <cederman@gaisler.com>
976 * config/sparc/sparc.c (fpop_insn_p): New function.
977 (sparc_do_work_around_errata): Insert NOP instructions to
978 prevent sequences that could trigger the TN-0012 errata for
980 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
981 * config/sparc/sparc.md (fix_gr712rc): New attribute.
982 (in_branch_annul_delay): Prevent floating-point instructions
983 in delay slot of annulled integer branch.
985 2017-11-29 Richard Biener <rguenther@suse.de>
987 PR tree-optimization/83202
988 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
990 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
991 nodes vectorized to the same stmts multiple times.
992 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
993 (vect_analyze_slp_instance): Adjust.
994 (scalar_stmts_to_slp_tree_map_t): New typedef.
995 (vect_schedule_slp_instance): Add a map recording the SLP node
996 representing the vectorized stmts for a set of scalar stmts.
997 Avoid code-generating redundancies.
998 (vect_schedule_slp): Allocate map and pass it down.
1000 2017-11-29 Nathan Sidwell <nathan@acm.org>
1003 * tree.c (build_complex_type): Fix canonicalization. Only fill in
1006 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
1008 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
1009 cases for printing LDP/STP memory addresses.
1010 (aarch64_print_address_internal): Renamed from
1011 aarch64_print_operand_address, added parameter, add Pmode check.
1012 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
1013 (aarch64_print_operand_address): Indirect to
1014 aarch64_print_address_internal.
1015 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
1016 'y' operand output specifier.
1018 2017-11-29 Jakub Jelinek <jakub@redhat.com>
1021 * tree.c (build_simple_mem_ref_loc): Handle
1022 get_addr_base_and_unit_offset returning a MEM_REF.
1025 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
1028 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
1031 2017-11-28 David Malcolm <dmalcolm@redhat.com>
1033 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
1035 (layout::move_to_column): Likewise.
1037 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1040 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
1041 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
1043 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1044 Martin Jambor <mjambor@suse.cz>
1047 * tree.h (expr_type_first_operand_type_p): Declare
1048 * tree.c (expr_type_first_operand_type_p): New function.
1049 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
1050 (ipa_value_from_jfunc): Adjust declaration.
1051 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
1052 Use it as result type for arithmetics, unless it is NULL in which case
1053 be more conservative.
1054 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
1055 ipa_get_jf_pass_through_result.
1056 (propagate_vals_across_pass_through): Likewise.
1057 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
1058 is to propagate_vals_across_pass_through.
1059 (propagate_constants_across_call): Pass PARM_TYPE to
1060 propagate_scalar_across_jump_function.
1061 (find_more_scalar_values_for_callers_subset): Pass parameter type to
1062 ipa_value_from_jfunc.
1063 (cgraph_edge_brings_all_scalars_for_node): Likewise.
1064 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
1065 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
1066 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
1067 target_type, pass it to ipa_value_from_jfunc.
1068 (update_indirect_edges_after_inlining): Pass parameter type to
1069 try_make_edge_direct_simple_call.
1071 2017-11-28 Jeff Law <law@redhat.com>
1073 * gimple-ssa-evrp-analyze.c
1074 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
1075 refine ranges if scev_initialized_p returns true.
1076 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
1078 2017-11-28 Julia Koval <julia.koval@intel.com>
1080 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
1081 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
1082 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
1083 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
1084 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
1085 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
1086 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
1087 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
1088 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
1089 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
1090 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
1091 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
1092 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
1093 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
1094 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
1095 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
1096 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
1097 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
1098 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
1099 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
1100 __builtin_ia32_vpshrd_v2di_mask): New builtins.
1101 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
1103 2017-11-28 Julia Koval <julia.koval@intel.com>
1105 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
1106 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
1107 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
1108 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
1109 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
1110 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
1111 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
1112 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
1113 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
1114 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
1115 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
1116 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
1117 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
1118 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
1119 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
1120 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
1121 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
1122 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
1123 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
1124 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
1125 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
1126 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
1127 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
1128 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
1129 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
1130 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
1131 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
1132 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
1133 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
1134 __builtin_ia32_vpshld_v2di_mask): New builtins.
1135 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1136 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
1138 2017-11-28 Richard Biener <rguenther@suse.de>
1140 PR tree-optimization/80776
1141 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
1143 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
1145 (evrp_range_analyzer::record_ranges_from_incoming_edges):
1146 If the incoming edge is an effective fallthru because the other
1147 edge only reaches a __builtin_unreachable () then record ranges
1148 derived from the controlling condition in SSA info.
1149 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
1150 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1152 2017-11-28 Olivier Hainque <hainque@adacore.com>
1154 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
1155 on mingw build hosts.
1157 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1160 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
1162 2017-11-28 Richard Biener <rguenther@suse.de>
1165 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
1166 copies generated from memcpy use a character array as reference
1169 2017-11-28 Julia Koval <julia.koval@intel.com>
1170 Sebastian Peryt <sebastian.peryt@intel.com>
1172 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
1173 c-family/c-cilkplus.o, c-family/array-notation-common.o,
1174 cilk-common.o, cilk.h, cilk-common.c): Remove.
1176 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
1177 * builtins.c (is_builtin_name): Remove cilkplus condition.
1178 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
1179 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
1180 cilk-builtins.def, cilkplus.def): Remove.
1181 * cif-code.def (CILK_SPAWN): Remove.
1182 * cilk-builtins.def: Delete.
1183 * cilk-common.c: Ditto.
1185 * cilkplus.def: Ditto.
1186 * config/darwin.h (fcilkplus): Delete.
1187 * cppbuiltin.c: Ditto.
1188 * doc/extend.texi: Remove cilkplus doc.
1189 * doc/generic.texi: Ditto.
1190 * doc/invoke.texi: Ditto.
1191 * doc/passes.texi: Ditto.
1192 * gcc.c (fcilkplus): Remove.
1193 * gengtype.c (cilk.h): Remove.
1194 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
1196 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
1198 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
1200 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
1201 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
1202 cilkplus conditions.
1203 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
1204 inline_read_section): Ditto.
1205 * ipa-inline-analysis.c (cilk.h): Remove.
1206 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
1207 * lto-wrapper.c (merge_and_complain, append_compiler_options,
1208 append_linker_options): Remove condition for fcilkplus.
1209 * lto/lto-lang.c (cilk.h): Remove.
1210 (lto_init): Remove condition for fcilkplus.
1211 * omp-expand.c (expand_cilk_for_call): Delete.
1212 (expand_omp_taskreg, expand_omp_for_static_chunk,
1213 expand_omp_for): Remove cilkplus
1215 (expand_cilk_for): Delete.
1216 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
1217 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
1218 execute_lower_omp, diagnose_sb_0): Ditto.
1219 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
1220 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1221 * tree-nested.c: Ditto.
1222 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
1223 (dump_generic_node): Ditto.
1224 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
1225 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
1227 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
1229 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1231 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
1232 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
1233 for vector mode and !TARGET_SIMD.
1235 2017-11-28 Jakub Jelinek <jakub@redhat.com>
1237 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
1239 * tree.h (SWITCH_LABELS): Remove.
1240 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
1241 assert SWITCH_BODY is non-NULL.
1242 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
1244 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
1246 PR tree-optimization/80788
1247 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
1248 has TREE_OVERFLOW set, call drop_tree_overflow.
1250 2017-11-28 Richard Biener <rguenther@suse.de>
1252 PR tree-optimization/83158
1253 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more
1256 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
1259 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
1260 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
1262 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
1264 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
1265 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
1267 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1270 * varasm.c (compare_constant): Take the mode of the constants into
1271 account when comparing floating point constants.
1273 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
1275 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
1276 from explicit instantiation of debug_helper.
1277 * vec.h (DEFINE_DEBUG_VEC): Ditto.
1279 2017-11-27 Richard Biener <rguenther@suse.de>
1281 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
1284 2017-11-27 Richard Biener <rguenther@suse.de>
1286 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
1287 (type_hash_canon): Also clear int_cst_hash_table entry for
1289 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
1291 2017-11-27 Tamar Christina <tamar.christina@arm.com>
1293 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1294 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1295 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
1297 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1299 * hash-map.h (gt_cleare_cache): Avoid UB.
1301 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
1303 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
1304 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
1305 (decide_unroll_runtime_iterations): Likewise.
1306 (decide_unroll_stupid): Likewise.
1308 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1311 * config/i386/i386.md: Add a loop with incssp.
1313 2017-11-27 Martin Jambor <mjambor@suse.cz>
1315 PR tree-optimization/81248
1316 * tree-sra.c (splice_param_accesses): Remove size check.
1317 (decide_one_param_reduction): Fix size check.
1318 * gimple-pretty-print.c (dump_profile): Silence warning.
1319 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
1321 2017-11-27 Jakub Jelinek <jakub@redhat.com>
1324 * dbxout.c (lastlineno): New variable.
1325 (dbx_debug_hooks): Use dbxout_switch_text_section as
1326 switch_text_section debug hook.
1327 (dbxout_function_end): Switch to current_function_section
1328 rather than function_section. If crtl->has_bb_partition,
1329 output just one N_FUN, depending on in_cold_section_p.
1330 (dbxout_source_line): Remember last lineno in lastlineno.
1331 (dbxout_switch_text_section): New function.
1332 (dbxout_function_decl): Adjust dbxout_block caller.
1333 (dbx_block_with_cold_children): New function.
1334 (dbxout_block): Return true if any LBRAC/RBRAC have been
1335 emitted. Use dbx_block_with_cold_children at depth == 0
1336 in second partition. Add PARENT_BLOCKNUM argument, pass
1337 it optionally adjusted to children. Output LBRAC/RBRAC
1338 around recursive call only if the block is in the current
1339 partition, if not and anything was output, emit empty
1341 * final.c (final_scan_insn): Compute cold_function_name
1342 before calling switch_text_section debug hook. Call
1343 that hook even if dwarf2out_do_frame if not emitting
1347 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
1348 TREE_READONLY decls.
1350 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
1352 PR rtl-optimization/82488
1353 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
1355 2017-11-26 Julia Koval <julia.koval@intel.com>
1357 * config/i386/i386.c (processor_target_table): Add skylake_cost for
1359 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
1362 2017-11-26 Julia Koval <julia.koval@intel.com>
1364 * config/i386/driver-i386.c (host_detect_local_cpu):
1365 Detect skylake-avx512.
1367 2017-11-26 Julia Koval <julia.koval@intel.com>
1369 * config.gcc: Add -march=cannonlake.
1370 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
1371 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
1372 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
1373 (PTA_CANNONLAKE): New.
1374 (processor_target_table): Add cannonlake.
1375 (ix86_option_override_internal): Ditto.
1376 (fold_builtin_cpu): Ditto.
1377 (get_builtin_code_for_version): Handle cannonlake.
1378 (M_INTEL_COREI7_CANNONLAKE): New.
1379 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
1380 * doc/invoke.texi: Add -march=cannonlake.
1382 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
1384 * plugin.c (add_new_plugin): Use platform-specific library extensions.
1385 (try_init_one_plugin): Alternative implementation for MinGW.
1386 * Makefile.in (plugin_implib): New.
1387 (gengtype-lex.c): Fix broken AIX workaround.
1388 * configure: Regenerate.
1389 * doc/plugins.texi: Document support for MinGW.
1391 2017-11-25 Jakub Jelinek <jakub@redhat.com>
1393 PR rtl-optimization/81553
1394 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
1395 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
1396 is a shift where C1 has different mode than the whole shift, use C1's
1397 mode for MULT rather than the shift's mode.
1400 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
1401 builtins not enabled in the currently selected ISA.
1403 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
1405 PR tree-optimization/71026
1406 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
1407 (insert_reciprocals): Change to insert reciprocals before a division
1408 by a square and to insert the square of a reciprocal.
1409 (execute_cse_reciprocals_1): Change to consider division by a square.
1410 (register_division_in): Add importance parameter.
1412 2017-11-24 Richard Biener <rguenther@suse.de>
1414 PR tree-optimization/82402
1415 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
1416 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
1418 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
1420 * match.pd (0-ptr): New transformation.
1422 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
1425 * ipa-inline.c (inline_small_functions): Set current badnes correctly
1426 when skipping checking.
1428 2017-11-24 Richard Biener <rguenther@suse.de>
1430 PR tree-optimization/83128
1431 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
1432 (vn_reference_lookup_3): Likewise.
1434 2017-11-24 Jakub Jelinek <jakub@redhat.com>
1437 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
1438 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
1441 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
1442 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
1443 check for the latter with gimple_call_builtin_p. Do not handle
1444 BUILT_IN_STPNCPY_CHK which is not a pass through call.
1446 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
1448 * config/arm/arm_neon.h: Fix pragma GCC push_options before
1451 2017-11-23 Julia Koval <julia.koval@intel.com>
1453 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
1454 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
1455 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
1456 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
1457 _mm512_maskz_expandloadu_epi16): New intrinsics.
1458 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
1459 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
1460 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
1461 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
1462 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
1463 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
1464 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
1465 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
1466 _mm256_maskz_expandloadu_epi8): New intrinsics.
1467 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
1468 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
1469 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
1470 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
1471 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
1472 * config/i386/sse.md (VI248_VLBW): New iterator.
1473 (expand<mode>_mask, expand<mode>_maskz): New patterns.
1475 2017-11-23 Julia Koval <julia.koval@intel.com>
1477 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
1478 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
1479 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
1480 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
1481 _mm512_mask_compressstoreu_epi16): New.
1482 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
1483 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
1484 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
1485 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
1486 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
1487 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
1488 _mm256_mask_compressstoreu_epi8): New.
1489 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
1490 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
1491 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
1492 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
1493 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
1494 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
1495 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
1496 __builtin_ia32_compresshi128_mask,
1497 __builtin_ia32_compressstoreuqi512_mask,
1498 __builtin_ia32_compressstoreuhi512_mask,
1499 __builtin_ia32_compressstoreuqi256_mask,
1500 __builtin_ia32_compressstoreuqi128_mask,
1501 __builtin_ia32_compressstoreuhi256_mask,
1502 __builtin_ia32_compressstoreuhi128_mask): New builtins.
1503 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
1505 (ix86_expand_special_args_builtin): Handle new types.
1506 (s4fma_expand): Handle new builtin array.
1507 * config/i386/immintrin.h: Include new headers.
1508 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
1509 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
1511 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1513 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
1514 fixes. Declare temp and g variables at the top in order to avoid
1515 {} in most of the cases.
1517 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
1519 * match.pd (ptr-0): New transformation.
1521 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1523 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
1524 (struct addr_mode_cost_table): New.
1525 (struct tune_params): Add field addr_mode_costs.
1526 * config/arm/arm.c (generic_addr_mode_costs): New.
1527 (arm_slowmul_tune): Initialise addr_mode_costs field.
1528 (arm_fastmul_tune): Likewise.
1529 (arm_strongarm_tune): Likewise.
1530 (arm_xscale_tune): Likewise.
1531 (arm_9e_tune): Likewise.
1532 (arm_marvell_pj4_tune): Likewise.
1533 (arm_v6t2_tune): Likewise.
1534 (arm_cortex_tune): Likewise.
1535 (arm_cortex_a8_tune): Likewise.
1536 (arm_cortex_a7_tune): Likewise.
1537 (arm_cortex_a15_tune): Likewise.
1538 (arm_cortex_a35_tune): Likewise.
1539 (arm_cortex_a53_tune): Likewise.
1540 (arm_cortex_a57_tune): Likewise.
1541 (arm_exynosm1_tune): Likewise.
1542 (arm_xgene1_tune): Likewise.
1543 (arm_cortex_a5_tune): Likewise.
1544 (arm_cortex_a9_tune): Likewise.
1545 (arm_cortex_a12_tune): Likewise.
1546 (arm_cortex_a73_tune): Likewise.
1547 (arm_v7m_tune): Likewise.
1548 (arm_cortex_m7_tune): Likewise.
1549 (arm_v6m_tune): Likewise.
1550 (arm_fa726te_tune): Likewise.
1551 (arm_mem_costs): Use table lookup to calculate cost of addressing
1554 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
1556 * config/arm/arm.c (arm_mem_costs): New function.
1557 (arm_rtx_costs_internal): Use arm_mem_costs.
1559 2017-11-23 Mark Wielaard <mark@klomp.org>
1561 * dwarf2out.c (init_sections_and_labels): Use generation to create
1562 unique ranges_section_label and ranges_base_label. Return generation.
1563 (output_rnglists): Add generation argument. Use generation to create
1564 unique ranges labels.
1565 (dwarf2out_finish): Get generation from init_sections_and_labels
1566 and pass generation to output_rnglists.
1568 2017-11-23 Mike Stump <mikestump@comcast.net>
1569 Eric Botcazou <ebotcazou@adacore.com>
1571 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
1572 * cfgloop.h (struct loop): Add unroll field.
1573 * function.h (struct function): Add has_unroll bitfield.
1574 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
1575 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
1576 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
1577 (pass_rtl_unroll_loops::gate): Likewise.
1578 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
1579 for which loop->unroll==1.
1580 (decide_unroll_constant_iterations): Use note for consistency and
1581 take loop->unroll into account. Return early if loop->unroll is set.
1582 Fix thinko in existing test.
1583 (decide_unroll_runtime_iterations): Use note for consistency and
1584 take loop->unroll into account.
1585 (decide_unroll_stupid): Likewise.
1586 * lto-streamer-in.c (input_cfg): Read loop->unroll.
1587 * lto-streamer-out.c (output_cfg): Write loop->unroll.
1588 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
1590 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
1591 (print_loop): Print loop->unroll if set.
1592 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
1593 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
1594 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
1596 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
1597 loop->unroll is set and smaller than the trip count. Otherwise bypass
1598 entirely the heuristics if loop->unroll is set. Remove dead note.
1599 Fix off-by-one bug in other note.
1600 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
1601 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
1603 (tree_unroll_loops_completely): Make static.
1604 (pass_complete_unroll::execute): Use correct type for variable.
1605 (pass_complete_unrolli::execute): Fix formatting.
1606 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
1608 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
1610 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
1611 enable when TARGET_PREFER_AVX128 is set.
1613 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1615 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
1617 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
1618 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
1620 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1622 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
1623 frequencies. Fix estimation of aggregate parameters.
1625 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
1627 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
1630 2017-11-23 Tom de Vries <tom@codesourcery.com>
1632 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
1634 2017-11-23 Tom de Vries <tom@codesourcery.com>
1636 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
1639 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
1642 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
1643 sibcall_value_pcrel_fdpic): Use local variable instead of
1645 (calli_tbr_rel): Add missing operand 2.
1646 (call_valuei_tbr_rel): Add missing operand 3.
1648 2017-11-23 Jakub Jelinek <jakub@redhat.com>
1651 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
1652 bitpos/bitsize covering the whole destination, use store_expr only if
1653 the complex mode is the same. Otherwise, use expand_normal and if
1654 it returns CONCAT, subreg each part separately instead of trying to
1655 subreg the whole result.
1657 2017-11-23 Richard Biener <rguenther@suse.de>
1659 PR tree-optimization/23094
1660 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
1661 come here from walking over backedges in the first iteration.
1662 (vn_reference_lookup_3): Skip clobbers that store the same value.
1664 2017-11-23 Richard Biener <rguenther@suse.de>
1666 PR tree-optimization/81403
1667 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
1668 a block we need a leader relative to.
1669 (phi_translate_1): For nary processing require a leader from
1670 get_representative_for given we run expression simplification
1671 using match-and-simplify. Remove previous fix.
1673 2017-11-22 Jeff Law <law@redhat.com>
1675 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
1676 Use new method allocate_value_range rather than accessing the
1677 vrp_value_range_pool data member directly.
1678 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
1679 to use extract_range_from_stmt method to avoid need for
1680 extract_range_from_assignment method.
1681 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
1682 method rather than setting values_propgated data member directly.
1683 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
1684 and values propagated data members and extract_range_from_assignment
1685 method. Reorder private data members to conform to standards.
1686 Add new methods set_lattice_propagation_complete and
1687 allocate_value_range.
1689 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
1691 PR rtl-optimization/83030
1692 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
1693 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
1694 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
1695 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
1696 CROSSING_JUMP_P flag is already set before setting it.
1697 * cfgrtl.c (fixup_partition_crossing): Likewise.
1698 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
1701 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1703 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
1704 Handle the case where both arguments are using gen_const_vec_series.
1706 2017-11-22 David Malcolm <dmalcolm@redhat.com>
1709 * pretty-print.c (pp_format): Move quoting implementation to
1710 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
1711 to pass address of "quote" local.
1712 (pp_begin_quote): New function.
1713 (pp_end_quote): New function.
1714 * pretty-print.h (printer_fn): Convert penultimate param from bool
1716 (pp_begin_quote): New decl.
1717 (pp_end_quote): New decl.
1718 * tree-diagnostic.c (default_tree_printer): Convert penultimate
1719 param from bool to bool *.
1720 * tree-diagnostic.h (default_tree_printer): Likewise.
1722 2017-11-22 Jeff Law <law@redhat.com>
1724 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
1725 Initialize vr_values.
1726 (evrp_range_analyzer::try_find_new_range): Call methods attached to
1727 vr_values via vr_values class instance rather than delegators.
1728 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
1729 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
1730 (evrp_range_analyzer::push_value_range): Likewise.
1731 (evrp_range_analyzer::pop_value_range): Likewise.
1732 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
1733 most delegators. Those remaining are exposed as public interfaces.
1734 Make vr_values a pointer and private.
1735 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
1737 (evrp_range_analyzer::get_vr_value): New method.
1738 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
1739 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
1740 (class evrp_dom_walker): Attach evrp_folder class, initialize
1741 it in the ctor. Remove temporary delegators.
1742 (evrp_dom_walker::before_dom_children): Call methods in attached
1743 evrp_range_analyzer class via class instance pointer. Use
1744 free value_range_constant_singleton to remove need for
1745 op_with_constant_singleton_value delegator method. Do not
1746 create a vrp_prop class instance for every call! Narrow
1747 scope of a couple variables.
1748 (evrp_dom_walker::cleanup): Call methods in attached
1749 evrp_range_analyzer class via class instance pointer.
1750 * vr-values.h (class vr_values): Privatize many methods and
1753 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1755 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
1756 * cfgexpand.c (expand_debug_expr): Remove handling for them.
1757 * expr.c (expand_expr_real_2): Likewise.
1758 * fold-const.c (const_unop): Likewise.
1759 * optabs-tree.c (optab_for_tree_code): Likewise.
1760 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1761 * tree-inline.c (estimate_operator_cost): Likewise.
1762 * tree-pretty-print.c (dump_generic_node): Likewise.
1763 (op_code_prio): Likewise.
1764 (op_symbol_code): Likewise.
1765 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
1766 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
1767 * internal-fn.c (direct_internal_fn_optab): New function.
1768 (direct_internal_fn_array, direct_internal_fn_supported_p
1769 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
1770 * fold-const-call.c (fold_const_reduction): New function.
1771 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
1773 * tree-vect-loop.c: Include internal-fn.h.
1774 (reduction_code_for_scalar_code): Rename to...
1775 (reduction_fn_for_scalar_code): ...this and return an internal
1777 (vect_model_reduction_cost): Take an internal_fn rather than
1779 (vect_create_epilog_for_reduction): Likewise. Build calls rather
1781 (vectorizable_reduction): Use internal functions rather than tree
1782 codes for the reduction operation. Update calls to the functions
1784 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
1785 Use calls to internal functions rather than REDUC tree codes.
1786 * config/aarch64/aarch64-simd.md: Update comment accordingly.
1788 2017-11-22 Olivier Hainque <hainque@adacore.com>
1790 * config/vxworks.c (vxworks_override_options): Pick default
1791 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
1792 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
1793 DWARF_GNAT_ENCODINGS_DEFAULT.
1794 * config/vxworksae.h: Likewise.
1796 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
1798 PR tree-optimization/83104
1799 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
1802 2017-11-22 Marek Polacek <polacek@redhat.com>
1803 H.J. Lu <hongjiu.lu@intel.com>
1804 Jason Merrill <jason@redhat.com>
1809 * calls.c (initialize_argument_information): Call
1810 warn_parameter_passing_abi target hook.
1811 (store_one_arg): Use 0 for empty record size. Don't push 0 size
1812 argument onto stack.
1813 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
1814 * common.opt: Update -fabi-version description.
1815 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
1816 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
1818 (ix86_is_empty_record): New function.
1819 (ix86_warn_parameter_passing_abi): New function.
1820 (TARGET_EMPTY_RECORD_P): Redefine.
1821 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
1822 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
1823 * doc/tm.texi: Regenerated.
1824 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
1825 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
1826 * dwarf2out.c (get_ultimate_context): Move to tree.c.
1827 * explow.c (hard_function_value): Call arg_int_size_in_bytes
1828 instead of int_size_in_bytes.
1829 * expr.c (copy_blkmode_to_reg): Likewise.
1830 * function.c (aggregate_value_p): Return 0 for empty types.
1831 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi target hook.
1832 (locate_and_pad_parm): Call arg size_in_bytes instead
1834 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
1835 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
1836 * target.def (empty_record_p, warn_parameter_passing_abi): New target
1838 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
1839 (std_gimplify_va_arg_expr): Skip empty records. Call
1840 arg_size_in_bytes instead size_in_bytes.
1841 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
1842 * tree-core.h (tree_type_common): Add empty_flag.
1843 (tree_decl_common): Update comments.
1844 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
1846 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1847 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
1849 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
1850 * tree.c (default_is_empty_type): New function.
1851 (default_is_empty_record): New function.
1852 (arg_int_size_in_bytes): New function.
1853 (arg_size_in_bytes): New function.
1854 (get_ultimate_context): New function.
1855 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
1856 TRANSLATION_UNIT_WARN_EMPTY_P.
1857 (default_is_empty_record, arg_int_size_in_bytes,
1858 arg_size_in_bytes, get_ultimate_context): Declare.
1860 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1862 * config/arm/arm.c (cmse_clear_registers): New function.
1863 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
1864 code by call to cmse_clear_registers.
1865 (cmse_nonsecure_entry_clear_before_return): Likewise.
1867 2017-11-22 Tamar Christina <tamar.christina@arm.com>
1869 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
1870 (vdot_s32, vdotq_s32): New.
1871 (vdot_lane_u32, vdotq_lane_u32): New.
1872 (vdot_lane_s32, vdotq_lane_s32): New.
1875 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1878 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
1879 for unsigned values with fewer HWIs than the precision.
1880 (test_overflow): New function.
1881 (wide_int_cc_tests): Call it.
1883 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
1885 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
1886 has the same mode class as Pmode.
1888 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1890 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
1891 padding_bits_to_clear_ptr.
1892 (cmse_nonsecure_entry_clear_before_return): Likewise.
1894 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1896 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
1897 auto_sbitap instead of integer bitfield to control register needing
1900 2017-11-22 Jakub Jelinek <jakub@redhat.com>
1902 PR tree-optimization/83044
1903 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
1904 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
1905 that need the upper bound. Subtract offset from
1906 get_addr_base_and_unit_offset only if positive and subtract it
1907 before division by eltsize rather than after it.
1910 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
1911 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
1912 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
1916 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
1917 expand_binop with *mul_widen_optab, make sure at least one of the
1918 operands doesn't have VOIDmode.
1921 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
1923 PR rtl-optimization/82044
1924 PR tree-optimization/82042
1925 * dse.c (record_store): Check for overflow.
1926 (check_mem_read_rtx): Properly check for overflow if width == -1, call
1927 add_wild_read instead of clear_rhs_from_active_local_stores on
1928 overflow and log it into dump_file.
1930 2017-11-22 Richard Biener <rguenther@suse.de>
1932 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
1933 fake edges to exit when looking for a place to insert.
1934 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
1936 (insert_into_preds_of_block): Commit edge insertion immediately,
1937 assert that doesn't require new BBs.
1938 (fini_pre): Release expressions.
1939 (pass_pre::execute): Shuffle things around a bit, if the fn
1940 is too large do not compute AVAIL either as this is really the
1943 2017-11-22 Richard Biener <rguenther@suse.de>
1945 PR tree-optimization/83089
1946 * tree-if-conv.c (pass_if_conversion::execute): If anything
1947 changed reset SCEV and free the number of iteration estimates.
1949 2017-11-21 Martin Sebor <msebor@redhat.com>
1951 PR tree-optimization/82945
1952 * calls.h (warn_nonstring_bound): Remove unused function.
1954 2017-11-21 Martin Sebor <msebor@redhat.com>
1956 PR tree-optimization/82945
1957 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
1958 * calls.h (maybe_warn_nonstring_arg): Declare new function.
1959 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
1961 (initialize_argument_information): Call maybe_warn_nonstring_arg.
1962 * calls.h (get_attr_nonstring_decl): Declare new function.
1963 * doc/extend.texi (attribute nonstring): Update.
1964 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
1965 get_attr_nonstring_decl and handle it.
1966 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
1967 detection of nul-termination.
1968 (strlen_to_stridx): Change to a pointer.
1969 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
1970 (pass_strlen::execute): Same.
1972 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
1974 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
1975 for the new option -mprefer-vector-width=[none|128|256|512].
1976 * config/i386/i386.c (ix86_target_string): remove old style options
1977 -mprefer-avx256 and make -mprefer-avx128 as alias.
1978 (ix86_option_override_internal): Apply defaults for the
1979 -mprefer-vector-width=[128|256] option.
1980 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
1981 Implement macros to work with -mprefer-vector-width=.
1982 * config/i386/i386.opt: Implemented option
1983 -mprefer-vector-width=[none|128|256|512].
1984 * doc/invoke.texi: Documentation for
1985 -mprefer-vector-width=[none|128|256|512].
1987 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
1989 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
1991 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
1993 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
1995 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
1997 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
1999 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
2000 * expr.c (expand_expr_real_2): Likewise.
2001 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
2002 fold_binary_loc): Likewise.
2003 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
2004 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
2005 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
2006 MINUS_EXPR transformations.
2007 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
2008 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
2009 * tree-inline.c (estimate_operator_cost): Likewise.
2010 * tree-pretty-print.c (dump_generic_node, op_code_prio,
2011 op_symbol_code): Likewise.
2012 * tree-vect-stmts.c (vectorizable_operation): Likewise.
2013 * vr-values.c (extract_range_from_binary_expr): Likewise.
2014 * varasm.c (initializer_constant_valid_p_1): Likewise.
2015 * tree.def: New tree code POINTER_DIFF_EXPR.
2017 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
2019 * config/i386/i386.md (*bswap<mode>2_movbe): Add
2020 integer suffix to movbe mnemonic.
2021 (*bswaphi2_movbe): Ditto.
2022 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
2024 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2027 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
2028 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
2029 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
2031 2017-11-21 Martin Liska <mliska@suse.cz>
2033 * tree-inline.c (expand_call_inline): Remove not needed
2036 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
2037 Jakub Jelinek <jakub@redhat.com>
2040 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
2041 Remove static keyword from f variable.
2043 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2045 PR tree-optimization/83086
2046 * gimple-ssa-store-merging.c
2047 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
2048 rather than n.base_addr.
2050 2017-11-21 Martin Liska <mliska@suse.cz>
2052 PR rtl-optimization/82044
2053 PR tree-optimization/82042
2054 * dse.c (check_mem_read_rtx): Check for overflow.
2056 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2058 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
2061 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2063 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
2064 and its corresponding call arg location note.
2066 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
2067 Andrew Burgess <andrew.burgess@embecosm.com>
2069 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
2071 (arc_return_slot_offset): Likewise.
2072 (arc_eh_return_address_location): New declaration.
2073 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
2074 (MUST_SAVE_REGISTER): Add exception handler case.
2075 (MUST_SAVE_RETURN_ADDR): Likewise.
2076 (arc_frame_pointer_required): Likewise.
2077 (arc_frame_pointer_needed): New function.
2078 (arc_compute_frame_size): Changed.
2079 (arc_expand_prologue): Likewise.
2080 (arc_expand_epilogue): Likewise.
2081 (arc_initial_elimination_offset): Likewise.
2082 (arc_return_slot_offset): Delete.
2083 (arc_eh_return_address_location): New function.
2084 (arc_builtin_setjmp_frame_value): Likewise.
2085 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
2086 (EH_RETURN_STACKADJ_RTX): Define.
2087 (EH_RETURN_HANDLER_RTX): Likewise.
2088 * config/arc/arc.md (eh_return): Delete.
2090 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
2092 * print-tree.h (debug_vec_tree): Remove prototype.
2093 * gdbinit.in (pvt): Remove macro.
2095 2017-11-21 Jakub Jelinek <jakub@redhat.com>
2097 PR tree-optimization/83047
2098 * gimple-ssa-store-merging.c
2099 (imm_store_chain_info::output_merged_store): If the loads with the
2100 same vuse are in different basic blocks, for load_gsi pick a load
2101 location that is dominated by the other loads.
2104 * config/i386/i386.c (ix86_memmodel_check): Start
2105 -Winvalid-memory-model diagnostics with lowercase letter.
2108 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
2109 set in_cold_section_p to the partition containing loc_list->first.
2110 When seeing loc_list->last_before_switch node, update secname and
2111 perform range_across_switch second partition handling only after that.
2114 * run-rtl-passes.c: Include debug.h.
2115 (run_rtl_passes): Call debug_hooks->assembly_start.
2116 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
2120 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
2121 OPTAB_DIRECT in calls to expand_simple_binop.
2123 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2126 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
2128 2017-11-20 David Malcolm <dmalcolm@redhat.com>
2131 * spellcheck.h (best_match::blithely_get_best_candidate): New
2134 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2136 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
2137 with lower case letter.
2139 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
2141 * config/i386/i386.md (bswaphi2): New expander.
2142 (*bswaphi2_movbe): New insn pattern.
2143 (bswaphi -> rorhi pepehole2): New peephole pattern.
2145 2017-11-20 Jeff Law <law@redhat.com>
2147 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
2148 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
2149 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
2150 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
2151 gimple-ssa-evrp-analyze.h.
2153 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
2154 set BB_VISITED here.
2155 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
2157 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2159 * tree-predcom.c: Add general comment on Store-Store chains.
2160 (split_data_refs_to_components): Postpone clearing eliminate_store_p
2162 (get_chain_last_ref_at): Rename into...
2163 (get_chain_last_write_at): ...this.
2164 (get_chain_last_write_before_load): New function.
2165 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
2166 CT_STORE_STORE when write reference is added.
2167 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
2168 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
2169 (initialize_root_vars_store_elim_1): Ditto.
2170 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
2171 definition is created.
2172 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
2173 chain by replacing it with dominant stored value.
2175 2017-11-20 Bin Cheng <bin.cheng@arm.com>
2177 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
2179 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
2181 * vr-values.c (extract_range_from_binary_expr): Use a full range
2184 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
2186 * config/arm/arm.md (R4_REGNUM): Define constant.
2187 (nonsecure_call_internal): Remove r4 clobber.
2188 (nonsecure_call_value_internal): Likewise.
2189 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
2190 clobber and resequence match_operands.
2191 (nonsecure_call_value_reg_thumb1_v5): Likewise.
2192 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
2193 (nonsecure_call_value_reg_thumb2): Likewise.
2195 2017-11-20 Jakub Jelinek <jakub@redhat.com>
2197 PR tree-optimization/78821
2198 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
2199 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
2200 to the address of the base rather than the base itself.
2201 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
2202 (find_bswap_or_nop_finalize): New function.
2203 (find_bswap_or_nop): Use it.
2204 (bswap_replace): Return a tree rather than bool, change first
2205 argument from gimple * to gimple_stmt_iterator, allow inserting
2206 into an empty sequence, allow ins_stmt to be NULL - then emit
2207 all stmts into gsi. Fix up MEM_REF address gimplification.
2208 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
2209 (struct store_immediate_info): Add N and INS_STMT non-static
2211 (store_immediate_info::store_immediate_info): Initialize them
2212 from newly added ctor args.
2213 (merged_store_group::apply_stores): Formatting fixes. Sort by
2215 (stmts_may_clobber_ref_p): For stores call also
2216 refs_anti_dependent_p.
2217 (gather_bswap_load_refs): New function.
2218 (imm_store_chain_info::try_coalesce_bswap): New method.
2219 (imm_store_chain_info::coalesce_immediate_stores): Use it.
2220 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
2221 (imm_store_chain_info::output_merged_store): Fail if number of
2222 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
2223 and NOP_EXPR rhs_code.
2224 (pass_store_merging::process_store): Compute n and ins_stmt, if
2225 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
2226 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
2228 (pass_store_merging::execute): Calculate dominators.
2230 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
2231 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
2232 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
2233 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2234 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
2235 class pass_optimize_bswap, bswap_replace,
2236 pass_optimize_bswap::execute): Moved to ...
2237 * gimple-ssa-store-merging.c: ... this file.
2238 Include optabs-tree.h.
2239 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
2240 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
2241 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
2242 anonymous namespace, remove static keywords.
2243 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
2244 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
2246 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
2249 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
2251 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
2253 * vec.h (debug_helper): New function.
2254 (DEFINE_DEBUG_VEC): New macro.
2255 * hash-set.h (debug_helper): New function.
2256 (DEFINE_DEBUG_HASH_SET): New macro.
2257 * cfg.c (debug_slim (edge)): New function.
2258 Call DEFINE_DEBUG_VEC for edges.
2259 Call DEFINE_DEBUG_HASH_SET for edges.
2260 * cfghooks.c (debug_slim (basic_block)): New function.
2261 Call DEFINE_DEBUG_VEC for basic blocks.
2262 Call DEFINE_DEBUG_HASH_SET for basic blocks.
2263 * print-tree.c (debug_slim): New function to handle trees.
2264 Call DEFINE_DEBUG_VEC for trees.
2265 Call DEFINE_DEBUG_HASH_SET for trees.
2266 (debug (vec<tree, va_gc>) &): Remove.
2267 (debug (<vec<tree, va_gc>) *): Remove.
2268 * print-rtl.c (debug_slim): New function to handle const_rtx.
2269 Call DEFINE_DEBUG_VEC for rtx_def.
2270 Call DEFINE_DEBUG_VEC for rtx_insn.
2271 Call DEFINE_DEBUG_HASH_SET for rtx_def.
2272 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
2273 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
2274 (debug (vec<rtx_insn *> *ptr): Remove.
2275 (debug_insn_vector): Remove.
2276 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
2278 2017-11-20 Tom de Vries <tom@codesourcery.com>
2280 PR rtl-optimization/82020
2281 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
2282 IF_THEN_ELSE condition.
2284 2017-11-19 Jeff Law <law@redhat.com>
2286 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2287 of degenerates resulting from ignoring an edge.
2289 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2292 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
2294 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2297 * profile-count.c (profile_count::to_sreal_scale): Fix return value
2298 for uninitialied counts.
2300 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2303 * tree-inline.c (estimate_num_insns): Set to 1 at least.
2305 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
2308 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
2310 2017-11-19 Tom de Vries <tom@codesourcery.com>
2312 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
2315 2017-11-19 Tom de Vries <tom@codesourcery.com>
2317 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
2320 2017-11-19 Tom de Vries <tom@codesourcery.com>
2322 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
2325 2017-11-19 Tom de Vries <tom@codesourcery.com>
2327 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
2329 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
2331 2017-11-19 Tom de Vries <tom@codesourcery.com>
2333 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
2335 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
2336 semicolon after MCORE_EXPORT_NAME call.
2338 2017-11-19 Tom de Vries <tom@codesourcery.com>
2341 * vmsdbgout.c (vmsdbgout_early_finish): New function.
2342 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
2344 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
2346 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
2347 (cgraph_edge::clone): Cleanup updating of profile.
2348 * ipa-cp.c (update_profiling_info): Likewise.
2349 * ipa-inline-transform.c (inline_transform): Likewise.
2350 * ipa-inline.c (inline_small_functions): Add missing space to dump.
2351 * ipa-split.c (execute_split_functions): Do not split when function
2353 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
2354 * profile-count.c (profile_count::dump): Add global0.
2355 (profile_count::to_cgraph_frequency): Do not ICE when entry is
2357 (profile_count::to_sreal_scale): Likewise.
2358 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
2359 (profile_count::combine_with_ipa_count): New function.
2360 * profile-count.h (profile_guessed_global0adjusted): New.
2361 (profile_count::adjusted_zero): New.
2362 (profile_count::global0adjusted): New.
2363 (profile_count::combine_with_ipa_count): New.
2364 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
2365 correct profile of return block of split functions.
2366 (copy_cfg_body): Remove unused profile_count.
2367 (copy_body): Likewise.
2368 (expand_call_inline): Update.
2369 (tree_function_versioning): Update.
2371 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
2373 * hash-set.h (hash_set::empty): New.
2374 * tree-ssa-threadbackward.h: Delete.
2375 * tree-ssa-threadbackward.c (class thread_jumps): New.
2376 Move max_threaded_paths into class.
2377 (fsm_find_thread_path): Remove arguments that are now in class.
2378 (profitable_jump_thread_path): Rename to...
2379 (thread_jumps::profitable_jump_thread_path): ...this.
2380 (convert_and_register_jump_thread_path): Rename to...
2381 (thread_jumps::convert_and_register_current_path): ...this.
2382 (check_subpath_and_update_thread_path): Rename to...
2383 (thread_jumps::check_subpath_and_update_thread_path): ...this.
2384 (register_jump_thread_path_if_profitable): Rename to...
2385 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
2386 (handle_phi): Rename to...
2387 (thread_jumps::handle_phi): ...this.
2388 (handle_assignment): Rename to...
2389 (thread_jumps::handle_assignment): ...this.
2390 (fsm_find_control_statement_thread_paths): Rename to...
2391 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
2392 (find_jump_threads_backwards): Rename to...
2393 (thread_jumps::find_jump_threads_backwards): ...this.
2394 Initialize path local data.
2395 (pass_thread_jumps::execute): Call find_jump_threads_backwards
2396 from within thread_jumps class.
2397 (pass_early_thread_jumps::execute): Same.
2399 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2401 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
2403 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2405 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
2406 to canonical location.
2408 2017-11-17 Steve Ellcey <sellcey@cavium.com>
2411 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
2413 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
2415 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
2417 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
2418 rcrt1.o%s/grcrt1.o%s for -static-pie.
2420 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2422 * i386.c (ix86_multiplication_cost, ix86_division_cost,
2423 ix86_shift_rotate_cost): Break out from ...
2424 (ix86_rtx_costs): ... here.
2425 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
2428 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2430 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
2431 when entry block was promoted unlikely.
2432 (estimate_bb_frequencies): Increase frequency scale.
2433 * profile-count.h (profile_count): Export precision info.
2435 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2437 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
2438 disturbin profile of entry block.
2440 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2442 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
2445 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
2447 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
2448 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
2450 2017-11-17 Jeff Law <law@redhat.com>
2452 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
2453 from evrp_dom_walker class. Various methods moved into new class.
2454 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
2455 (evrp_range_analyzer::enter): New method.
2456 (evrp_range_analyzer::leave): New method.
2457 (evrp_dom_walker): Remove delegators no longer needed by this class.
2458 Replace vr_values data member with evrp_range_analyzer
2460 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
2461 method extracted from evrp_dom_walker::before_dom_children.
2462 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
2463 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
2465 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
2466 Add private copy constructor and move assignment operators.
2467 Privatize methods and class data where trivially possible.
2468 (evrp_dom_walker::cleanup): New function, extracted from
2469 execute_early_vrp. Simplify access to class data.
2471 * vr-values.h (get_output_for_vrp): Prototype.
2472 * vr-values.c (get_output_for_vrp): New function extracted from
2473 vrp_visit_assignment_or_call and extract_range_from_stmt.
2474 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
2476 2017-11-17 Luis Machado <luis.machado@linaro.org>
2478 * config/aarch64/aarch64.c
2479 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
2480 (qdf24xx_tunings) <autoprefetcher_model>: Set to
2481 tune_params::AUTOPREFETCHER_WEAK.
2483 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2486 * config/arm/arm.c (arm_valid_target_attribute_rec):
2487 Parse "arch=" and "+<ext>".
2488 (arm_valid_target_attribute_tree): Re-init global options.
2489 (arm_option_override): Make non-static.
2490 (arm_options_perform_arch_sanity_checks): Make errors fatal.
2491 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
2492 (__ARM_FEATURE_CRC32): Support undef.
2493 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
2494 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
2496 2017-11-17 David Malcolm <dmalcolm@redhat.com>
2498 * gdbinit.in (break-on-diagnostic): New command.
2500 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2502 * config/i386/i386.c (ix86_expand_epilogue): Change simple
2503 return to indirect jump for EH return if control-flow
2504 protection is enabled. Change explicit 'false' argument in
2505 pro_epilogue_adjust_stack with a value of
2507 * config/i386/i386.md (simple_return_indirect_internal):
2508 Remove SImode restriction to support 64-bit.
2510 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2512 * combine.c (added_notes_insn): New.
2513 (try_combine): Handle added_notes_insn like added_links_insn.
2514 Rewrite return value code.
2515 (distribute_notes): Set added_notes_insn to the earliest insn we added
2518 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
2520 PR rtl-optimization/82621
2521 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
2522 dest of one of those SETs is unused.
2524 2017-11-17 Richard Biener <rguenther@suse.de>
2527 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
2528 * tree-pretty-print.c (dump_generic_node): Handle
2529 annot_expr_parallel_kind.
2530 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
2531 * gimplify.c (gimple_boolify): Likewise.
2533 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2535 * config.gcc (extra_headers): Add cet.h for x86 targets.
2536 * config/i386/cet.h: New file.
2537 * doc/install.texi: Add --enable-cet/--disable-cet.
2539 2017-11-17 Richard Biener <rguenther@suse.de>
2541 PR tree-optimization/83017
2542 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
2543 (gen_parallel_loop): Properly count iterations.
2544 (parallelize_loops): Handle loop->can_be_parallel independent
2545 of flag_loop_parallelize_all. Make static profitability test match
2547 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
2548 * invoke.texi (parloops-min-per-thread): Document.
2550 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
2552 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
2553 upstreaming review comments.
2555 2017-11-17 Tamar Christina <tamar.christina@arm.com>
2557 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
2558 with fast unaligned access.
2559 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
2561 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
2563 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
2564 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
2566 2017-11-17 Richard Biener <rguenther@suse.de>
2568 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
2569 folding of references.
2571 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
2574 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
2575 of replacing call to strncmp with corresponding call to strcmp when
2578 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
2580 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
2581 option prefer-avx256 for skylake-avx512 configuration.
2582 * config/i386/i386.c (ix86_option_override_internal): Ditto.
2583 (get_builtin_code_for_version): Ditto.
2585 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2586 Monk Chiang <sh.chiang04@gmail.com>
2588 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
2589 (FIXED_REGISTERS): Reserve more register numbers.
2590 (CALL_USED_REGISTERS): Likewise.
2591 (REG_ALLOC_ORDER): Likewise.
2592 (REG_CLASS_CONTENTS): Likewise.
2593 (REGISTER_NAMES): Likewise.
2595 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
2596 Kito Cheng <kito.cheng@gmail.com>
2598 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
2600 * config/nds32/iterators.md: Add vector mode iterators and attributes.
2602 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
2604 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
2605 parameter list for vec_splats.
2607 2017-11-16 Joseph Myers <joseph@codesourcery.com>
2609 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
2611 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
2613 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
2616 * doc/install.texi: Document incompatibility of Perl >=5.6.26
2617 with the required version of automake 1.11.6.
2619 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
2621 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
2623 (DU_C2_3_power9): Correct reservation combinations.
2624 (FP_DIV_power9, VEC_DIV_power9): New.
2625 (power9-alu): Split out rotate/shift...
2626 (power9-rot): ...to here, correct dispatch resource.
2627 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
2629 (power9-fp): Correct latency.
2630 (power9-sdiv): Add div/sqrt resource.
2631 (power9-ddiv): Correct latency, add div/sqrt resource.
2632 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
2633 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
2635 (power9-qpdiv, power9-qpmul): Adjust resource usage.
2637 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2639 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
2640 switch statement mapping KF built-ins to TF built-ins if we don't
2641 have the proper ISA 3.0 assembler support.
2643 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2645 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
2646 (lower_emutls_function_body): Do not compute it.
2648 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2650 * ipa-split.c (split_bb_info): Turn time to sreal.
2651 (split_point): Likewise.
2652 (dump_split_point): Likewise.
2653 (fine_split_points): Likewise.
2654 (execute_split_functions): Only zero split_bbs; turn time to sreals.
2656 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2658 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
2661 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2663 * predict.c (combine_predictions_for_bb): Preserve zero predicted
2665 (expensive_function_p): Remove useless assert.
2666 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
2668 2017-11-16 Martin Sebor <msebor@redhat.com>
2670 PR tree-optimization/82588
2671 PR tree-optimization/82583
2672 * tree-vrp.c (check_array_ref): Handle flexible array members,
2673 string literals, and inner indices.
2674 (search_for_addr_array): Add detail to diagnostics.
2676 2017-11-16 Nathan Sidwell <nathan@acm.org>
2680 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
2681 (SET_DECL_ASSEMBLER_NAME): Forward to
2682 overwrite_decl_assembler_name.
2683 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
2684 (overwrite_decl_assembler_name): Declare.
2685 * tree.c (overwrite_decl_assembler_name): New.
2686 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
2687 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
2688 (LANG_HOOKS_INITIALIZER): Add it.
2689 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
2690 * langhooks.c (lhd_set_decl_assembler_name): Use
2691 SET_DECL_ASSEMBLER_NAME.
2692 (lhd_overwrite_decl_assembler_name): Default implementation.
2694 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2695 Jackson Woodruff <jackson.woodruff@arm.com>
2697 PR tree-optimization/71026
2698 * match.pd: Canonicalize constant multiplies in division.
2700 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
2702 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
2703 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
2704 Remove OPT_fomit_frame_pointer entry.
2705 * common/config/arc/arc-common.c: Likewise.
2706 * common/config/arm/arm-common.c: Likewise.
2707 * common/config/avr/avr-common.c: Likewise.
2708 * common/config/c6x/c6x-common.c: Likewise.
2709 * common/config/cr16/cr16-common.c: Likewise.
2710 * common/config/cris/cris-common.c: Likewise.
2711 * common/config/epiphany/epiphany-common.c: Likewise.
2712 * common/config/fr30/fr30-common.c: Likewise.
2713 * common/config/frv/frv-common.c: Likewise.
2714 * common/config/ia64/ia64-common.c: Likewise.
2715 * common/config/iq2000/iq2000-common.c: Likewise.
2716 * common/config/lm32/lm32-common.c: Likewise.
2717 * common/config/m32r/m32r-common.c: Likewise.
2718 * common/config/mcore/mcore-common.c: Likewise.
2719 * common/config/microblaze/microblaze-common.c: Likewise.
2720 * common/config/mips/mips-common.c: Likewise.
2721 * common/config/mmix/mmix-common.c: Likewise.
2722 * common/config/mn10300/mn10300-common.c: Likewise.
2723 * common/config/nios2/nios2-common.c: Likewise.
2724 * common/config/pa/pa-common.c: Likewise.
2725 * common/config/pdp11/pdp11-common.c: Likewise.
2726 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
2727 * common/config/riscv/riscv-common.c: Likewise.
2728 * common/config/rs6000/rs6000-common.c: Likewise.
2729 * common/config/rx/rx-common.c: Likewise.
2730 * common/config/s390/s390-common.c: Likewise.
2731 * common/config/sh/sh-common.c: Likewise.
2732 * common/config/sparc/sparc-common.c: Likewise.
2733 * common/config/tilegx/tilegx-common.c: Likewise.
2734 * common/config/tilepro/tilepro-common.c: Likewise.
2735 * common/config/v850/v850-common.c: Likewise.
2736 * common/config/visium/visium-common.c: Likewise.
2737 * common/config/xstormy16/xstormy16-common.c: Likewise.
2738 * common/config/xtensa/xtensa-common.c: Likewise.
2739 * invoke.texi (-fomit-frame-pointer): Update documentation.
2741 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
2743 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
2745 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2747 * cfg.c (scale_bbs_frequencies_int,
2748 cale_bbs_frequencies_gcov_type): Remove.
2749 * cfg.h (scale_bbs_frequencies_int,
2750 cale_bbs_frequencies_gcov_type): Remove.
2752 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2754 * tree-ssa-loop-manip.c
2755 (scale_dominated_blocks_in_loop): Update to profile counts.
2756 (tree_transform_and_unroll_loop): Likewise.
2758 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2760 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
2761 scale_bbs_frequencies_int.
2763 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2765 * final.c (compute_alignments): Use counts rather than frequencies.
2767 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2769 * cfgloopanal.c: Include sreal.h
2770 (average_num_loop_insns): Use counts and sreal for accounting.
2772 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
2774 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
2777 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2778 Kito Cheng <kito.cheng@gmail.com>
2780 * config/nds32/constraints.md: Provide more constraints.
2781 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
2782 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
2783 support constraints usage.
2785 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2787 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
2789 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
2790 Kito Cheng <kito.cheng@gmail.com>
2792 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
2793 * config/nds32/nds32.opt: Refine the layout.
2794 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
2795 TARGET_EXT_STRING): Support new options.
2796 * config/nds32/nds32.h: Likewise.
2797 * config/nds32/nds32.md: Likewise.
2798 * config/nds32/nds32-predicates.c: Likewise.
2799 * config/nds32/constraints.md: Likewise.
2800 * common/config/nds32/nds32-common.c: Likewise.
2802 2017-11-16 Julia Koval <julia.koval@intel.com>
2805 * config/i386/gfniintrin.h: Add sse check.
2806 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
2808 2017-11-16 Julia Koval <julia.koval@intel.com>
2810 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
2811 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
2812 (ix86_handle_option): Handle -mavx512vbmi2.
2813 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
2814 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2815 * config/i386/i386-c.c (__AVX512VBMI2__): New.
2816 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
2817 (ix86_valid_target_attribute_inner_p): Ditto.
2818 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
2819 * config/i386/i386.opt (mavx512vbmi2): New option.
2820 * doc/invoke.texi: Add new option.
2822 2017-11-16 Julia Koval <julia.koval@intel.com>
2824 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
2825 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
2826 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
2827 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
2828 _mm512_gf2p8mul_epi8): New intrinsics.
2829 * config/i386/i386-builtin-types.def
2830 (V64QI_FTYPE_V64QI_V64QI): New type.
2831 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
2832 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
2833 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
2834 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
2835 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
2836 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2838 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
2840 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
2841 explicitly as a stream of bytes.
2843 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2845 * config/rs6000/altivec.h (vec_xst_be): New #define.
2846 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
2847 and externalize from *altivec_vperm_<mode>_internal.
2848 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
2850 (XL_BE_V8HI): Likewise.
2851 (XL_BE_V4SI): Likewise.
2852 (XL_BE_V4SI): Likewise.
2853 (XL_BE_V2DI): Likewise.
2854 (XL_BE_V4SF): Likewise.
2855 (XL_BE_V2DF): Likewise.
2856 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
2857 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
2858 all array entries with these keys: VSX_BUILTIN_VEC_XL,
2859 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
2860 VSX_BUILTIN_VEC_XST_BE.
2861 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
2862 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
2864 (altivec_init_builtins): Replace conditional calls to def_builtin
2865 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
2866 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
2867 with unconditional calls. Remove calls to def_builtin for
2868 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
2869 __builtin_vec_xst_be.
2870 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
2871 to define_expand, and add alternate RTL generation for P8.
2872 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
2873 vsx_ld_elemrev_v8hi.
2874 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
2875 add alternate RTL generation for P8.
2876 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
2877 vsx_ld_elemrev_v16qi.
2878 (vsx_st_elemrev_v8hi): Convert define_insn
2879 to define_expand, and add alternate RTL generation for P8.
2880 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
2881 vsx_st_elemrev_v8hi.
2882 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
2883 add alternate RTL generation for P8.
2884 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
2885 vsx_st_elemrev_v16qi.
2887 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
2890 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
2891 TARGET_AVX512ER check.
2892 (ix86_option_override_internal): Set MASK_VZEROUPPER if
2893 neither -mzeroupper nor -mno-zeroupper is used and
2894 TARGET_EMIT_VZEROUPPER is set.
2895 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
2896 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
2898 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2900 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2901 folding of vector compares.
2902 (fold_build_vec_cmp): New helper function.
2903 (fold_compare_helper): New helper function.
2904 (builtin_function_type): Add compare builtins to the list of functions
2905 having unsigned arguments. Cosmetic updates to comment indentation.
2906 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
2907 the not+eq combination.
2909 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2911 PR tree-optimization/82726
2912 PR tree-optimization/70754
2913 * tree-predcom.c (order_drefs_by_pos): New function.
2914 (combine_chains): Move code setting has_max_use_after to...
2915 (try_combine_chains): ...here. New parameter. Sort combined chains
2916 according to position information.
2917 (tree_predictive_commoning_loop): Update call to above function.
2918 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
2920 2017-11-15 Bin Cheng <bin.cheng@arm.com>
2922 PR tree-optimization/82726
2924 2017-01-23 Bin Cheng <bin.cheng@arm.com>
2926 PR tree-optimization/70754
2927 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
2928 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
2929 combined stmt before it if not NULL.
2930 (combine_chains): Process refs reversely and compute dominance point
2934 2017-02-23 Bin Cheng <bin.cheng@arm.com>
2936 PR tree-optimization/79663
2937 * tree-predcom.c (combine_chains): Process refs in reverse order
2938 only for ZERO length chains, and add explaining comment.
2940 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2942 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
2943 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
2944 * doc/invoke.texi (ARM Options): Add armv8.3-a.
2946 2017-11-15 Tamar Christina <tamar.christina@arm.com>
2948 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
2950 2017-11-15 Martin Liska <mliska@suse.cz>
2952 * tree-cfg.c (pass_warn_function_return::execute):
2953 Compare warn_return_type for greater than zero.
2955 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
2959 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
2960 to return true on Xeon and not on Xeon Phi.
2961 (ix86_check_avx256_register): Changed to ...
2962 (ix86_check_avx_upper_register): ... this. Add extra check for
2963 VALID_AVX512F_REG_OR_XI_MODE.
2964 (ix86_avx_u128_mode_needed): Changed
2965 ix86_check_avx256_register to ix86_check_avx_upper_register.
2966 (ix86_check_avx256_stores): Changed to ...
2967 (ix86_check_avx_upper_stores): ... this. Changed
2968 ix86_check_avx256_register to ix86_check_avx_upper_register.
2969 (ix86_avx_u128_mode_after): Changed
2970 avx_reg256_found to avx_upper_reg_found. Changed
2971 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
2972 (ix86_avx_u128_mode_entry): Changed
2973 ix86_check_avx256_register to ix86_check_avx_upper_register.
2974 (ix86_avx_u128_mode_exit): Ditto.
2975 * config/i386/i386.h: (host_detect_local_cpu): New define.
2977 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2979 * config/arm/xgene1.md (xgene1): Split into automatons
2980 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
2981 (xgene1_f_load): Adjust reservations and/or types.
2982 (xgene1_f_store): Likewise.
2983 (xgene1_load_pair): Likewise.
2984 (xgene1_store_pair): Likewise.
2985 (xgene1_fp_load1): Likewise.
2986 (xgene1_load1): Likewise.
2987 (xgene1_store1): Likewise.
2988 (xgene1_move): Likewise.
2989 (xgene1_alu): Likewise.
2990 (xgene1_simd): Likewise.
2991 (xgene1_bfm): Likewise.
2992 (xgene1_neon_load1): Likewise.
2993 (xgene1_neon_store1): Likewise.
2994 (xgene1_neon_logic): Likewise.
2995 (xgene1_neon_st1): Likewise.
2996 (xgene1_neon_ld1r): Likewise.
2997 (xgene1_alu_cond): Added.
2998 (xgene1_shift_reg): Likwise.
2999 (xgene1_bfx): Likewise.
3000 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
3002 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3005 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
3007 (can_widen_mult_without_libcall): New function.
3008 (expand_mul_overflow): If only checking unsigned mul overflow,
3009 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
3010 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
3011 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
3012 (expand_DIVMOD): Formatting fix.
3013 * expmed.h (expand_mult): Add NO_LIBCALL argument.
3014 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
3015 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
3017 PR tree-optimization/82977
3018 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
3019 constructed temporary to strlen_to_stridx.put.
3021 2017-11-15 Martin Liska <mliska@suse.cz>
3023 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
3024 * configure: Regenerate.
3026 2017-11-15 Martin Liska <mliska@suse.cz>
3029 * config/sh/sh-mem.cc: Use proper probability for
3032 2017-11-14 Jeff Law <law@redhat.com>
3034 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
3035 the red zone for stack_clash_protection_final_dynamic_probe targets
3036 when the total dynamic stack size is zero bytes.
3038 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
3039 blocks is post order.
3041 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
3043 * dumpfile.h (TDF_COMPARE_DEBUG): New.
3044 * final.c (rest_of_clean_state): Set it for the
3045 -fcompare-debug dump.
3046 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
3047 class when TDF_COMPARE_DEBUG is set.
3049 * dwarf2out.c (gen_producer_string): Discard
3052 2017-11-15 Joseph Myers <joseph@codesourcery.com>
3055 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
3056 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
3057 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
3058 (__TGMATH_REAL_2_3): Remove macros.
3059 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
3060 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
3062 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
3064 (remquo): Define using __TGMATH_REAL_3.
3066 2017-11-14 Jeff Law <law@redhat.com>
3068 * vr-values.c: New file with contents extracted from tree-vrp.c.
3069 * Makefile.in (OBJS): Add vr-values.o
3070 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
3071 (set_value_range, set_and_canonicalize_value_range): Likewise.
3072 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
3073 (value_range_constant_singleton, symbolic_range_p): Likewise.
3074 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
3075 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
3076 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
3077 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
3078 (find_case_label_range, find_case_label_index): Likewise.
3079 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
3080 (range_int_cst_singleton_p, value_inside_range): Likewise.
3081 (get_single_symbol): Likewise.
3082 (switch_update): Move structure definition here.
3083 (to_remove_edges, to_update_switch_stmts): Provide externs.
3084 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
3085 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
3086 (vrp_val_is_min, set_value_range): Likewise.
3087 (set_and_canonicalize_value_range, copy_value_range): Likewise.
3088 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
3089 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
3090 (range_is_nonnull, range_int_cst_p): Likewwise.
3091 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
3092 (get_single_symbol, operand_less_p): Likewise
3093 (compare_values_warnv, compare_values): Likewise.
3094 (value_inside_range, value_range_constant_singleton): Likewise.
3095 (zero_nonzero_bitgs_from_vr): Likewise.
3096 (extract_range_from_binary_expr_1): Likewise.
3097 (overflow_comparison_p): Likewise.
3098 (to_remove_edges, to_update_switch_stmts): Likewise.
3099 (find_case_label-index, find_case_label_range): Likewise.
3100 (switch_update, set_value_range_to_nonnegative): Remove.
3101 (set_value_range_to_truthvalue): Likewise.
3102 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
3103 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
3104 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
3105 (find_case_label_ranges, test_for_singularity): Likewise.
3106 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
3107 (x_vr_values): Move to its remaining use site.
3109 2017-11-10 Jeff Law <law@redhat.com>
3111 * vr-values.h (VR_INITIALIZER): Move #define here.
3112 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
3113 * Makefile.in (OBJS): Add tree-evrp.o
3114 * tree-vrp.h (assert_info): Move structure definition here.
3115 (set_value_range_to_varying): Prototype.
3116 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3117 (infer_value_range, register_edge_assert_for): Likewise.
3118 (stmt_interesting_for_vrp): Likewise.
3119 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
3120 (set_value_range_to_varying): No longer static.
3121 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
3122 (infer_value_range, register_edge_assert_for): Likewise.
3124 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3126 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
3127 generate the XXBRD instruction.
3129 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
3130 (rs6000_builtin_type_compatible): Treat _Float128 and long double
3131 as being compatible if -mabi=ieeelongdouble.
3132 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
3133 to setup float128 built-ins with hardware support.
3134 (BU_FLOAT128_HW_2): Likewise.
3135 (BU_FLOAT128_HW_3): Likewise.
3136 (BU_FLOAT128_HW_VSX_1): Likewise.
3137 (BU_FLOAT128_HW_VSX_2): Likewise.
3138 (scalar_extract_expq): Change float128 built-in functions to
3139 accommodate having both KFmode and TFmode functions. Use the
3140 KFmode variant as the default.
3141 (scalar_extract_sigq): Likewise.
3142 (scalar_test_neg_qp): Likewise.
3143 (scalar_insert_exp_q): Likewise.
3144 (scalar_insert_exp_qp): Likewise.
3145 (scalar_test_data_class_qp): Likewise.
3146 (sqrtf128_round_to_odd): Delete processing the round to odd
3147 built-in functions as special built-in functions, and define them
3148 as float128 built-ins. Use the KFmode variant as the default.
3149 (truncf128_round_to_odd): Likewise.
3150 (addf128_round_to_odd): Likewise.
3151 (subf128_round_to_odd): Likewise.
3152 (mulf128_round_to_odd): Likewise.
3153 (divf128_round_to_odd): Likewise.
3154 (fmaf128_round_to_odd): Likewise.
3155 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
3156 support for KFmode and TFmode xststdcqp calls.
3157 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
3158 point, switch the built-in handlers for the get/set float128
3159 exponent, get float128 mantissa, float128 test built-ins, and the
3160 float128 round to odd built-in functions. Eliminate creating the
3161 float128 round to odd built-in functions as special built-ins.
3162 (rs6000_init_builtins): Eliminate special creation of the float128
3163 round to odd built-in functions.
3164 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
3165 function insns to support both TFmode and KFmode variants.
3166 (xsxsigqp_<mode>): Likewise.
3167 (xsiexpqpf_<mode>): Likewise.
3168 (xsiexpqp_<mode>): Likewise.
3169 (xststdcqp_<mode>): Likewise.
3170 (xststdcnegqp_<mode>): Likewise.
3171 (xststdcqp_<mode>): Likewise.
3173 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3175 * tree-ssa-threadupdate.c (compute_path_counts): Remove
3176 unused path_in_freq_ptr parameter.
3177 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
3179 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3181 * ipa-inline.c (edge_badness): Dump sreal frequency.
3182 (compute_inlined_call_time): Match natural implementaiton ...
3183 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
3184 forgotten division by CGRAPH_FREQ_BASE.
3186 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3188 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
3189 Solaris 11. Update comment.
3190 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
3192 * config/sol2.h (STARTFILE_SPEC): Likewise.
3193 * configure: Regenerate.
3195 2017-11-14 Carl Love <cel@us.ibm.com>
3197 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
3198 le_ and be_ prefixes to swap* variables. Remove
3199 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
3202 2017-11-14 Jason Merrill <jason@redhat.com>
3204 Support GTY((cache)) on hash_map.
3205 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
3206 (ggc_cache_remove): Override it instead of ggc_mx.
3207 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
3208 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
3209 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
3210 (simple_cache_map_traits): Override maybe_mx.
3211 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
3212 (hash_map): Friend gt_cleare_cache.
3213 (gt_cleare_cache): New.
3214 * tree.h (tree_cache_traits): New hash_map traits class.
3215 (tree_cache_map): New typedef.
3217 2017-11-14 Richard Biener <rguenther@suse.de>
3219 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
3220 paramter and handling.
3221 (cleanup_control_flow_bb): Likewise.
3222 (cleanup_control_flow_pre): New helper performing a DFS walk
3223 to call cleanup_control_flow_bb in PRE order.
3224 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
3225 via cleanup_control_flow_pre.
3227 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
3229 * config/aarch64/aarch64-simd.md
3230 (aarch64_simd_bsl<mode>_internal): Remove DImode.
3231 (*aarch64_simd_bsl<mode>_alt): Likewise.
3232 (aarch64_simd_bsldi_internal): New.
3233 (aarch64_simd_bsldi_alt): Likewise.
3235 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3237 * tracer.c (better_p): Do not compare frequencies.
3238 * reg-stack.c (better_edge): Likewise.
3239 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
3242 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3244 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
3245 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
3246 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
3247 * lto-streamer-in.c (input_function): Use update_max_bb_count.
3248 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
3249 * predict.c (maybe_hot_frequency_p): Inline to ...
3250 (maybe_hot_count_p): ... here; rewrite to counts.
3251 (counts_to_freqs): Rename to ...
3252 (update_max_bb_count): ... this one.
3253 (expensive_function_p): Use counts.
3254 (estimate_bb_frequencies): Update.
3255 (rebuild_frequencies): Update.
3256 * predict.h (counts_to_freqs): Rename to ...
3257 (update_max_bb_count): ... this one.
3258 * profile.c (compute_branch_probabilities): Add debug info
3259 * tree-inline.c (expand_call_inline): Update debug info.
3260 (optimize_inline_calls): Use update_max_bb_count..
3261 (tree_function_versioning): Use update_max_bb_count..
3262 * value-prof.c (gimple_value_profile_transformations):
3263 Do not use update_max_bb_count.
3265 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3267 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
3268 always use frequencies.
3270 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3272 * bb-reorder.c: Remove frequencies from comments.
3273 (better_edge_p): Use profile counts.
3274 (find_traces): Dump profile counts.
3275 (rotate_loop): Use profile counts.
3276 (find_traces_1_round): Likewise.
3277 (connect_better_edge_p): Use counts instead of probabilities for
3279 (copy_bb_p): Drop early check for non-0 frequency.
3280 (sanitize_hot_paths): Update comments.
3282 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3284 * ipa-split.c (struct split_point): Add count.
3285 (consider_split): Do not compute incoming frequency; compute incoming
3286 count and store it to split_point.
3287 (split_function): Set count of the call to split part correctly.
3289 2017-11-13 Carl Love <cel@us.ibm.com>
3291 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
3293 2017-11-13 Tom Tromey <tom@tromey.com>
3295 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
3297 2017-11-13 Carl Love <cel@us.ibm.com>
3299 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
3300 Add support for builtins:
3301 unsigned int vec_first_{,miss}_match_{,or_eos}index,
3302 vector {un,}signed {char,int,short},
3303 vector {un,}signed {char,int,short}) arguments.
3304 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
3305 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
3306 Add BU_P9V_AV_2 expansions for the builtins.
3307 * config/rs6000/altivec.h (vec_first_match_index,
3308 vec_first_mismatch_index, vec_first_match_or_eos_index,
3309 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
3310 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
3311 new extern declaration.
3312 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
3313 * config/rs6000/vsx.md (first_match_index_<mode>,
3314 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
3315 first_mismatch_or_eos_index_<mode>): Add define expand.
3316 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
3317 * doc/extend.texi: Update the built-in documenation file for the new
3320 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
3322 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
3323 into the min/max operations for _Float<N> and _Float<N>X types.
3325 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3328 * dwarf2out.c (do_eh_frame): New static variable.
3329 (dwarf2out_begin_prologue): Set it.
3330 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
3332 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3334 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
3336 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
3338 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
3341 * debug.h (dwarf2out_do_eh_frame): Declare.
3342 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
3343 (dwarf2out_do_frame): Use it.
3344 (dwarf2out_do_cfi_asm): Likewise.
3345 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
3346 (dwarf2out_assembly_start): Likewise.
3347 (dwarf2out_begin_prologue): Fix comment.
3348 * toplev.c (compile_file): Always call dwarf2out_frame_finish
3349 if the target needs either debug or unwind DWARF2 info.
3350 * lto-opts.c (lto_write_options): Do not save -fexceptions,
3351 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
3352 -ftrapping-math, -ftrapv and -fwrapv.
3354 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
3356 * cgraph.c (cgraph_edge::sreal_frequency): New function.
3357 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
3358 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
3359 (estimate_edge_size_and_time): Likewise.
3360 (ipa_merge_fn_summary_after_inlining): Likewise.
3361 * ipa-inline.c (cgraph_freq_base_rec): Remove.
3362 (compute_uninlined_call_time): Use sreal_frequency.
3363 (compute_inlined_call_time): Likewise.
3364 (ipa_inline): Do not initialize cgraph_freq_base_rec.
3365 * profile-count.c: Include sreal.h.
3366 (profile_count::to_sreal_scale): New.
3367 * profile-count.h: Forward declare sreal.
3368 (profile_count::to_sreal_scale): Declare.
3370 2017-11-13 Nathan Sidwell <nathan@acm.org>
3372 * diagnostic.c (maybe_line_and_column): New.
3373 (diagnostic_get_location_text): Use it.
3374 (diagnostic_report_current_module): Likewise.
3375 (test_diagnostic_get_location_text): Add tests.
3377 2017-11-13 Luis Machado <luis.machado@linaro.org>
3379 * doc/md.texi (Specifying processor pipeline description): Fix
3380 incorrect latency for the div instruction example.
3382 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3384 PR tree-optimization/78821
3385 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
3386 that bit_not_p is the same.
3387 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
3388 (split_group): Count precisely bit_not_p bits in each statement.
3389 (invert_op): New function.
3390 (imm_store_chain_info::output_merged_store): Use invert_op to
3391 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
3392 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
3394 2017-11-13 Martin Liska <mliska@suse.cz>
3396 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
3397 (struct source_info): Likewise.
3398 (add_branch_counts): Likewise.
3399 (add_line_counts): Likewise.
3400 (function_summary): Likewise.
3401 (output_intermediate_line): Likewise.
3402 (generate_results): Likewise.
3404 2017-11-13 Martin Liska <mliska@suse.cz>
3406 * gcov.c (struct block_info): Remove typedef for block_t.
3407 (struct line_info): Likewise.
3408 (line_info::has_block): Likewise.
3409 (EXIT_BLOCK): Likewise.
3410 (unblock): Likewise.
3411 (circuit): Likewise.
3412 (get_cycles_count): Likewise.
3413 (process_file): Likewise.
3414 (read_graph_file): Likewise.
3415 (solve_flow_graph): Likewise.
3416 (find_exception_blocks): Likewise.
3417 (add_line_counts): Likewise.
3418 (accumulate_line_info): Likewise.
3419 (output_line_details): Likewise.
3421 2017-11-13 Martin Liska <mliska@suse.cz>
3423 * gcov.c (struct arc_info): Remove typedef for arc_t.
3424 (struct line_info): Likewise.
3425 (add_branch_counts): Likewise.
3426 (output_branch_count): Likewise.
3427 (function_info::~function_info): Likewise.
3428 (circuit): Likewise.
3429 (output_intermediate_line): Likewise.
3430 (read_graph_file): Likewise.
3431 (solve_flow_graph): Likewise.
3432 (find_exception_blocks): Likewise.
3433 (add_line_counts): Likewise.
3434 (accumulate_line_info): Likewise.
3435 (output_line_details): Likewise.
3436 (output_function_details): Likewise.
3438 2017-11-13 Martin Liska <mliska@suse.cz>
3440 * gcov.c (struct function_info): Remove typedef for function_t.
3441 (struct source_info): Likewise.
3442 (source_info::get_functions_at_location): Likewise.
3443 (solve_flow_graph): Likewise.
3444 (find_exception_blocks): Likewise.
3445 (add_line_counts): Likewise.
3446 (output_intermediate_file): Likewise.
3447 (process_file): Likewise.
3448 (generate_results): Likewise.
3449 (release_structures): Likewise.
3450 (read_graph_file): Likewise.
3451 (read_count_file): Likewise.
3452 (accumulate_line_counts): Likewise.
3453 (output_lines): Likewise.
3455 2017-11-13 Martin Liska <mliska@suse.cz>
3457 * gcov.c (function_info::function_info): Remove num_counts
3458 and add vector<gcov_type>.
3459 (function_info::~function_info): Use the vector.
3460 (process_file): Likewise.
3461 (read_graph_file): Likewise.
3462 (read_count_file): Likewise.
3463 (solve_flow_graph): Likewise.
3465 2017-11-13 Martin Liska <mliska@suse.cz>
3467 * gcov.c (function_info::is_artificial): New function.
3468 (process_file): Erase all artificial early.
3469 (generate_results): Skip as all artificial are already
3472 2017-11-13 Martin Liska <mliska@suse.cz>
3474 * gcov.c (read_graph_file): Store to global vector of functions.
3475 (read_count_file): Iterate the vector.
3476 (process_file): Likewise.
3477 (generate_results): Likewise.
3478 (release_structures): Likewise.
3480 2017-11-13 Jakub Jelinek <jakub@redhat.com>
3482 PR tree-optimization/82954
3483 * gimple-ssa-store-merging.c
3484 (imm_store_chain_info::coalesce_immediate_stores): If
3485 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
3487 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
3489 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
3490 Upddate call to ENDIAN_LANE_N.
3491 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
3492 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
3493 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
3494 and use aarch64_endian_lane_rtx.
3495 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
3497 2017-11-12 Tom de Vries <tom@codesourcery.com>
3499 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
3501 2017-11-12 Tom de Vries <tom@codesourcery.com>
3503 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
3505 2017-11-12 Tom de Vries <tom@codesourcery.com>
3507 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
3509 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3510 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
3511 * defaults.h (ASM_OUTPUT_LABELREF): Same.
3513 2017-11-11 Martin Sebor <msebor@redhat.com>
3516 * doc/extend.texi (attribute nonstring): Remove spurious argument.
3519 * prefic.c (translate_name): Replace strncpy with memcpy to
3520 avoid -Wstringop-truncation.
3522 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3524 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
3526 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3528 * predict.c (maybe_hot_frequency_p): Do not use cfun.
3530 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3532 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
3535 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
3537 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
3538 enable generating XXBRH if the value is in a vector register.
3539 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
3540 value is in a vector register.
3541 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
3542 register to register bswap64's instead of doing the GPR sequence
3543 used on previous machines.
3544 (bswapdi2_xxbrd): New insn.
3545 (bswapdi2_reg): Disallow on ISA 3.0.
3546 (register to register bswap64 splitter): Do not split the insn on
3547 ISA 3.0 systems that use XXBRD.
3549 2017-11-10 Martin Sebor <msebor@redhat.com>
3552 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
3554 (find_subframework_file): Same.
3556 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3558 * auto-profile.c (afdo_indirect_call): Drop frequency.
3559 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
3560 (cgraph_node::create_edge): Drop frequency argument.
3561 (cgraph_node::create_indirect_edge): Drop frequency argument.
3562 (cgraph_edge::make_speculative): Drop frequency arguments.
3563 (cgraph_edge::resolve_speculation): Do not update frequencies
3564 (cgraph_edge::dump_edge_flags): Do not dump frequency.
3565 (cgraph_node::dump): Check consistency in IPA mode.
3566 (cgraph_edge::maybe_hot_p): Use IPA counter.
3567 (cgraph_edge::verify_count_and_frequency): Rename to ...
3568 (cgraph_edge::verify_count): ... this one; drop frequency checking.
3569 (cgraph_node::verify_node): Update.
3570 * cgraph.h (struct cgraph_edge): Drop frequency.
3571 (cgraph_edge::frequency): New function.
3572 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
3574 (cgraph_edge::rebuild_edges): Likewise.
3575 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
3576 (duplicate_thunk_for_node): Do not pass frequency.
3577 (cgraph_node::create_clone): Scale only counts.
3578 (cgraph_node::create_virtual_clone): Do not pass frequency.
3579 (cgraph_node::create_edge_including_clones): Do not pass frequency.
3580 (cgraph_node::create_version_clone): Do not pass frequency.
3581 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
3582 (cgraph_node::expand_thunk): Do not pass frequency.
3583 (cgraph_node::create_wrapper): Do not pass frequency.
3584 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
3586 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
3587 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
3588 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
3589 (gather_caller_stats): Use frequency function.
3590 (ipcp_cloning_candidate_p): Use frequency function.
3591 (ipcp_propagate_stage): Use frequency function.
3592 (get_info_about_necessary_edges): Use frequency function.
3593 (update_profiling_info): Update only IPA profile.
3594 (update_specialized_profile): Use frequency functoin.
3595 (perhaps_add_new_callers): Update only IPA profile.
3596 * ipa-devirt.c (ipa_devirt): Use IPA profile.
3597 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
3598 (dump_ipa_call_summary): Use frequency function.
3599 (estimate_edge_size_and_time): Use frequency function.
3600 (ipa_merge_fn_summary_after_inlining): Use frequency function.
3601 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
3602 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
3603 (update_noncloned_counts): ... ths one; scale counts only.
3604 (clone_inlined_nodes): Do not scale frequency.
3605 (inline_call): Do not pass frequency.
3606 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
3607 (compute_inlined_call_time): Use IPA profile.
3608 (want_inline_small_function_p): Use IPA profile.
3609 (want_inline_self_recursive_call_p): Use IPA profile.
3610 (edge_badness): Use IPA profile.
3611 (lookup_recursive_calls): Use IPA profile.
3612 (recursive_inlining): Do not pass frequency.
3613 (resolve_noninline_speculation): Do not update frequency.
3614 (inline_small_functions): Collect max of IPA profile.
3615 (dump_overall_stats): Dump IPA porfile.
3616 (dump_inline_stats): Dump IPA porfile.
3617 (ipa_inline): Collect IPA stats.
3618 * ipa-inline.h (clone_inlined_nodes): Update prototype.
3619 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
3620 (ipa_propagate_frequency): Use frequency function.
3621 (ipa_profile): Cleanup.
3622 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
3623 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
3624 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
3625 (input_node): Do not stream frequency.
3626 (input_edge): Do not stream frequency.
3627 (merge_profile_summaries): Scale only IPA profiles.
3628 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
3629 * predict.c (drop_profile): Do not recompute frequency.
3630 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
3631 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
3632 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
3633 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
3634 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
3635 * tree-inline.c (copy_bb): Do not scale frequency.
3636 (expand_call_inline): Do not scale frequency.
3637 (tree_function_versioning): Do not scale frequency.
3638 * ubsan.c (ubsan_create_edge): Do not pass frequency.
3640 2017-11-10 Julia Koval <julia.koval@intel.com>
3642 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
3643 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
3644 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
3645 (_mm256_maskz_gf2p8affine_epi64_epi8)
3646 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
3647 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
3648 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
3649 (__builtin_ia32_vgf2p8affineqb_v32qi)
3650 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
3651 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
3653 2017-11-10 Tamar Christina <tamar.christina@arm.com>
3657 (arm_option_override): Refactor.
3658 (arm_option_reconfigure_globals): New.
3659 (arm_options_perform_arch_sanity_checks): New.
3660 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
3662 (arm_options_perform_arch_sanity_checks): Likewise
3664 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
3666 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
3667 (power9-qpmul): New.
3668 * rs6000/rs6000.md ("type" attr): Add qmul.
3669 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
3670 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
3671 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
3673 2017-11-10 Martin Sebor <msebor@redhat.com>
3676 * builtins.c (compute_objsize): Handle arrays that
3677 compute_builtin_object_size likes to fail for. Make extern.
3678 * builtins.h (compute_objsize): Declare.
3679 (check_strncpy_sizes): New function.
3680 (expand_builtin_strncpy): Call check_strncpy_sizes.
3681 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
3682 -Wstringop-truncation.
3683 (gimple_fold_builtin_strncat): Same.
3684 * gimple.c (gimple_build_call_from_tree): Set call location.
3685 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
3686 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
3687 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
3688 (handle_builtin_strlen): Use strlen_to_stridx.
3689 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
3691 Use strlen_to_stridx.
3692 (pass_strlen::execute): Release strlen_to_stridx.
3693 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
3694 (-Wstringop-truncation): Document new option.
3696 2017-11-10 Martin Liska <mliska@suse.cz>
3698 PR gcov-profile/82702
3699 * gcov.c (main): Handle intermediate files in a different
3701 (get_gcov_intermediate_filename): New function.
3702 (output_gcov_file): Remove support of intermediate files.
3703 (generate_results): Allocate intermediate file.
3704 (release_structures): Clean-up properly fn_end.
3705 (output_intermediate_file): Start iterating with line 1.
3707 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3709 PR tree-optimization/82929
3710 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3711 ops_swapped_p non-static data member.
3712 (store_immediate_info::store_immediate_info): Clear it.
3713 (imm_store_chain_info::coalesce_immediate_stores): If swapping
3714 ops set ops_swapped_p.
3715 (count_multiple_uses): Handle ops_swapped_p.
3717 2017-11-10 Martin Liska <mliska@suse.cz>
3719 * coverage.c (coverage_init): Stream information about
3720 support of has_unexecuted_blocks.
3721 * doc/gcov.texi: Document that.
3722 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
3723 * gcov.c (read_graph_file): Likewise.
3724 (output_line_beginning): Fix a small issue with
3727 2017-11-10 Bin Cheng <bin.cheng@arm.com>
3729 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
3730 reference of trivial component.
3732 2017-11-10 Jakub Jelinek <jakub@redhat.com>
3735 * gimple-ssa-store-merging.c
3736 (pass_store_merging::terminate_all_aliasing_chains): For
3737 gimple_store_p stmts also call refs_output_dependent_p.
3739 PR rtl-optimization/82913
3740 * compare-elim.c (try_merge_compare): Punt if def_insn is not
3743 2017-11-09 Jeff Law <law@redhat.com>
3745 * vr-values.h: New file with vr_values class.
3746 * tree-vrp.c: Include vr-values.h
3747 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
3748 data objects into the vr_values class.
3749 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
3750 (get_value_range): Make it a member function within vr_values class.
3751 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
3752 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
3753 (op_with_constant_singleton_value_range): Likewise.
3754 (extract_range_for_var_from_comparison_expr): Likewise.
3755 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
3756 (extract_range_from_binary_expr): Likewise.
3757 (extract_range_from_unary_expr): Likewise.
3758 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
3759 (check_for_binary_op_overflow, extract_range_basic): Likewise.
3760 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
3761 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
3762 (compare_name_with_value, compare_names): Likewise.
3763 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
3764 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
3765 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
3766 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
3767 (extract_range_from_phi_node): Likewise.
3768 (simplify_truth_ops_using_ranges): Likewise.
3769 (simplify_div_or_mod_using_ranges): Likewise.
3770 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
3771 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
3772 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
3773 (simplify_float_conversion_using_ranges): Likewise.
3774 (simplify_internal_call_using_ranges): Likewise.
3775 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
3776 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
3777 poitner via x_vr_values for calls into gimple folder.
3778 (vrp_initialize_lattice): Make this the vr_values ctor.
3779 (vrp_free_lattice): Make this the vr_values dtor.
3780 (set_vr_value): New function.
3781 (class vrp_prop): Add vr_values data member. Add various member
3782 functions as well as member functions that delegate to vr_values.
3783 (check_array_ref): Make a member function within vrp_prop class.
3784 (search_for_addr_array, vrp_initialize): Likewise.
3785 (vrp_finalize): Likewise. Revamp to avoid direct access to
3786 vr_value, values_propagated, etc.
3787 (check_array_bounds): Extract vrp_prop class instance pointer from
3788 walk info structure. Use it to call member functions.
3789 (check_all_array_refs): Make a member function within vrp_prop class.
3790 Smuggle class instance pointer via walk info structure.
3791 (x_vr_values): New local static.
3792 (vrp_valueize): Use x_vr_values to get class instance.
3793 (vr_valueize_1): Likewise.
3794 (class vrp_folder): Add vr_values data member. Add various member
3795 functions as well as member functions that delegate to vr_values.
3796 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
3797 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
3798 class instance from vr_values. Use it to call member functions.
3799 (vrp_dom_walker): Add vr_values data member.
3800 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
3801 instance via x_vr_values.
3802 (identify_jump_threads): Accept vr_values as argument. Store
3803 it into the walker structure.
3804 (evrp_dom_walker): Add vr_values class data member. Add various
3806 (evrp_dom_walker::try_find_new_range): Use vr_values data
3807 member to access the memory allocator.
3808 (evrp_dom_walker::before_dom_children): Store vr_values class
3809 instance into the vrp_folder class.
3810 (evrp_dom_walker::push_value_range): Rework to avoid direct
3811 access to num_vr_values and vr_value.
3812 (evrp_dom_walker::pop_value_range): Likewise.
3813 (execute_early_vrp): Remove call to vrp_initialize_lattice.
3814 Use vr_values to get to dump_all_value_ranges member function.
3815 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
3816 and simplify_cond_using_ranges_2 via vrp_prop class instance.
3817 Pass vr_values class instance down to identify_jump_threads.
3818 Remove call to vrp_free_lattice.
3819 (debug_all_value_ranges): Remove.
3821 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
3822 (vrp_folder): Likewise.
3824 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
3825 Get it from the existing bitmap instead.
3826 (vrp_intersect_ranges_1): Likewise.
3828 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3830 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
3832 (store_immediate_info::store_immediate_info): Add bitnotp argument,
3833 set bit_not_p to it.
3834 (imm_store_chain_info::coalesce_immediate_stores): Break group
3835 if bit_not_p is different.
3836 (count_multiple_uses, split_group,
3837 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
3838 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
3839 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
3840 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
3841 to store_immediate_info ctor.
3843 2017-11-09 Jim Wilson <jimw@sifive.com>
3845 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
3846 (scan_prog_file): Likewise.
3848 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3850 * bb-reorder.c (max_entry_frequency): Remove.
3851 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
3852 connect_traces, push_to_next_round_p): Remove prototypes.
3853 (find_traces_1_round): Use counts only.
3854 (push_to_next_round_p): Likewise.
3855 (find_traces): Likewise.
3856 (rotate_loop): Likewise.
3857 (find_traces_1_round): Likewise.
3858 (connect_traces): Likewise.
3859 (edge_order): Likewise.
3861 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
3863 * config/arm/arm.c (output_return_instruction): Add comments to
3864 indicate requirement for cmse_nonsecure_entry return to account
3865 for the size of clearing instruction output here.
3866 (thumb_exit): Likewise.
3867 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
3868 return in hardfloat mode.
3870 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
3872 * config/rs6000/rs6000.c (machine_function): Add a bool,
3873 "toc_is_wrapped_separately".
3874 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
3875 if it wasn't explicitly set or unset, we are optimizing for speed, and
3876 doing separate shrink-wrapping.
3877 (rs6000_get_separate_components): Enable the TOC component if
3878 saving the TOC register in the prologue.
3879 (rs6000_components_for_bb): Handle the TOC component.
3880 (rs6000_emit_prologue_components): Store the TOC register where needed.
3881 (rs6000_set_handled_components): Mark TOC as handled, if handled.
3882 (rs6000_emit_prologue): Don't save the TOC if that is already done.
3884 2017-11-09 Martin Jambor <mjambor@suse.cz>
3886 * ipa-param-manipulation.c: New file.
3887 * ipa-param-manipulation.h: Likewise.
3888 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
3889 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
3890 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
3891 (ipa_parm_adjustment): Likewise.
3892 (ipa_parm_adjustment_vec): Likewise.
3893 (ipa_get_vector_of_formal_parms): Moved declaration to
3894 ipa-param-manipulation.h.
3895 (ipa_get_vector_of_formal_parm_types): Likewise.
3896 (ipa_modify_formal_parameters): Likewise.
3897 (ipa_modify_call_arguments): Likewise.
3898 (ipa_combine_adjustments): Likewise.
3899 (ipa_dump_param_adjustments): Likewise.
3900 (ipa_modify_expr): Likewise.
3901 (ipa_get_adjustment_candidate): Likewise.
3902 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
3903 ipa-param-manipulation.c.
3904 (ipa_get_vector_of_formal_parm_types): Likewise.
3905 (ipa_modify_formal_parameters): Likewise.
3906 (ipa_modify_call_arguments): Likewise.
3907 (ipa_modify_expr): Likewise.
3908 (get_ssa_base_param): Likewise.
3909 (ipa_get_adjustment_candidate): Likewise.
3910 (index_in_adjustments_multiple_times_p): Likewise.
3911 (ipa_combine_adjustments): Likewise.
3912 (ipa_dump_param_adjustments): Likewise.
3913 * tree-sra.c: Also include ipa-param-manipulation.h
3914 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
3917 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3918 Alan Hayward <alan.hayward@arm.com>
3919 David Sherwood <david.sherwood@arm.com>
3921 * doc/sourcebuild.texi (vect_masked_store): Document.
3923 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3924 Alan Hayward <alan.hayward@arm.com>
3925 David Sherwood <david.sherwood@arm.com>
3927 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
3929 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3930 Alan Hayward <alan.hayward@arm.com>
3931 David Sherwood <david.sherwood@arm.com>
3933 * doc/sourcebuild.texi (vect_variable_length): Document.
3935 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3936 Alan Hayward <alan.hayward@arm.com>
3937 David Sherwood <david.sherwood@arm.com>
3939 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
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 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
3947 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3948 Alan Hayward <alan.hayward@arm.com>
3949 David Sherwood <david.sherwood@arm.com>
3951 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
3952 previously undocumented selectors.
3953 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
3955 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3957 * doc/rtl.texi (const_vector): Say that elements can be
3958 const_wide_ints too.
3959 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
3960 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
3961 (gen_vec_duplicate): Use it instead of CONSTANT_P.
3962 * optabs.c (expand_vector_broadcast): Likewise.
3964 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3965 Alan Hayward <alan.hayward@arm.com>
3966 David Sherwood <david.sherwood@arm.com>
3968 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
3969 scaled index even if the unscaled address was invalid.
3970 Don't increase the complexity of using a scale in that case.
3972 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
3973 Alan Hayward <alan.hayward@arm.com>
3974 David Sherwood <david.sherwood@arm.com>
3976 * doc/rtl.texi: Rewrite the subreg rules so that they partition
3977 the inner register into REGMODE_NATURAL_SIZE bytes rather than
3978 UNITS_PER_WORD bytes.
3979 * emit-rtl.c (validate_subreg): Divide subregs into blocks
3980 based on REGMODE_NATURAL_SIZE of the inner mode.
3981 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
3982 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
3983 * expmed.c (lowpart_bit_field_p): Divide the value up into
3984 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
3985 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
3986 whether something is likely to occupy more than one register.
3988 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3991 * ipa-inline-transform.c (update_noncloned_frequencies): Use
3992 profile_count::adjust_for_ipa_scaling.
3993 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
3994 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
3996 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
3998 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4000 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4001 (split_group): Add total_orig and total_new arguments, estimate the
4002 number of statements related to the store group without store merging
4003 and with store merging.
4004 (imm_store_chain_info::output_merged_store): Adjust split_group
4005 callers, punt if estimated number of statements with store merging
4006 is not smaller than estimated number of statements without it.
4008 (handled_load): Remove has_single_use checks.
4009 (pass_store_merging::process_store): Likewise.
4011 2017-11-09 Richard Biener <rguenther@suse.de>
4013 PR tree-optimization/82902
4014 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
4016 2017-11-09 Martin Liska <mliska@suse.cz>
4019 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
4022 2017-11-09 Martin Liska <mliska@suse.cz>
4024 PR tree-optimization/82669
4025 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
4027 2017-11-09 Martin Liska <mliska@suse.cz>
4029 PR gcov-profile/48463
4030 * coverage.c (coverage_begin_function): Output also end locus
4031 of a function and information whether the function is
4033 * gcov-dump.c (tag_function): Parse and print the information.
4034 * gcov.c (INCLUDE_MAP): Add include.
4035 (INCLUDE_SET): Likewise.
4036 (struct line_info): Move earlier in the source file because
4037 of vector<line_info> in function_info structure.
4038 (line_info::line_info): Likewise.
4039 (line_info::has_block): Likewise.
4040 (struct source_info): Add new member index.
4041 (source_info::get_functions_at_location): New function.
4042 (function_info::group_line_p): New function.
4043 (output_intermediate_line): New function.
4044 (output_intermediate_file): Use the mentioned function.
4045 (struct function_start): New.
4046 (struct function_start_pair_hash): Likewise.
4047 (process_file): Add code that identifies group functions.
4048 Assign lines either to global or function scope.
4049 (generate_results): Skip artificial functions.
4050 (find_source): Assign index for each source file.
4051 (read_graph_file): Read new flag artificial and end_line.
4052 (add_line_counts): Assign it either to global of function scope.
4053 (accumulate_line_counts): Isolate core of the function to
4054 accumulate_line_info and call it for both function and global
4056 (accumulate_line_info): New function.
4057 (output_line_beginning): Fix GNU coding style.
4058 (print_source_line): New function.
4059 (output_line_details): Likewise.
4060 (output_function_details): Likewise.
4061 (output_lines): Iterate both source (global) scope and function
4063 (struct function_line_start_cmp): New class.
4064 * doc/gcov.texi: Reflect changes in documentation.
4066 2017-11-09 Jakub Jelinek <jakub@redhat.com>
4069 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
4070 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
4071 and similarly for not instead of neg.
4073 2017-11-08 Andi Kleen <ak@linux.intel.com>
4075 * config/i386/i386.opt: Add -mforce-indirect-call.
4076 * config/i386/predicates.md: Check for flag_force_indirect_call.
4077 * doc/invoke.texi: Document -mforce-indirect-call
4079 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
4081 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
4083 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
4084 config/riscv/riscv.c (predict.h): New include.
4085 (riscv_slow_unaligned_access_p): No longer static.
4086 (riscv_block_move_straight): Add require.
4087 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4089 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4092 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
4093 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
4094 nonimmediate_operand predicate for operand 1 instead of
4097 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4099 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
4101 * config/aarch64/constraints.md (Uml): New constraint.
4102 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
4105 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4107 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
4108 of two vec_duplicates into a vec_concat.
4110 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4112 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4113 Simplify vec_merge of vec_duplicate and vec_concat.
4114 * config/aarch64/constraints.md (Utq): New constraint.
4115 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
4118 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4120 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
4121 Simplify vec_merge of vec_duplicate and const_vector.
4122 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
4124 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
4125 mode iterator. Update predicate on operand 1 to
4126 handle non-const_vec constants. Delete constraints.
4127 (*aarch64_combinez_be<mode>): Likewise for operand 2.
4129 2017-11-08 Jakub Jelinek <jakub@redhat.com>
4131 PR tree-optimization/78821
4132 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
4134 (store_operand_info::store_operand_info): Initialize it to false.
4135 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
4136 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
4137 store in the group, and if chain_info is non-NULL, to ignore altogether
4139 (compatible_load_p): Fail if bit_not_p does not match.
4140 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
4141 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
4142 (pass_store_merging::process_store): Adjust
4143 terminate_all_aliasing_chains calls to pass NULL in all current spots,
4144 call terminate_all_aliasing_chains newly when adding a store into
4145 a chain with non-NULL chain_info.
4147 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4149 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
4151 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
4153 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4155 (aarch64_layout_frame): Initialise emit_frame_chain.
4156 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
4157 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
4159 2017-11-08 Martin Liska <mliska@suse.cz>
4161 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
4162 of gimple_call_internal_p.
4164 2017-11-07 Tom de Vries <tom@codesourcery.com>
4166 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
4168 2017-11-07 Tom de Vries <tom@codesourcery.com>
4170 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4173 2017-11-08 Martin Liska <mliska@suse.cz>
4176 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
4178 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4180 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
4182 2017-11-07 Jakub Jelinek <jakub@redhat.com>
4185 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
4186 (*cmp<mode>_ccz_1): New insn with $k alternative.
4189 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
4190 RTX_COMM_COMPARE as commutative as well.
4191 (ix86_binary_operator_ok): Formatting fix.
4192 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
4193 *<code><mode>3<mask_name><round_saeonly_name>,
4194 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
4195 *<s>mul<mode>3_highpart<mask_name>,
4196 *vec_widen_umult_even_v16si<mask_name>,
4197 *vec_widen_umult_even_v8si<mask_name>,
4198 *vec_widen_umult_even_v4si<mask_name>,
4199 *vec_widen_smult_even_v16si<mask_name>,
4200 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
4201 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
4202 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
4203 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
4204 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
4205 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
4206 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
4207 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
4208 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
4209 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
4210 ix86_binary_operator_ok. Formatting fixes.
4211 (*<plusminus_insn><mode>3<mask_name><round_name>,
4212 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
4215 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
4217 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
4218 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
4219 eq and ne if TARGET_ISEL.
4220 (cmp): New code_iterator.
4221 (UNS, UNSU_, UNSIK): New code_attrs.
4222 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
4223 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
4225 ("eq<mode>3"): ... this.
4226 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
4228 ("ne<mode>3"): ... this.
4230 2017-11-07 Julia Koval <julia.koval@intel.com>
4233 * common/config/i386/i386-common.c
4234 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
4235 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
4236 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
4237 * config/i386/i386.opt: Ditto.
4238 * config/i386/i386.c (ix86_target_string): Ditto.
4239 (ix86_option_override_internal): Ditto.
4240 (ix86_init_mpx_builtins): Move MPX to args2.
4241 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
4242 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
4243 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
4244 __builtin_ia32_vgf2p8affineinvqb_v32qi,
4245 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
4246 __builtin_ia32_vgf2p8affineinvqb_v16qi,
4247 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
4249 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
4252 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
4253 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
4254 (zero-extendsidi peephole2): Remove peephole.
4256 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
4259 * stor-layout.c: Include attribs.h.
4260 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
4261 explicit lookup of "aligned" attribute.
4263 2017-11-07 Andrew Waterman <andrew@sifive.com>
4265 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
4266 (riscv_expand_block_move): Likewise.
4267 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
4269 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
4270 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
4271 * config/riscv/riscv.c (riscv_block_move_straight): New function.
4272 (riscv_adjust_block_mem): Likewise.
4273 (riscv_block_move_loop): Likewise.
4274 (riscv_expand_block_move): Likewise.
4275 * config/riscv/riscv.md (movmemsi): New pattern.
4277 2017-11-07 Michael Clark <michaeljclark@mac.com>
4279 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
4280 (MUSL_DYNAMIC_LINKER): Likewise.
4282 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
4284 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
4285 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
4288 2017-11-07 Richard Biener <rguenther@suse.de>
4290 * match.pd: Fix build.
4292 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
4293 Jackson Woodruff <jackson.woodruff@arm.com>
4295 PR tree-optimization/71026
4296 * match.pd: Canonicalize negate in division.
4298 2017-11-07 Sudakshina Das <sudi.das@arm.com>
4301 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
4303 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4305 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
4306 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
4308 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
4310 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
4311 non-scalar integral types.
4312 * match.pd (negate_expr_p): Handle MINUS_EXPR.
4313 (-(A-B), -(~A)): New transformations.
4315 2017-11-07 Tom de Vries <tom@codesourcery.com>
4317 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
4318 semicolon after "do {} while (0)".
4319 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4320 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4321 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4322 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4323 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4325 2017-11-07 Tom de Vries <tom@codesourcery.com>
4327 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
4328 after "do {} while (0)".
4329 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4330 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4331 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4332 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4333 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
4335 2017-11-07 Tom de Vries <tom@codesourcery.com>
4338 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
4340 (arm_rtx_costs_internal): Add missing semicolon after
4341 HANDLE_NARROW_SHIFT_ARITH call.
4343 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4345 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4346 disable isel if it was not set explicitly.
4348 2017-11-06 James Bowman <james.bowman@ftdichip.com>
4350 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
4351 (add_type_attribute) likewise.
4353 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
4355 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
4356 of struct ix86_frame.
4357 (ix86_initial_elimination_offset): Likewise.
4358 (ix86_expand_split_stack_prologue): Likewise.
4360 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
4362 * tree-vrp.h (enum value_range_type): Update stale comment.
4364 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4365 Alan Hayward <alan.hayward@arm.com>
4366 David Sherwood <david.sherwood@arm.com>
4368 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
4369 (aarch64_expand_vec_perm_const): Take the number of units too.
4370 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
4371 (aarch64_expand_vec_perm_const): Likewise.
4372 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
4373 (vec_perm<mode>): Update accordingly.
4375 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4376 Alan Hayward <alan.hayward@arm.com>
4377 David Sherwood <david.sherwood@arm.com>
4379 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
4380 Take the number of units too.
4381 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
4382 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
4383 but check for a vector mode before rather than after the call.
4384 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
4385 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
4386 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
4387 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
4388 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
4389 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
4390 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
4391 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
4392 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
4393 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
4394 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
4395 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
4396 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
4397 (aarch64_sqdmull2_n<mode>): Update accordingly.
4399 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4400 Alan Hayward <alan.hayward@arm.com>
4401 David Sherwood <david.sherwood@arm.com>
4403 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
4404 the number of units too.
4405 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
4406 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
4407 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
4408 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
4409 (vec_store_lanesxi<mode>): Update accordingly.
4411 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4412 Alan Hayward <alan.hayward@arm.com>
4413 David Sherwood <david.sherwood@arm.com>
4415 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
4416 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
4417 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
4418 of units rather than the mode.
4419 * config/aarch64/iterators.md (nunits): New mode attribute.
4420 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4421 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
4422 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
4423 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
4424 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
4425 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
4426 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
4427 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
4428 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
4429 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
4430 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4431 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
4432 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
4433 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
4434 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
4435 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
4436 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
4437 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
4438 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
4439 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
4440 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
4441 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
4442 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
4443 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4444 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
4445 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
4446 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4447 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
4448 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
4449 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
4450 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
4451 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
4452 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
4453 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
4454 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
4455 (aarch64_simd_vec_setv2di): Likewise.
4457 2017-11-06 Carl Love <cel@us.ibm.com>
4459 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
4461 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
4462 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
4464 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
4466 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
4467 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
4468 Add power 8 macro expansions.
4469 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
4470 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
4471 power 8 instructions. (VSX_XXBR): Add iterator.
4473 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
4475 * config/arm/arm.md (predicable_short_it): Change default to "no",
4476 improve documentation, remove uses that are identical to the default.
4477 (enabled_for_depr_it): Rename to enabled_for_short_it.
4478 * gcc/config/arm/arm-fixed.md (predicable_short_it):
4479 Remove default uses.
4480 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
4481 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
4482 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
4483 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
4485 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
4488 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
4489 float128 helper macros, which are no longer used after deleting
4490 the old 'q' built-in functions, and moving the round to odd
4491 built-in functions to being special built-in functions.
4492 (BU_FLOAT128_2): Likewise.
4493 (BU_FLOAT128_1_HW): Likewise.
4494 (BU_FLOAT128_2_HW): Likewise.
4495 (BU_FLOAT128_3_HW): Likewise.
4496 (FABSQ): Delete old 'q' built-in functions.
4497 (COPYSIGNQ): Likewise.
4498 (SQRTF128_ODD): Move round to odd built-in functions to be
4499 special built-in functions, so that we can handle
4500 -mabi=ieeelongdouble.
4501 (TRUNCF128_ODD): Likewise.
4502 (ADDF128_ODD): Likewise.
4503 (SUBF128_ODD): Likewise.
4504 (MULF128_ODD): Likewise.
4505 (DIVF128_ODD): Likewise.
4506 (FMAF128_ODD): Likewise.
4507 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
4508 built-in names to 'f128'.
4509 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
4510 old 'q' built-in functions, as the machine independent code for
4511 'f128' built-in functions handles this.
4512 (rs6000_expand_builtin): Add expansion for float128 round to odd
4513 functions, keying off on -mabi=ieeelongdouble of whether to use
4514 the KFmode or TFmode variant.
4515 (rs6000_init_builtins): Initialize the _Float128 round to odd
4517 * doc/extend.texi (PowerPC Built-in Functions): Document the old
4518 _Float128 'q' built-in functions are now mapped into the new
4519 'f128' built-in functions.
4521 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
4523 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
4524 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
4526 2017-11-06 David Malcolm <dmalcolm@redhat.com>
4529 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
4530 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
4531 * toplev.c: Include "ipa-fnsummary.h".
4532 (toplev::finalize): Call ipa_fnsummary_c_finalize.
4534 2017-11-06 Jakub Jelinek <jakub@redhat.com>
4536 PR tree-optimization/82838
4537 * gimple-ssa-store-merging.c
4538 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
4539 on a separate gimple_seq which is then appended to seq.
4541 2017-11-06 Jeff Law <law@redhat.com>
4544 * config/i386/i386.c (PROBE_INTERVAL): Remove.
4545 (get_probe_interval): New functions.
4546 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
4547 (ix86_adjust_stack_and_probe): Likewise.
4548 (output_adjust_stack_and_probe): Likewise.
4549 (ix86_emit_probe_stack_range): Likewise.
4550 (ix86_expand_prologue): Likewise.
4552 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4554 PR tree-optimization/82816
4555 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
4556 if the modes of the two types are the same.
4557 (convert_plusminus_to_widen): Likewise.
4559 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4561 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
4563 (usadv16qi): New define_expand.
4564 (usadv8hi): New define_expand.
4566 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
4569 * ipa-inline-transform.c (update_noncloned_frequencies): Always
4571 (inline_transform): Likewise.
4572 * predict.c (counts_to_freqs): Remove useless conditional.
4573 * profile-count.h (profile_count::apply_scale): Move sanity check.
4574 * tree-inline.c (copy_bb): Always scale.
4575 (copy_cfg_body): Likewise.
4577 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
4580 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
4582 (*cmp_ite0): Add enabled_for_depr_it attribute.
4583 (*cmp_ite1): Likewise.
4585 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
4587 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
4590 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
4592 * tree-vrp.c (vrp_int_const_binop): Return true on success and
4593 return the value by pointer.
4594 (extract_range_from_multiplicative_op_1): Update accordingly.
4595 Return as soon as an operation fails.
4597 2017-11-05 Tom de Vries <tom@codesourcery.com>
4600 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
4601 (DEF_SANITIZER_BUILTIN): ... here.
4602 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
4603 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
4605 2017-11-05 Tom de Vries <tom@codesourcery.com>
4608 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
4610 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
4611 ASM_OUTPUT_BEFORE_CASE_LABEL call.
4612 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
4614 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4615 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4616 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
4618 2017-11-05 Tom de Vries <tom@codesourcery.com>
4621 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
4624 2017-11-04 Michael Clark <michaeljclark@mac.com>
4626 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
4627 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
4629 (*addsi3_extended): Likewise.
4630 (*addsi3_extended2): Likewise.
4631 (<optab>si3): Likewise.
4632 (<optab>di3): Likewise.
4633 (<optab><mode>3): Likewise.
4634 (<*optabe>si3_internal): Likewise.
4635 (zero_extendqi<SUPERQI:mode>2): Likewise.
4636 (*add<mode>hi3): Likewise.
4637 (*xor<mode>hi3): Likewise.
4638 (<optab>di3): Likewise.
4639 (*<optab>si3_extend): Likewise.
4640 (*sge<u>_<X:mode><GPR:mode>): Likewise.
4641 (*slt<u>_<X:mode><GPR:mode>): Likewise.
4642 (*sle<u>_<X:mode><GPR:mode>): Likewise.
4644 2017-11-04 Andrew Waterman <andrew@sifive.com>
4646 * config/riscv/riscv.c (riscv_option_override): Conditionally set
4647 TARGET_STRICT_ALIGN based upon -mtune argument.
4649 2017-11-04 Andrew Waterman <andrew@sifive.com>
4651 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
4653 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
4655 * config/i386/i386.c (choose_basereg): Use optional scratch
4656 register and add assertion.
4657 (x86_emit_outlined_ms2sysv_save): Use scratch register when
4658 needed, and don't allocate stack.
4659 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
4660 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
4661 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
4663 2017-11-03 Jeff Law <law@redhat.com>
4665 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
4666 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
4667 to probe at the start of a noreturn function.
4669 2017-11-03 Jakub Jelinek <jakub@redhat.com>
4671 PR tree-optimization/78821
4672 * gimple-ssa-store-merging.c: Update the file comment.
4673 (MAX_STORE_ALIAS_CHECKS): Define.
4674 (struct store_operand_info): New type.
4675 (store_operand_info::store_operand_info): New constructor.
4676 (struct store_immediate_info): Add rhs_code and ops data members.
4677 (store_immediate_info::store_immediate_info): Add rhscode, op0r
4678 and op1r arguments to the ctor, initialize corresponding data members.
4679 (struct merged_store_group): Add load_align_base and load_align
4681 (merged_store_group::merged_store_group): Initialize them.
4682 (merged_store_group::do_merge): Update them.
4683 (merged_store_group::apply_stores): Pick the constant for
4684 encode_tree_to_bitpos from one of the two operands, or skip
4685 encode_tree_to_bitpos if neither operand is a constant.
4686 (class pass_store_merging): Add process_store method decl. Remove
4687 bool argument from terminate_all_aliasing_chains method decl.
4688 (pass_store_merging::terminate_all_aliasing_chains): Remove
4689 var_offset_p argument and corresponding handling.
4690 (stmts_may_clobber_ref_p): New function.
4691 (compatible_load_p): New function.
4692 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
4693 if there is overlap and rhs_code is not INTEGER_CST. For
4694 non-overlapping stores terminate group if rhs is not mergeable.
4695 (get_alias_type_for_stmts): Change first argument from
4696 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
4697 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
4698 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
4700 (get_location_for_stmts): Change first argument from
4701 auto_vec<gimple *> & to vec<gimple *> &.
4702 (struct split_store): Remove orig_stmts data member, add orig_stores.
4703 (split_store::split_store): Create orig_stores rather than orig_stmts.
4704 (find_constituent_stmts): Renamed to ...
4705 (find_constituent_stores): ... this. Change second argument from
4706 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
4707 to info structures rather than the statements.
4708 (split_group): Rename ALLOW_UNALIGNED argument to
4709 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
4710 it. Adjust find_constituent_stores caller.
4711 (imm_store_chain_info::output_merged_store): Handle rhs_code other
4712 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
4713 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
4714 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
4715 (mem_valid_for_store_merging): New function.
4716 (handled_load): New function.
4717 (pass_store_merging::process_store): New method.
4718 (pass_store_merging::execute): Use process_store method. Adjust
4719 terminate_all_aliasing_chains caller.
4721 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4723 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
4724 Return true for more constants, symbols and label references.
4725 (aarch64_valid_floating_const): Remove unused function.
4727 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4730 * config/aarch64/aarch64.c (aarch64_layout_frame):
4731 Undo forcing of LR at bottom of frame.
4733 2017-11-03 Jeff Law <law@redhat.com>
4736 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
4737 for int_registers_saved.
4739 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
4740 extracted from tree-ssa-dom.c.
4741 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
4742 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
4743 (record_equivalences_from_incoming_edge): Add additional argument
4744 to single_pred_edge_ignoring_loop_edges call.
4745 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
4746 (uncprop_dom_walker::before_dom_children): Add additional argument
4747 to single_pred_edge_ignoring_loop_edges call.
4748 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
4749 single_pred_edge_ignoring_loop_edges rather than open coding.
4750 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
4752 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
4754 * match.pd (-(-A)): Rewrite.
4756 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4758 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
4759 (rs6000_emit_int_cmove): New declaration.
4760 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
4761 (rs6000_emit_sISEL): Delete.
4762 (rs6000_emit_int_cmove): Make non-static.
4763 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
4764 instead of rs6000_emit_sISEL.
4766 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
4768 * asan.c (create_cond_insert_point): Maintain profile.
4769 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
4771 * basic-block.h (struct basic_block_def): Remove frequency.
4772 (EDGE_FREQUENCY): Use to_frequency
4773 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
4775 (find_traces): Update to use to_frequency.
4776 (find_traces_1_round): Likewise; use only IPA counts.
4777 (bb_to_key): Likewise.
4778 (connect_traces): Use IPA counts only.
4779 (copy_bb_p): Update to use to_frequency.
4780 (fix_up_crossing_landing_pad): Likewise.
4781 (sanitize_hot_paths): Likewise.
4782 * bt-load.c (basic_block_freq): Likewise.
4783 * cfg.c (init_flow): Set count_max to uninitialized.
4784 (check_bb_profile): Remove frequencies; check counts.
4785 (dump_bb_info): Do not dump frequencies.
4786 (update_bb_profile_for_threading): Update counts only.
4787 (scale_bbs_frequencies_int): Likewise.
4788 (MAX_SAFE_MULTIPLIER): Remove.
4789 (scale_bbs_frequencies_gcov_type): Update counts only.
4790 (scale_bbs_frequencies_profile_count): Update counts only.
4791 (scale_bbs_frequencies): Update counts only.
4792 * cfg.h (struct control_flow_graph): Add count-max.
4793 (update_bb_profile_for_threading): Update prototype.
4794 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
4795 (find_many_sub_basic_blocks): Likewise.
4796 * cfgcleanup.c (try_forward_edges): Likewise.
4797 (try_crossjump_to_edge): Likewise.
4798 * cfgexpand.c (expand_gimple_cond): Likewise.
4799 (expand_gimple_tailcall): Likewise.
4800 (construct_init_block): Likewise.
4801 (construct_exit_block): Likewise.
4802 * cfghooks.c (verify_flow_info): Check consistency of counts.
4803 (dump_bb_for_graph): Do not dump frequencies.
4804 (split_block_1): Do not update frequencies.
4805 (split_edge): Do not update frequencies.
4806 (make_forwarder_block): Do not update frequencies.
4807 (duplicate_block): Do not update frequencies.
4808 (account_profile_record): Do not update frequencies.
4809 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
4810 for global heuristics.
4811 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
4812 (expected_loop_iterations_unbounded): Use counts only.
4813 * cfgloopmanip.c (scale_loop_profile): Simplify.
4814 (create_empty_loop_on_edge): Simplify
4816 (duplicate_loop_to_header_edge): Simplify
4817 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
4818 (update_br_prob_note): Take care of removing note when profile
4820 (relink_block_chain): Do not dump frequency.
4821 (rtl_account_profile_record): Use to_frequency.
4822 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
4823 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
4824 (cgraph_update_edges_for_call_stmt_node): Likewise.
4825 (cgraph_edge::verify_count_and_frequency): Update.
4826 (cgraph_node::verify_node): Temporarily disable frequency verification.
4827 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
4828 to_cgraph_frequency.
4829 (cgraph_edge::rebuild_edges): Convert to ipa counts.
4830 * cgraphunit.c (init_lowered_empty_function): Do not initialize
4832 (cgraph_node::expand_thunk): Update profile.
4833 * except.c (dw2_build_landing_pads): Do not update frequency.
4834 * final.c (compute_alignments): Use to_frequency.
4835 (dump_basic_block_info): Do not dump frequency.
4836 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
4837 (dump_gimple_bb_header): Do not dump frequency.
4838 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
4840 * gimple-streamer-in.c (input_bb): Do not stream frequency.
4841 * gimple-streamer-out.c (output_bb): Do not stream frequency.
4842 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
4843 (init_before_recovery): Do not update frequency.
4844 (sched_create_recovery_edges): Do not update frequency.
4845 * hsa-gen.c (convert_switch_statements): Do not update frequency.
4846 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
4847 (ipa_cp_c_finalize): Set max_count to uninitialized.
4848 * ipa-fnsummary.c (get_minimal_bb): Use counts.
4849 (param_change_prob): Use counts.
4850 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
4852 * ipa-split.c (consider_split): Use to_frequency.
4853 (split_function): Use to_frequency.
4854 * ira-build.c (loop_compare_func): Likewise.
4855 (mark_loops_for_removal): Likewise.
4856 (mark_all_loops_for_removal): Likewise.
4857 * loop-doloop.c (doloop_modify): Do not update frequency.
4858 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
4860 * lto-streamer-in.c (input_function): Update count_max.
4861 * omp-expand.c (expand_omp_taskreg): Update count_max.
4862 * omp-simd-clone.c (simd_clone_adjust): Update profile.
4863 * predict.c (maybe_hot_frequency_p): Use to_frequency.
4864 (maybe_hot_count_p): Use ipa counts only.
4865 (maybe_hot_bb_p): Simplify.
4866 (maybe_hot_edge_p): Simplify.
4867 (probably_never_executed): Do not take frequency argument.
4868 (probably_never_executed_bb_p): Do not pass frequency.
4869 (probably_never_executed_edge_p): Likewise.
4870 (combine_predictions_for_bb): Check that profile is nonzero.
4871 (propagate_freq): Do not set frequency.
4872 (drop_profile): Simplify.
4873 (counts_to_freqs): Simplify.
4874 (expensive_function_p): Use to_frequency.
4875 (propagate_unlikely_bbs_forward): Simplify.
4876 (determine_unlikely_bbs): Simplify.
4877 (estimate_bb_frequencies): Add hack to silence graphite issues.
4878 (compute_function_frequency): Use ipa counts.
4879 (pass_profile::execute): Update.
4880 (rebuild_frequencies): Use counts only.
4881 (force_edge_cold): Use counts only.
4882 * profile-count.c (profile_count::dump): Dump new count types.
4883 (profile_count::differs_from_p): Check compatiblity.
4884 (profile_count::to_frequency): New function.
4885 (profile_count::to_cgraph_frequency): New function.
4886 * profile-count.h (struct function): Declare.
4887 (enum profile_quality): Add profile_guessed_local and
4888 profile_guessed_global0.
4889 (class profile_proability): Decrease number of bits to 29;
4890 update from_reg_br_prob_note and to_reg_br_prob_note.
4891 (class profile_count: Update comment; decrease number of bits
4892 to 61. Check compatibility.
4893 (profile_count::compatible_p): New private member function.
4894 (profile_count::ipa_p): New member function.
4895 (profile_count::operator<): Handle global zero correctly.
4896 (profile_count::operator>): Handle global zero correctly.
4897 (profile_count::operator<=): Handle global zero correctly.
4898 (profile_count::operator>=): Handle global zero correctly.
4899 (profile_count::nonzero_p): New member function.
4900 (profile_count::force_nonzero): New member function.
4901 (profile_count::max): New member function.
4902 (profile_count::apply_scale): Handle IPA scalling.
4903 (profile_count::guessed_local): New member function.
4904 (profile_count::global0): New member function.
4905 (profile_count::ipa): New member function.
4906 (profile_count::to_frequency): Declare.
4907 (profile_count::to_cgraph_frequency): Declare.
4908 * profile.c (OVERLAP_BASE): Delete.
4909 (compute_frequency_overlap): Delete.
4910 (compute_branch_probabilities): Do not use compute_frequency_overlap.
4911 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
4912 * sched-ebb.c (rank): Use counts only.
4913 * shrink-wrap.c (handle_simple_exit): Use counts only.
4914 (try_shrink_wrapping): Use counts only.
4915 (place_prologue_for_one_component): Use counts only.
4916 * tracer.c (find_best_predecessor): Use to_frequency.
4917 (find_trace): Use to_frequency.
4918 (tail_duplicate): Use to_frequency.
4919 * trans-mem.c (expand_transaction): Do not update frequency.
4920 * tree-call-cdce.c: Do not update frequency.
4921 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
4922 (gimple_merge_blocks): Likewise.
4923 (gimple_split_edge): Likewise.
4924 (gimple_duplicate_sese_region): Likewise.
4925 (gimple_duplicate_sese_tail): Likewise.
4926 (move_sese_region_to_fn): Likewise.
4927 (gimple_account_profile_record): Likewise.
4928 (insert_cond_bb): Likewise.
4929 * tree-complex.c (expand_complex_div_wide): Likewise.
4930 * tree-eh.c (lower_resx): Update profile.
4931 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
4933 (initialize_cfun): Do not initialize frequencies
4934 (freqs_to_counts): Delete.
4935 (copy_cfg_body): Ignore count parameter.
4936 (copy_body): Update.
4937 (expand_call_inline): Update count_max.
4938 (optimize_inline_calls): Update count_max.
4939 (tree_function_versioning): Update count_max.
4940 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
4941 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
4943 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
4944 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
4945 (try_peel_loop): Likewise.
4946 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
4948 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
4949 (tree_transform_and_unroll_loop): Do not use frequencies
4950 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
4951 Use reliable prediction only.
4952 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
4953 * tree-ssa-sink.c (select_best_block): Use to_frequency.
4954 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
4955 probability scaling.
4956 * tree-ssa-threadupdate.c (create_block_for_threading): Do
4957 not update frequency
4958 (any_remaining_duplicated_blocks): Likewise.
4959 (update_profile): Likewise.
4960 (estimated_freqs_path): Delete.
4961 (freqs_to_counts_path): Delete.
4962 (clear_counts_path): Delete.
4963 (ssa_fix_duplicate_block_edges): Likewise.
4964 (duplicate_thread_path): Likewise.
4965 * tree-switch-conversion.c (gen_inbound_check): Use counts.
4966 * tree-tailcall.c (decrease_profile): Do not update frequency.
4967 (eliminate_tail_call): Likewise.
4968 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
4969 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
4970 (optimize_mask_stores): Likewise.
4971 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4972 * ubsan.c (ubsan_expand_null_ifn): Update profile.
4973 (ubsan_expand_ptr_ifn): Update profile.
4974 * value-prof.c (gimple_ic): Simplify.
4975 * value-prof.h (gimple_ic): Update prototype.
4976 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
4977 * ipa-inline.c (compute_uninlined_call_time): Be sure that
4979 (want_inline_self_recursive_call_p): Likewise.
4980 (resolve_noninline_speculation): Only cummulate defined counts.
4981 (inline_small_functions): Use nonzero_p.
4982 (ipa_inline): Do not access freed node.
4984 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
4986 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4987 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
4989 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
4991 * config/riscv/riscv.c (riscv_legitimize_move): Handle
4992 non-legitimate address.
4994 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
4996 * config/rs6000/rs6000.md (*lt0_disi): Delete.
4997 (*lt0_<mode>di, *lt0_<mode>si): New.
4999 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
5001 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
5002 TARGET_PAIRED_FLOAT.
5004 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
5005 Jim Wilson <jim.wilson@linaro.org>
5007 * config/aarch64/aarch64-cores.def (saphira): New CPU.
5008 * config/aarch64/aarch64-tune.md: Regenerated.
5009 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
5010 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
5012 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
5014 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
5015 (arc_expand_prologue): Restore blink for millicode.
5016 * config/arc/linux.h (LINK_EH_SPEC): Defined.
5018 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
5021 * config/i386/i386.c (ix86_vector_duplicate_value): Use
5022 gen_vec_duplicate after forcing the scalar into a register.
5024 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
5026 * combine (try_combine): Print the insns input to try_combine to the
5029 2017-11-02 Steve Ellcey <sellcey@cavium.com>
5032 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
5033 Remove second argument from aarch64_process_target_attr call.
5034 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
5036 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
5038 (aarch64_handle_attr_arch): Remove second argument.
5039 (aarch64_handle_attr_cpu): Ditto.
5040 (aarch64_handle_attr_tune): Ditto.
5041 (aarch64_handle_attr_isa_flags): Ditto.
5042 (aarch64_process_one_target_attr): Ditto.
5043 (aarch64_process_target_attr): Ditto.
5044 (aarch64_option_valid_attribute_p): Remove second argument.
5045 on aarch64_process_target_attr call.
5047 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5049 * diagnostic.c: Include "selftest-diagnostic.h".
5050 (selftest::assert_location_text): New function.
5051 (selftest::test_diagnostic_get_location_text): New function.
5052 (selftest::diagnostic_c_tests): Call it.
5054 2017-11-02 David Malcolm <dmalcolm@redhat.com>
5056 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
5057 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
5058 (class selftest::test_diagnostic_context): Move to...
5059 * selftest-diagnostic.c: New file.
5060 * selftest-diagnostic.h: New file.
5062 2017-11-02 James Bowman <james.bowman@ftdichip.com>
5064 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
5065 offset range for FT32B.
5066 * config/ft32/ft32.h: option "mcompress" enables relaxation.
5067 * config/ft32/ft32.md: Add TARGET_NOPM.
5068 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
5069 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
5071 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
5073 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
5075 2017-11-02 Jeff Law <law@redhat.com>
5077 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
5078 virtual keyword on FINAL OVERRIDE members.
5080 * tree-ssa-propagate.h (ssa_propagation_engine): Group
5081 virtuals together. Add virtual destructor.
5082 (substitute_and_fold_engine): Similarly.
5084 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
5086 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5088 2017-11-02 Richard Biener <rguenther@suse.de>
5090 PR tree-optimization/82795
5091 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
5093 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5095 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
5097 (gcc_GAS_CHECK_FEATURE): Remove.
5098 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
5099 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
5100 * configure: Regenerate.
5102 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
5104 * config/arc/arc.c (hwloop_optimize): Account for empty
5107 2017-11-02 Richard Biener <rguenther@suse.de>
5110 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
5111 Truncate ARRAY_REF index and element size.
5113 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5115 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
5117 2017-11-01 Jeff Law <law@redhat.com>
5119 * tree-ssa-ccp.c (ccp_folder): New class derived from
5120 substitute_and_fold_engine.
5121 (ccp_folder::get_value): New member function.
5122 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
5123 (ccp_fold_stmt): Remove prototype.
5124 (ccp_finalize): Call substitute_and_fold from the ccp_class.
5125 * tree-ssa-copy.c (copy_folder): New class derived from
5126 substitute_and_fold_engine.
5127 (copy_folder::get_value): Renamed from get_value.
5128 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
5129 * tree-vrp.c (vrp_folder): New class derived from
5130 substitute_and_fold_engine.
5131 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
5132 (vrp_folder::get_value): New member function.
5133 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
5134 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
5135 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
5136 provide a class interface to folder/substitute routines.
5137 (ssa_prop_fold_stmt_fn): Remove typedef.
5138 (ssa_prop_get_value_fn): Likewise.
5139 (subsitute_and_fold): Remove prototype.
5140 (replace_uses_in): Likewise.
5141 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
5142 Renamed from replace_uses_in. Call the virtual member function
5143 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
5144 (substitute_and_fold_dom_walker): Remove initialization of
5145 data member entries for calbacks. Add substitute_and_fold_engine
5146 member and initialize it.
5147 (substitute_and_fold_dom_walker::before_dom_children0: Use the
5148 member functions for get_value, replace_phi_args_in c
5149 replace_uses_in, and fold_stmt calls.
5150 (substitute_and_fold_engine::substitute_and_fold): Renamed from
5151 substitute_and_fold. Remove assert. Update ctor call.
5153 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
5154 (ssa_prop_visit_phi_fn): Likewise.
5155 (class ssa_propagation_engine): New class to provide an interface
5157 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
5159 (ssa_prop_visit_phi): Likewise.
5160 (ssa_propagation_engine::simulate_stmt): Moved into class.
5161 Call visit_phi/visit_stmt from the class rather than via
5162 file scoped static variables.
5163 (ssa_propagation_engine::simulate_block): Moved into class.
5164 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
5165 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
5166 set file scoped statics for the visit_stmt/visit_phi callbacks.
5167 * tree-complex.c (complex_propagate): New class derived from
5168 ssa_propagation_engine.
5169 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
5170 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
5171 (tree_lower_complex): Call ssa_propagate via the complex_propagate
5173 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
5174 ssa_propagation_engine.
5175 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
5176 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
5177 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
5178 * tree-ssa-copy.c (copy_prop): New class derived from
5179 ssa_propagation_engine.
5180 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
5181 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
5182 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
5183 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
5184 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
5185 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
5186 (execute_vrp): Call ssa_propagate from the vrp_prop class.
5188 2017-11-01 Jakub Jelinek <jakub@redhat.com>
5190 PR rtl-optimization/82778
5191 PR rtl-optimization/82597
5192 * compare-elim.c (struct comparison): Add in_a_setter field.
5193 (find_comparison_dom_walker::before_dom_children): Remove killed
5194 bitmap and df_simulate_find_defs call, instead walk the defs.
5195 Compute last_setter and initialize in_a_setter. Merge definitions
5196 with first initialization for a few variables.
5197 (try_validate_parallel): Use insn_invalid_p instead of
5198 recog_memoized. Return insn rather than just the pattern.
5199 (try_merge_compare): Fix up comment. Don't uselessly test if
5200 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
5202 (execute_compare_elim_after_reload): Remove df_chain_add_problem
5205 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5206 Alan Hayward <alan.hayward@arm.com>
5207 David Sherwood <david.sherwood@arm.com>
5209 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
5210 aarch64_hard_regno_nregs to get the number of registers
5213 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5214 Alan Hayward <alan.hayward@arm.com>
5215 David Sherwood <david.sherwood@arm.com>
5217 * config/aarch64/constraints.md (Upl): Rename to...
5219 * config/aarch64/aarch64.md
5220 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
5223 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5224 Alan Hayward <alan.hayward@arm.com>
5225 David Sherwood <david.sherwood@arm.com>
5227 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
5228 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
5231 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5232 Alan Hayward <alan.hayward@arm.com>
5233 David Sherwood <david.sherwood@arm.com>
5235 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
5236 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
5237 (aarch64_evpc_dup): Generate rtl direcly, rather than using
5239 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
5240 of a single element.
5241 * config/aarch64/iterators.md: Add a comment above the permute
5242 unspecs to say that they are generated directly by
5243 aarch64_expand_vec_perm_const.
5244 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
5246 2017-11-01 Nathan Sidwell <nathan@acm.org>
5248 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
5250 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
5252 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
5253 and medany code models, and describe what they do.
5255 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5257 Revert accidental duplicate:
5259 * combine.c (can_change_dest_mode): Reject changes in
5260 REGMODE_NATURAL_SIZE.
5262 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
5264 PR rtl-optimization/64682
5265 PR rtl-optimization/69567
5266 PR rtl-optimization/69737
5267 PR rtl-optimization/82683
5268 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
5269 register mentioned in the note, drop the note, unless it came from I3,
5270 in which case it should go to I3 again.
5272 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5274 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
5275 and return false if not.
5276 (clear_bytes_written_by, live_bytes_read): Update accordingly.
5278 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5280 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
5281 range is known to be empty.
5283 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5284 Alan Hayward <alan.hayward@arm.com>
5285 David Sherwood <david.sherwood@arm.com>
5287 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
5288 and CONST_VECTOR_NUNITS instead of computing the number of units from
5289 the byte sizes of the vector and element.
5290 (simplify_binary_operation_1): Likewise.
5291 (simplify_const_binary_operation): Likewise.
5292 (simplify_ternary_operation): Likewise.
5294 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5295 Alan Hayward <alan.hayward@arm.com>
5296 David Sherwood <david.sherwood@arm.com>
5298 * var-tracking.c (INT_MEM_OFFSET): Replace with...
5299 (int_mem_offset): ...this new function.
5300 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
5301 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
5302 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
5305 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5306 Alan Hayward <alan.hayward@arm.com>
5307 David Sherwood <david.sherwood@arm.com>
5309 * lower-subreg.c (interesting_mode_p): New function.
5310 (compute_costs, find_decomposable_subregs, decompose_register)
5311 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
5312 (resolve_clobber, dump_choices): Use it.
5314 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5315 Alan Hayward <alan.hayward@arm.com>
5316 David Sherwood <david.sherwood@arm.com>
5318 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
5320 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5321 Alan Hayward <alan.hayward@arm.com>
5322 David Sherwood <david.sherwood@arm.com>
5324 * alias.c (find_base_value, find_base_term): Only process integer
5325 truncations. Check the precision rather than the size.
5327 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5328 Alan Hayward <alan.hayward@arm.com>
5329 David Sherwood <david.sherwood@arm.com>
5331 * machmode.h (is_narrower_int_mode): New function
5332 * optabs.c (expand_float, expand_fix): Use it.
5333 * dwarf2out.c (rotate_loc_descriptor): Likewise.
5335 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5336 Alan Hayward <alan.hayward@arm.com>
5337 David Sherwood <david.sherwood@arm.com>
5339 * rtl.h (narrower_subreg_mode): New function.
5340 * ira-color.c (update_costs_from_allocno): Use it.
5342 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5343 Alan Hayward <alan.hayward@arm.com>
5344 David Sherwood <david.sherwood@arm.com>
5346 * optabs-query.h (convert_optab_p): New function, split out from...
5347 (convert_optab_handler): ...here.
5348 (widening_optab_handler): Delete.
5349 (find_widening_optab_handler): Remove permit_non_widening parameter.
5350 (find_widening_optab_handler_and_mode): Likewise. Provide an
5351 override that operates on mode class wrappers.
5352 * optabs-query.c (widening_optab_handler): Delete.
5353 (find_widening_optab_handler_and_mode): Remove permit_non_widening
5354 parameter. Assert that the two modes are the same class and that
5355 the "from" mode is narrower than the "to" mode. Use
5356 convert_optab_handler instead of widening_optab_handler.
5357 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
5358 instead of widening_optab_handler.
5359 * expr.c (expand_expr_real_2): Update calls to
5360 find_widening_optab_handler.
5361 * optabs.c (expand_widen_pattern_expr): Likewise.
5362 (expand_binop_directly): Take the insn_code as a parameter.
5363 (expand_binop): Only call find_widening_optab_handler for
5364 conversion optabs; use optab_handler otherwise. Update calls
5365 to find_widening_optab_handler and expand_binop_directly.
5366 Use convert_optab_handler instead of widening_optab_handler.
5367 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
5368 find_widening_optab_handler and use scalar_mode rather than
5370 (convert_plusminus_to_widen): Likewise.
5372 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5373 Alan Hayward <alan.hayward@arm.com>
5374 David Sherwood <david.sherwood@arm.com>
5376 * machmode.h (fixed_size_mode): New class.
5377 * rtl.h (get_pool_mode): Return fixed_size_mode.
5378 * gengtype.c (main): Add fixed_size_mode.
5379 * target.def (get_raw_result_mode): Return a fixed_size_mode.
5380 (get_raw_arg_mode): Likewise.
5381 * doc/tm.texi: Regenerate.
5382 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
5383 * targhooks.c (default_get_reg_raw_mode): Likewise.
5384 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
5385 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
5386 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
5387 (msp430_get_raw_result_mode): Likewise.
5388 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
5389 * dbxout.c (dbxout_parms): Require fixed-size modes.
5390 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
5391 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
5392 * omp-low.c (lower_oacc_reductions): Likewise.
5393 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
5394 (simplify_subreg): Update accordingly.
5395 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
5396 (force_const_mem): Update accordingly. Return NULL_RTX for modes
5397 that aren't fixed-size.
5398 (get_pool_mode): Return a fixed_size_mode.
5399 (output_constant_pool_2): Take a fixed_size_mode.
5401 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5402 Alan Hayward <alan.hayward@arm.com>
5403 David Sherwood <david.sherwood@arm.com>
5405 * doc/rtl.texi (vec_series): Document.
5406 (const): Say that the operand can be a vec_series.
5407 * rtl.def (VEC_SERIES): New rtx code.
5408 * rtl.h (const_vec_series_p_1): Declare.
5409 (const_vec_series_p): New function.
5410 * emit-rtl.h (gen_const_vec_series): Declare.
5411 (gen_vec_series): Likewise.
5412 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
5413 (gen_vec_series): Likewise.
5414 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
5415 * simplify-rtx.c (simplify_unary_operation): Handle negations
5417 (simplify_binary_operation_series): New function.
5418 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
5419 (test_vector_ops_series): New function.
5420 (test_vector_ops): Call it.
5421 * config/powerpcspe/altivec.md (altivec_lvsl): Use
5422 gen_const_vec_series.
5423 (altivec_lvsr): Likewise.
5424 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
5426 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5427 Alan Hayward <alan.hayward@arm.com>
5428 David Sherwood <david.sherwood@arm.com>
5430 * doc/rtl.texi (const): Update description of address constants.
5431 Say that vector constants are allowed too.
5432 * common.md (E, F): Use CONSTANT_P instead of checking for
5434 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
5435 checking for CONST_VECTOR.
5436 * expmed.c (make_tree): Use build_vector_from_val for a CONST
5438 * expr.c (expand_expr_real_2): Check for vector modes instead
5439 of checking for CONST_VECTOR.
5440 * rtl.h (const_vec_p): New function.
5441 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
5442 (unwrap_const_vec_duplicate): Handle them here too.
5444 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5445 David Malcolm <dmalcolm@redhat.com>
5446 Alan Hayward <alan.hayward@arm.com>
5447 David Sherwood <david.sherwood@arm.com>
5449 * rtl.h (vec_duplicate_p): New function.
5450 * selftest-rtl.c (assert_rtx_eq_at): New function.
5451 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
5452 (assert_rtx_eq_at): Declare.
5453 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
5454 * selftest-run-tests.c (selftest::run_tests): Call it.
5455 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
5456 (simplify_unary_operation_1): Recursively handle vector duplicates.
5457 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
5459 (simplify_subreg): Handle subregs of vector duplicates.
5460 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
5461 (selftest::simplify_rtx_c_tests): New functions.
5463 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5464 Alan Hayward <alan.hayward@arm.com>
5465 David Sherwood <david.sherwood@arm.com>
5467 * emit-rtl.h (gen_const_vec_duplicate): Declare.
5468 (gen_vec_duplicate): Likewise.
5469 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
5471 (gen_const_vector): ...here.
5472 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
5473 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
5474 whose elements are all equal.
5475 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
5476 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
5477 (simplify_relational_operation): Likewise.
5478 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
5480 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
5481 (aarch64_expand_vector_init): Likewise.
5482 * config/arm/arm.c (neon_vdup_constant): Likewise.
5483 (neon_expand_vector_init): Likewise.
5484 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
5485 (arm_block_set_unaligned_vect): Likewise.
5486 (arm_block_set_aligned_vect): Likewise.
5487 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
5488 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
5489 (expand_vec_perm_even_odd_pack): Likewise.
5490 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
5491 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
5492 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
5493 gen_const_vec_duplicate.
5494 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
5495 * config/mips/mips.c (mips_gen_const_int_vector): Use
5496 gen_const_vec_duplicate.
5497 (mips_expand_vector_init): Use CONST0_RTX.
5498 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
5499 (define_split): Use gen_const_vec_duplicate.
5500 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
5501 (define_split): Use gen_const_vec_duplicate.
5502 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
5503 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
5504 * config/spu/spu.c (spu_const): Likewise.
5506 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
5507 Alan Hayward <alan.hayward@arm.com>
5508 David Sherwood <david.sherwood@arm.com>
5510 * combine.c (can_change_dest_mode): Reject changes in
5511 REGMODE_NATURAL_SIZE.
5513 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
5515 * configure.ac (--enable-libssp): New.
5516 (gcc_cv_libc_provides_ssp): Check for explicit setting before
5517 trying to determine target-specific default. Adjust indentation.
5518 * configure: Regenerated.
5519 * doc/install.texi (Configuration): Expand --disable-libssp
5522 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
5524 config/i386/i386.c (ix86_expand_epilogue): Correct stack
5527 2017-10-31 Martin Jambor <mjambor@suse.cz>
5530 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
5532 2017-10-31 David Malcolm <dmalcolm@redhat.com>
5534 * auto-profile.c (autofdo_source_profile::read): Use
5535 UNKNOWN_LOCATION rather than 0.
5536 * diagnostic-core.h (warning_at_rich_loc): Rename to...
5537 (warning_at): ...this overload.
5538 (warning_at_rich_loc_n): Rename to...
5539 (warning_n): ...this overload.
5540 (error_at_rich_loc): Rename to...
5541 (error_at): ...this overload.
5542 (pedwarn_at_rich_loc): Rename to...
5543 (pedwarn): ...this overload.
5544 (permerror_at_rich_loc): Rename to...
5545 (permerror): ...this overload.
5546 (inform_at_rich_loc): Rename to...
5547 (inform): ...this overload.
5548 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
5549 (diagnostic_n_impl_richloc): Rename to...
5550 (diagnostic_n_impl): ...this rich_location *-based decl.
5551 (inform_at_rich_loc): Rename to...
5552 (inform): ...this, and add an assertion.
5553 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
5554 (warning_at_rich_loc): Rename to...
5555 (warning_at): ...this, and add an assertion.
5556 (warning_at_rich_loc_n): Rename to...
5557 (warning_n): ...this, and add an assertion.
5558 (warning_n): Update location_t-based implementation for removal of
5559 location_t-based diagnostic_n_impl.
5560 (pedwarn_at_rich_loc): Rename to...
5561 (pedwarn): ...this, and add an assertion.
5562 (permerror_at_rich_loc): Rename to...
5563 (permerror): ...this, and add an assertion.
5564 (error_n): Update for removal of location_t-based diagnostic_n_impl.
5565 (error_at_rich_loc): Rename to...
5566 (error_at): ...this, and add an assertion.
5567 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
5568 (driver::do_spec_on_infiles): Likewise.
5569 * substring-locations.c (format_warning_va): Update for renaming
5570 of inform_at_rich_loc.
5572 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
5574 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
5575 _Float<N>X built-in functions so that the variant without the
5576 "__builtin_" prefix is only enabled for the GNU C and Objective C
5577 languages when they are in non-strict ANSI/ISO mode.
5578 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
5579 * target.def (floatn_builtin_p): Add a target hook to control
5580 whether _Float<N> and _Float<N>X built-in functions without the
5581 "__builtin_" prefix are enabled, and return true for C and
5582 Objective C in the default hook. Include langhooks.h in
5584 * targhooks.h (default_floatn_builtin_p): Likewise.
5585 * targhooks.c (default_floatn_builtin_p): Likewise.
5586 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
5587 floatn_builtin_p target hook.
5588 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
5590 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
5591 Eric Botcazou <ebotcazou@adacore.com>
5593 PR rtl-optimization/81803
5594 * lra-constraints.c (curr_insn_transform): Also reload the whole
5595 register for a strict subreg no wider than a word if this is for
5596 a WORD_REGISTER_OPERATIONS target.
5598 2017-10-31 Jason Merrill <jason@redhat.com>
5600 * gdbinit.in: Skip over inlines from timevar.h.
5602 2017-10-31 Martin Liska <mliska@suse.cz>
5604 * doc/gcov.texi: Document new option.
5605 * gcov.c (print_usage): Likewise print it.
5606 (process_args): Support the argument.
5607 (format_count): New function.
5608 (format_gcov): Use the function.
5610 2017-10-31 Martin Liska <mliska@suse.cz>
5612 * gcov.c (struct name_map): do not use typedef.
5613 Define operator== and operator<.
5614 (name_search): Remove.
5615 (name_sort): Remove.
5616 (main): Do not allocate names.
5617 (process_file): Add vertical space.
5618 (generate_results): Use std::find.
5619 (release_structures): Do not release memory.
5620 (find_source): Use std::find.
5622 2017-10-31 Martin Liska <mliska@suse.cz>
5624 * gcov.c (struct line_info): Remove it's typedef.
5625 (line_info::line_info): Add proper ctor.
5626 (line_info::has_block): Do not use a typedef.
5627 (struct source_info): Do not use typedef.
5628 (circuit): Likewise.
5629 (get_cycles_count): Likewise.
5630 (output_intermediate_file): Iterate via vector iterator.
5631 (add_line_counts): Use std::vector methods.
5632 (accumulate_line_counts): Likewise.
5633 (output_lines): Likewise.
5635 2017-10-31 Martin Liska <mliska@suse.cz>
5637 * gcov.c (struct source_info): Remove typedef.
5638 (source_info::source_info): Add proper ctor.
5639 (accumulate_line_counts): Use struct, not it's typedef.
5640 (output_gcov_file): Likewise.
5641 (output_lines): Likewise.
5642 (main): Do not allocate an array.
5643 (output_intermediate_file): Use size of vector container.
5644 (process_file): Resize the vector.
5645 (generate_results): Do not preallocate, use newly added vector
5647 (release_structures): Do not release sources.
5648 (find_source): Use vector methods.
5649 (add_line_counts): Do not use typedef.
5651 2017-10-31 Martin Liska <mliska@suse.cz>
5653 * doc/gcov.texi: Document that.
5654 * gcov.c (add_line_counts): Mark lines with a non-executed
5656 (output_line_beginning): Handle such lines.
5657 (output_lines): Pass new argument.
5658 (output_intermediate_file): Print it in intermediate format.
5660 2017-10-31 Martin Liska <mliska@suse.cz>
5662 * color-macros.h: New file.
5663 * diagnostic-color.c: Factor out color related to macros to
5665 * doc/gcov.texi: Document -k option.
5666 * gcov.c (INCLUDE_STRING): Include string.h.
5667 (print_usage): Add -k option.
5668 (process_args): Parse it.
5669 (pad_count_string): New function.
5670 (output_line_beginning): Likewise.
5671 (DEFAULT_LINE_START): New macro.
5672 (output_lines): Support color output.
5674 2017-10-31 Martin Liska <mliska@suse.cz>
5676 PR gcov-profile/82633
5677 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
5678 their interaction with GCOV infrastructure.
5679 * configure.ac: Add -fkeep-{inline,static}-functions to
5681 * configure: Regenerate.
5683 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
5686 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
5688 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
5691 * config/rs6000/rs6000.md (allocate_stack): Force update interval
5692 into a register if it does not fit into an immediate offset field.
5694 2017-10-31 Olivier Hainque <hainque@adacore.com>
5696 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
5698 2017-10-31 Julia Koval <julia.koval@intel.com>
5700 * config.gcc: Add gfniintrin.h.
5701 * config/i386/gfniintrin.h: New.
5702 * config/i386/i386-builtin-types.def
5703 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5704 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5705 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5706 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5707 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5708 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
5709 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
5710 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
5711 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
5712 V64QI_FTYPE_V64QI_V64QI_INT): New types.
5713 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5714 * config/i386/immintrin.h: Include gfniintrin.h.
5715 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
5717 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
5719 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
5721 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5723 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
5724 (arm_ashldi3_1bit): Remove pattern.
5725 (ashrdi3): Remove shift by 1 expansion.
5726 (arm_ashrdi3_1bit): Remove pattern.
5727 (lshrdi3): Remove shift by 1 expansion.
5728 (arm_lshrdi3_1bit): Remove pattern.
5729 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
5730 cost of ashldi3 by 1.
5731 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
5732 (<shift>di3_neon): Likewise.
5734 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5736 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
5737 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
5738 and (*aarch64_simd_mov<VQ:mode>).
5739 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
5740 pattern alternative.
5741 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
5742 attributes to match pattern alternative.
5744 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
5746 * config.gcc (powerpc*-*-*): Add emmintrin.h.
5747 * config/rs6000/emmintrin.h: New file.
5748 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
5750 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
5752 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
5753 * (movdi_vfp_cortexa8): Remove pattern.
5755 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5757 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
5759 (Specific, alpha*-dec-osf5.1): Remove.
5760 (Specific, mips-sgi-irix5): Remove.
5761 (Specific, mips-sgi-irix6): Remove.
5763 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5766 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
5767 arguments to clear_bit_region_be.
5769 2017-10-30 Jim Wilson <wilson@tuliptree.org>
5771 * gimplify.c: Include memmodel.h.
5773 2017-10-30 Martin Jambor <mjambor@suse.cz>
5775 * omp-grid.c (grid_attempt_target_gridification): Also insert a
5776 condition whether loop should be executed at all.
5778 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
5780 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5781 gimple folding of vec_madd() intrinsics.
5782 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
5783 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
5784 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
5786 2017-10-30 Richard Biener <rguenther@suse.de>
5788 PR tree-optimization/82762
5790 2017-10-23 Richard Biener <rguenther@suse.de>
5792 PR tree-optimization/82129
5794 2017-08-01 Richard Biener <rguenther@suse.de>
5796 PR tree-optimization/81181
5797 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5798 (compute_antic): ... end of iteration here.
5800 2017-10-30 Joseph Myers <joseph@codesourcery.com>
5802 * doc/invoke.texi (C Dialect Options): Document -std=c17,
5803 -std=iso9899:2017 and -std=gnu17.
5804 * doc/standards.texi (C Language): Document C17 support.
5805 * doc/cpp.texi (Overview): Mention -std=c17.
5806 (Standard Predefined Macros): Document C11 and C17 values of
5807 __STDC_VERSION__. Do not refer to C99 support as incomplete.
5808 * doc/extend.texi (Inline): Do not list individual options for
5809 standards newer than C99.
5810 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
5812 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
5815 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
5817 * asan.c (asan_finish_file): Align asan globals array by shadow
5820 2017-10-30 Jakub Jelinek <jakub@redhat.com>
5823 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
5824 (struct store_immediate_info): Add bitregion_start and bitregion_end
5826 (store_immediate_info::store_immediate_info): Add brs and bre
5827 arguments and initialize bitregion_{start,end} from those.
5828 (struct merged_store_group): Add bitregion_start, bitregion_end,
5829 align_base and mask fields. Drop unnecessary struct keyword from
5830 struct store_immediate_info. Add do_merge method.
5831 (clear_bit_region_be): Use memset instead of loop storing zeros.
5832 (merged_store_group::do_merge): New method.
5833 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
5834 stores as long as the surrounding bitregions have no gaps.
5835 (merged_store_group::merge_overlapping): Use do_merge.
5836 (merged_store_group::apply_stores): Test that bitregion_{start,end}
5837 is byte aligned, rather than requiring that start and width are
5838 byte aligned. Drop unnecessary struct keyword from
5839 struct store_immediate_info. Allocate and populate also mask array.
5840 Make start of the arrays relative to bitregion_start rather than
5841 start and size them according to bitregion_{end,start} difference.
5842 (struct imm_store_chain_info): Drop unnecessary struct keyword from
5843 struct store_immediate_info.
5844 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
5845 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
5846 struct keyword from struct store_immediate_info.
5847 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
5848 between stores as long as the surrounding bitregions have no gaps.
5850 (struct split_store): Add orig non-static data member.
5851 (split_store::split_store): Initialize orig to false.
5852 (find_constituent_stmts): Return store_immediate_info *, non-NULL
5853 if there is exactly a single original stmt. Change stmts argument
5854 to pointer from reference, if NULL, don't push anything to it. Add
5855 first argument, use it to optimize skipping over orig stmts that
5856 are known to be before bitpos already. Simplify.
5857 (split_group): Return unsigned int count how many stores are or
5858 would be needed rather than a bool. Add allow_unaligned argument.
5859 Change split_stores argument from reference to pointer, if NULL,
5860 only do a dry run computing how many stores would be produced.
5861 Rewritten algorithm to use both alignment and misalign if
5862 !allow_unaligned and handle bitfield stores with gaps.
5863 (imm_store_chain_info::output_merged_store): Set start_byte_pos
5864 from bitregion_start instead of start. Compute allow_unaligned
5865 here, if true, do 2 split_group dry runs to compute which one
5866 produces fewer stores and prefer aligned if equal. Punt if
5867 new count is bigger or equal than original before emitting any
5868 statements, rather than during that. Remove no longer needed
5869 new_ssa_names tracking. Replace num_stmts with
5870 split_stores.length (). Use 32-bit stack allocated entries
5871 in split_stores auto_vec. Try to reuse original store lhs/rhs1
5872 if possible. Handle bitfields with gaps.
5873 (pass_store_merging::execute): Ignore bitsize == 0 stores.
5874 Compute bitregion_{start,end} for the stores and construct
5875 store_immediate_info with that. Formatting fixes.
5877 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
5880 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
5881 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
5883 2017-10-29 Jim Wilson <wilson@tuliptree.org>
5885 * gimplify.c: Include tm_p.h.
5887 * common.opt (gcoff): Re-add as ignored option.
5888 (gcoff1, gcoff2, gcoff3): Likewise.
5890 * Makefile.in (OBJS): Delete sdbout.o.
5891 (GTFILES): Delete $(srcdir)/sdbout.c.
5892 * debug.h: Delete sdb_debug_hooks.
5893 * final.c: Delete sdbout.h include.
5894 (final_scan_insn): Delete SDB_DEBUG check.
5895 (rest_of_clean_state): Likewise.
5896 * output.h: Delete sdb_begin_function_line.
5899 * toplev.c: Delete sdbout.h include.
5900 (process_options): Delete SDB_DEBUG check.
5901 * tree-core.h (tree_type_common): Delete pointer field of
5903 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
5904 TYPE_SYMTAB_POINTER.
5905 * tree.h (TYPE_SYMTAB_POINTER): Delete.
5906 (TYPE_SYMTAB_IS_POINTER): Delete.
5907 (TYPE_SYMTAB_IS_DIE): Renumber.
5908 * xcoffout.c: Refer to former sdbout.c file.
5909 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
5911 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
5912 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
5913 (Debugging Options): Delete -gcoff.
5914 (-gstabs): Delete SDB reference.
5916 (-gcoff@var{level}): Delete.
5917 * doc/passes.texi (Debugging information output): Delete SDB and
5918 sdbout.c references.
5919 * doc/tm.texi: Regenerate.
5920 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
5921 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
5923 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
5924 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
5925 (SDB_DEBUGGING_INFO): Delete.
5926 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
5927 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
5928 * target.def (output_source_filename): Delete COFF reference.
5930 * common.opt (gcoff): Delete.
5931 (gxcoff+): Update Negative chain.
5932 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
5934 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
5935 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
5936 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
5938 (expand_function_start): Change sdb reference to past tense.
5939 (expand_function_end): Change sdb reference to past tense.
5940 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
5941 * opts.c (debug_type_names): Delete coff entry.
5942 (common_handle_option): Delete OPT_gcoff case.
5943 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
5945 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
5946 * config/cris/cris.h: Delete SDB reference in comment.
5947 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
5948 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
5949 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
5950 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
5952 (ix86_expand_prologue): Likewise.
5953 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
5954 * config/ia64/ia64.h: Likewise.
5955 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
5956 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
5958 * config/mmix/mmix.h: Likewise.
5959 * config/nds32/nds32.c: Likewise.
5960 * config/stormy/storym16.h: Likewise.
5961 * config/visium/visium.h: Likewise.
5962 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
5964 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
5966 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
5967 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
5968 FRAME_POINTER_REGNUM point at high end of local var area.
5970 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
5972 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
5973 Move comment around. Do not reset best_edge for a copiable
5974 destination if the copy would cause a partition change.
5975 (better_edge_p): Remove redundant check.
5977 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
5979 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
5981 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5983 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
5984 for math functions that have _Float<N> and _Float<N>X variants.
5985 (mathfn_built_in_2): Add support for math functions that have
5986 _Float<N> and _Float<N>X variants.
5987 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
5988 (expand_builtin_mathfn_ternary): Add support for fma with
5989 _Float<N> and _Float<N>X variants.
5990 (expand_builtin): Likewise.
5991 (fold_builtin_3): Likewise.
5992 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
5993 create math function _Float<N> and _Float<N>X variants as external
5995 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
5996 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
5997 the __builtin_ prefix and if not strict ansi, without the prefix.
5998 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
5999 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
6000 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
6001 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
6002 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
6003 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
6004 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
6005 function signatures for fma _Float<N> and _Float<N>X variants.
6006 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
6007 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
6008 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
6009 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
6010 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
6011 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
6012 * gencfn-macros.c (print_case_cfn): Add support for math functions
6013 that have _Float<N> and _Float<N>X variants.
6014 (print_define_operator_list): Likewise.
6015 (fltfn_suffixes): Likewise.
6017 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
6018 for math functions that have _Float<N> and _Float<N>X variants.
6019 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
6020 and _Float<N>X variants.
6021 (COPYSIGN): Likewise.
6024 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
6025 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
6027 (integer_valued_read_call_p): Likewise.
6028 * fold-const-call.c (fold_const_call_ss): Likewise.
6029 (fold_const_call_sss): Add support for copysign, fmin, and fmax
6030 _Float<N> and _Float<N>X variants.
6031 (fold_const_call_ssss): Add support for fma _Float<N> and
6032 _Float<N>X variants.
6033 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
6034 support for copysign and fma _Float<N> and _Float<N>X variants.
6035 (backprop::process_builtin_call_use): Likewise.
6036 * tree-call-cdce.c (can_test_argument_range); Add support for
6037 sqrt _Float<N> and _Float<N>X variants.
6038 (edom_only_function): Likewise.
6039 (get_no_error_domain): Likewise.
6040 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
6041 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
6042 copysign _Float<N> and _Float<N>X variants.
6043 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
6044 handled by machine independent code.
6045 (FMAF128): Likewise.
6046 * doc/cpp.texi (Common Predefined Macros): Document defining
6047 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
6048 fma _Float<N> and _Float<N>X variants.
6050 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
6053 * config/i386/i386-modes.def (CCFPU): Remove definition.
6054 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
6055 (ix86_cc_modes_compatible): Ditto.
6056 (ix86_expand_carry_flag_compare): Ditto.
6057 (ix86_expand_int_movcc): Ditto.
6058 (ix86_expand_int_addcc): Ditto.
6059 (ix86_reverse_condition): Ditto.
6060 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
6061 Return true/false for unordered/ordered fp comparisons.
6062 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
6063 (ix86_prepare_fp_compare_args): Update for rename.
6064 (ix86_expand_fp_compare): Update for rename. Generate unordered
6065 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
6066 (ix86_expand_sse_compare_and_jump): Ditto.
6067 * config/i386/predicates.md (fcmov_comparison_operator):
6068 Remove CCFPU mode handling.
6069 (ix86_comparison_operator): Ditto.
6070 (ix86_carry_flag_operator): Ditto.
6071 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
6072 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
6073 (*cmpu<mode>_cc_i387): Ditto.
6074 (FPCMP): Remove mode iterator.
6075 (unord): Remove mode attribute.
6076 (unord_subst): New define_subst transformation
6077 (unord): New define_subst attribute.
6079 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
6080 (*cmpi<unord>xf_i387): Ditto.
6081 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
6082 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
6083 using unord_subst transformation.
6084 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
6085 (round_saeonly): Also handle CCFP mode.
6086 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
6087 Remove UNSPEC_SAHF unspec handling.
6089 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6091 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
6093 2017-10-27 Jeff Law <law@redhat.com>
6095 * gimple-ssa-sprintf.c: Include domwalk.h.
6096 (class sprintf_dom_walker): New class, derived from dom_walker.
6097 (sprintf_dom_walker::before_dom_children): New function.
6098 (struct call_info): Moved into sprintf_dom_walker class
6099 (compute_formath_length, handle_gimple_call): Likewise.
6100 (sprintf_length::execute): Call the dominator walker rather
6101 than walking the statements.
6103 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
6104 gimple statement locations.
6105 (check_array_bounds): Corresponding changes. Get the statement's
6106 location directly from wi->stmt.
6108 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
6111 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
6113 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
6115 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
6116 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
6118 2017-10-27 Jakub Jelinek <jakub@redhat.com>
6121 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
6122 * config/i386/i386.c (maybe_get_pool_constant): Removed.
6123 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
6124 maybe_get_pool_constant.
6125 * config/i386/predicates.md (zero_extended_scalar_load_operand):
6128 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6130 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
6131 2.26 caveat. Update gas and gld versions.
6132 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
6135 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6137 * cgraph.h (set_malloc_flag): Declare.
6138 * cgraph.c (set_malloc_flag_1): New function.
6139 (set_malloc_flag): Likewise.
6140 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
6141 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
6143 (read_ipa_call_summary): Add support for reading is_return_callee.
6144 (write_ipa_call_summary): Stream is_return_callee.
6145 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
6146 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
6147 ipa-prop.h, ipa-fnsummary.h.
6148 (pure_const_names): Change to static.
6149 (malloc_state_e): Define.
6150 (malloc_state_names): Define.
6151 (funct_state_d): Add field malloc_state.
6152 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
6153 (check_retval_uses): New function.
6154 (malloc_candidate_p): Likewise.
6155 (analyze_function): Add support for malloc attribute.
6156 (pure_const_write_summary): Stream malloc_state.
6157 (pure_const_read_summary): Add support for reading malloc_state.
6158 (dump_malloc_lattice): New function.
6159 (propagate_malloc): New function.
6160 (warn_function_malloc): New function.
6161 (ipa_pure_const::execute): Call propagate_malloc and
6162 ipa_free_fn_summary.
6163 (pass_local_pure_const::execute): Add support for malloc attribute.
6164 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
6165 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
6167 2017-10-27 Martin Liska <mliska@suse.cz>
6169 PR gcov-profile/82457
6170 * doc/invoke.texi: Document that one needs a non-strict ISO mode
6171 for fork-like functions to be properly instrumented.
6173 2017-10-27 Richard Biener <rguenther@suse.de>
6176 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
6177 info when we redirected EH.
6179 2017-10-26 Michael Collison <michael.collison@arm.com>
6181 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
6183 (<optab>_trunchf<GPI:mode>2: New pattern.
6184 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
6185 * config/aarch64/iterators.md (wv): New mode attribute.
6186 (vf, VF): New mode attributes.
6187 (vgp, VGP): New mode attributes.
6188 (s): Update attribute with SImode and DImode prefixes.
6190 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6192 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
6193 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
6194 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
6195 (nios2_option_overide): Initialize it. Don't allow R0-relative
6196 addressing with PIC.
6197 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
6198 (nios2_symbolic_constant_p): Likewise.
6199 (nios2_legitimate_address_p): Likewise.
6200 (nios2_r0rel_section_name_p): New.
6201 (nios2_symbol_ref_in_r0rel_data_p): New.
6202 (nios2_emit_move_sequence): Handle r0rel_constant_p.
6203 (r0rel_constant_p): New.
6204 (nios2_print_operand_address): Handle r0rel_constant_p.
6205 (nios2_cdx_narrow_form_p): Likewise.
6206 * config/nios2/nios2.opt (mr0rel-sec=): New option.
6207 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
6208 (Nios II Options): Document -mr0rel-sec.
6210 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
6212 * config/nios2/nios2.c: Include xregex.h.
6213 (nios2_gprel_sec_regex): New.
6214 (nios2_option_overide): Initialize it. Don't allow GP-relative
6215 addressing with PIC.
6216 (nios2_small_section_name_p): Check for regex match.
6217 * config/nios2/nios2.opt (mgprel-sec=): New option.
6218 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
6219 (Nios II Options): Document -mgprel-sec.
6221 2017-10-26 Jim Wilson <wilson@tuliptree.org>
6223 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
6225 2017-10-26 Tom de Vries <tom@codesourcery.com>
6227 PR tree-optimization/82707
6228 * gimple.c (gimple_copy): Fix unsharing of
6229 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
6231 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
6233 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
6234 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
6235 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
6236 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
6237 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
6238 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
6239 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
6240 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
6241 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
6242 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
6243 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
6244 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
6245 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
6246 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
6247 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
6248 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
6249 _mm512_mask_cmpunord_ps_mask): New intrinsics.
6251 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6253 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
6255 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
6256 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
6257 warning to rs6000.c. Remove the Undocumented flag, since it has
6259 (-mabi=ibmlongdouble): Likewise.
6260 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
6261 already set, set the default format for long double.
6262 (rs6000_debug_reg_global): Print whether long double is IBM or
6264 (rs6000_option_override_internal): Rework setting long double
6265 format. Only warn if the user is changing the long double default
6266 and they did not use -Wno-psabi.
6267 * doc/invoke.texi (PowerPC options): Update the documentation for
6268 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
6270 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6271 Alan Hayward <alan.hayward@arm.com>
6272 David Sherwood <david.sherwood@arm.com>
6274 * rtl.h (wider_subreg_mode): New function.
6275 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
6276 rather than an unsigned int *.
6277 * ira-color.c (regno_max_ref_width): Replace with...
6278 (regno_max_ref_mode): ...this new variable.
6279 (coalesced_pseudo_reg_slot_compare): Update accordingly.
6280 Use wider_subreg_mode.
6281 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
6282 rather than an unsigned int *.
6283 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
6284 (process_alt_operands): Likewise.
6285 (invariant_p): Likewise.
6286 * lra-spills.c (assign_mem_slot): Likewise.
6287 (add_pseudo_to_slot): Likewise.
6288 * lra.c (collect_non_operand_hard_regs): Likewise.
6289 (add_regs_to_insn_regno_info): Likewise.
6290 * reload1.c (regno_max_ref_width): Replace with...
6291 (regno_max_ref_mode): ...this new variable.
6292 (reload): Update accordingly. Update call to
6293 ira_sort_regnos_for_alter_reg.
6294 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
6295 (init_eliminable_invariants): Update to use regno_max_ref_mode.
6296 (scan_paradoxical_subregs): Likewise.
6298 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6300 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
6301 (aarch64_frame): Add emit_frame_chain boolean.
6302 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6303 Move eh_return case to aarch64_layout_frame.
6304 (aarch64_layout_frame): Initialize emit_frame_chain.
6305 (aarch64_expand_prologue): Use emit_frame_chain.
6307 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6309 * config/aarch64/aarch64.c (aarch64_layout_frame):
6310 Ensure LR is always stored at the bottom of the callee-saves.
6311 Remove rarely used frame layout which saves callee-saves at top of
6312 frame, so the store of LR can be used as a valid probe in all cases.
6314 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
6316 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
6317 Improve unaligned TImode/TFmode base/offset split.
6319 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6320 Alan Hayward <alan.hayward@arm.com>
6321 David Sherwood <david.sherwood@arm.com>
6323 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
6324 * combine.c (find_single_use_1): Likewise.
6325 (expand_field_assignment): Likewise.
6326 (move_deaths): Likewise.
6327 * lra-constraints.c (simplify_operand_subreg): Likewise.
6328 (curr_insn_transform): Likewise.
6329 * lra.c (collect_non_operand_hard_regs): Likewise.
6330 (add_regs_to_insn_regno_info): Likewise.
6331 * rtlanal.c (reg_referenced_p): Likewise.
6332 (covers_regno_no_parallel_p): Likewise.
6334 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6336 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
6337 Don't print any bits outside the precision of the value.
6338 * wide-int.cc (test_printing): Add some new tests.
6340 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6342 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
6343 supports -xbrace_comment option.
6344 * configure: Regenerate.
6345 * config.in: Regenerate.
6346 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
6347 (ASM_CPU_SPEC): Use it.
6349 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
6351 * target.def (static_rtx_alignment): New hook.
6352 * targhooks.h (default_static_rtx_alignment): Declare.
6353 * targhooks.c (default_static_rtx_alignment): New function.
6354 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
6355 * doc/tm.texi: Regenerate.
6356 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
6357 instead of targetm.constant_alignment. Remove call to
6359 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6360 (cris_preferred_mininum_alignment): New function, split out from...
6361 (cris_constant_alignment): ...here.
6362 (cris_static_rtx_alignment): New function.
6363 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
6365 (ix86_constant_alignment): ...here.
6366 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6367 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6368 (mmix_static_rtx_alignment): New function.
6369 * config/spu/spu.c (spu_static_rtx_alignment): New function.
6370 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
6372 2017-10-26 Tamar Christina <tamar.christina@arm.com>
6375 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
6376 Add flag_trapping_math and flag_fp_int_builtin_inexact.
6378 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
6380 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
6381 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
6382 mark as a sign-extending load.
6383 (local_pic_load_u): Define.
6385 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
6388 * fold-const.c (operand_equal_for_comparison_p): Also return true
6389 if ARG0 is a simple variant of ARG1 with narrower precision.
6390 (fold_ternary_loc): Always pass unstripped operands to the predicate.
6392 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
6394 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
6396 * i386.h (processor_costs): Add gather_static, gather_per_elt,
6397 scatter_static, scatter_per_elt.
6398 * x86-tune-costs.h: Add new cost entries.
6400 2017-10-25 Richard Biener <rguenther@suse.de>
6402 * tree-ssa-sccvn.h (vn_eliminate): Declare.
6403 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
6404 class pass_fre): Move to ...
6405 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
6406 class pass_fre): ... here and adjust for statistics.
6408 2017-10-25 Jakub Jelinek <jakub@redhat.com>
6411 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
6412 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
6413 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
6414 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
6417 2017-10-25 Richard Biener <rguenther@suse.de>
6419 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
6420 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
6421 eliminate_push_avail, eliminate_insert): Move inside...
6422 (class eliminate_dom_walker): ... this class in preparation
6424 (fini_eliminate): Remove by merging with ...
6425 (eliminate): ... this function. Adjust for class changes.
6426 (pass_pre::execute): Remove fini_eliminate call.
6427 (pass_fre::execute): Likewise.
6429 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6432 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
6433 (VPERMI2, VPERMI2I): New mode iterators.
6434 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
6435 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
6437 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
6438 mode iterator. Remove 3 old define_insn patterns.
6439 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
6440 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
6441 VPERMI2 mode iterator, remove the other two expanders.
6442 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
6443 to use VPERMI2 mode iterator, add another alternative for vpermi2*
6444 instructions, remove the other two patterns.
6445 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
6446 mode iterator, remove the other two patterns.
6447 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
6448 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
6449 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
6450 and adjust argument order accordingly.
6451 (ix86_expand_vec_perm): Adjust caller.
6452 (expand_vec_perm_1): Likewise.
6453 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
6454 (expand_vec_perm_vpermt2_vpshub2): ... this.
6455 (ix86_expand_vec_perm_const_1): Adjust caller.
6456 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
6459 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
6460 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
6461 (vec_shl_<mode>): Remove unused expander.
6462 (avx512bw_<shift_insn><mode>3): New define_insn.
6463 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
6464 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
6466 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
6469 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
6472 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6474 PR rtl-optimization/82396
6475 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
6476 (autopref_multipass_init): Simplify initialization.
6477 (autopref_rank_data): Simplify sort order.
6478 * gcc/sched-int.h (autopref_multipass_data_): Remove
6479 multi_mem_insn_p, min_offset and max_offset.
6481 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
6484 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6485 Check special value of flag_omit_frame_pointer.
6486 (aarch64_can_eliminate): Likewise.
6487 (aarch64_override_options_after_change_1): Simplify handling of
6488 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
6490 2017-10-24 Richard Biener <rguenther@suse.de>
6492 PR tree-optimization/82697
6493 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
6494 zero for conditional load and unconditional store.
6496 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6498 * doc/install.texi: Document bootstrap-cet.
6500 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
6503 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
6504 ENDBR instruction at function entrance if function is only
6507 2017-10-24 Jakub Jelinek <jakub@redhat.com>
6510 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
6511 patterns to better describe from which operation the CF is computed.
6512 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
6513 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
6514 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
6515 is 0, use _0 suffixed expanders instead of emitting a comparison
6518 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
6520 * config/i386/i386.md(*movsf_internal, *movdf_internal):
6521 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
6523 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
6526 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
6527 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
6528 * loop-iv.c (iv_get_reaching_def): Likewise.
6529 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
6530 variable is promoted and the partition contains undefined values.
6532 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6534 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
6536 (nios2_address_cost): Define.
6537 (nios2_legitimize_address): Recognize (exp + constant) directly.
6538 (TARGET_ADDRESS_COST): Define.
6540 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6542 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
6543 (nios2_symbolic_memory_operand_p): Declare.
6544 (nios2_split_large_constant): Declare.
6545 (nios2_split_symbolic_memory_operand): Declare.
6546 * config/nios2/nios2.c: Adjust includes.
6547 (nios2_symbolic_constant_allowed): New.
6548 (nios2_symbolic_constant_p): New.
6549 (nios2_plus_symbolic_constant_p): New.
6550 (nios2_valid_addr_expr_p): Recognize addresses involving
6552 (nios2_legitimate_address_p): Likewise, also LO_SUM.
6553 (nios2_symbolic_memory_operand_p): New.
6554 (nios2_large_constant_p): New.
6555 (nios2_split_large_constant): New.
6556 (nios2_split_plus_large_constant): New.
6557 (nios2_split_symbolic_memory_operand): New.
6558 (nios2_legitimize_address): Code refactoring. Handle addresses
6559 involving symbolic constants.
6560 (nios2_emit_move_sequence): Likewise.
6561 (nios2_print_operand): Improve error output.
6562 (nios2_print_operand_address): Handle LO_SUM.
6563 (nios2_cdx_narrow_form_p): Likewise.
6564 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
6565 operands involving symbolic constants.
6566 (movhi_internal, movsi_internal): Likewise.
6567 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
6568 (extendhisi2, extendqi<mode>2): Likewise.
6570 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6572 * tree-pass.h (PROP_rtl_split_insns): Define.
6573 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
6575 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
6577 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
6579 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6582 * target.def (const_not_ok_for_debug_p): Default to
6583 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
6584 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
6585 * targhooks.c (default_const_not_ok_for_debug_p): New function.
6586 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
6587 which targetm.const_not_ok_for_debug_p returned true.
6588 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
6590 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
6592 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
6593 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
6594 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
6596 (ix86_const_not_ok_for_debug_p): New function.
6597 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
6598 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
6600 2017-10-23 David Malcolm <dmalcolm@redhat.com>
6603 * system.h: Conditionally include "unique-ptr.h" if
6604 INCLUDE_UNIQUE_PTR is defined.
6605 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
6606 of defining INCLUDE_UNIQUE_PTR before including "system.h".
6608 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
6610 * config/rl78/rl78.md: New define_expand "subdi3".
6612 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
6615 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
6616 DF_REF_INSN if DF_REF_INSN_INFO is false.
6618 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
6620 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
6621 xmm_move instead of sse_move.
6622 (sse_store_index): New function.
6623 (ix86_register_move_cost): Be more sensible about mismatch stall;
6624 model AVX moves correctly; make difference between sse->integer and
6626 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
6627 moves; make difference between SSE and AVX.
6628 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
6629 and zmm_move. Increase size of sse load and store tables;
6630 add unaligned load and store tables; add ssemmx_to_integer.
6631 * x86-tune-costs.h: Update all entries according to real
6632 move latencies from Agner Fog's manual and chip documentation.
6634 2017-10-23 Jakub Jelinek <jakub@redhat.com>
6637 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
6638 * config/i386/constraints.md (Wf): New constraint.
6639 * config/i386/i386.md (UNSPEC_SBB): New unspec.
6640 (cmp<dwi>_doubleword): Removed.
6641 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
6642 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
6643 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
6644 expand with cmp<dwi>_doubleword. For LTU and GEU use
6645 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
6647 * common.opt (gcolumn-info): Enable by default.
6648 * doc/invoke.texi (gcolumn-info): Document new default.
6650 2017-10-23 Richard Biener <rguenther@suse.de>
6652 PR tree-optimization/82672
6653 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
6654 Fold the stmt if we propagated into it.
6656 2017-10-23 Richard Biener <rguenther@suse.de>
6658 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
6659 (bitmap_remove_expr_from_set): ... this. All callers call this
6660 for non-constant values.
6661 (bitmap_set_subtract): Rename to...
6662 (bitmap_set_subtract_expressions): ... this. Adjust and
6664 (bitmap_set_contains_value): Remove superfluous check.
6665 (bitmap_set_replace_value): Inline into single caller ...
6666 (bitmap_value_replace_in_set): ... here and simplify.
6667 (dependent_clean): Merge into ...
6668 (clean): ... this using an overload. Adjust.
6669 (prune_clobbered_mems): Adjust.
6670 (compute_antic_aux): Likewise.
6671 (compute_partial_antic_aux): Likewise.
6673 2017-10-23 Richard Biener <rguenther@suse.de>
6675 PR tree-optimization/82129
6677 2017-08-01 Richard Biener <rguenther@suse.de>
6679 PR tree-optimization/81181
6680 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
6681 (compute_antic): ... end of iteration here.
6683 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6685 * target.def (starting_frame_offset): New hook.
6686 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
6687 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
6688 * doc/tm.texi.in: Regenerate.
6689 * hooks.h (hook_hwi_void_0): Declare.
6690 * hooks.c (hook_hwi_void_0): New function.
6691 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
6692 STARTING_FRAME_OFFSET.
6693 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
6694 * reload1.c (reload): Likewise.
6695 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
6696 instead of STARTING_FRAME_OFFSET.
6697 * function.c (try_fit_stack_local): Likewise.
6698 (assign_stack_local_1): Likewise
6699 (instantiate_virtual_regs): Likewise.
6700 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
6701 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
6702 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
6703 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
6704 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
6705 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
6706 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
6707 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
6708 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
6709 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
6710 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
6711 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
6712 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
6713 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
6714 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
6715 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
6716 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
6717 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
6718 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
6719 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
6720 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
6721 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
6722 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
6723 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
6724 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
6725 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
6726 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
6727 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
6728 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
6729 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
6730 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
6731 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
6732 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
6733 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
6734 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
6735 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
6736 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
6737 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
6738 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
6739 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
6740 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
6741 * config/avr/avr.c (avr_starting_frame_offset): Make static and
6742 return a HOST_WIDE_INT.
6743 (avr_builtin_setjmp_frame_value): Use it instead of
6744 STARTING_FRAME_OFFSET.
6745 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6746 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
6747 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
6749 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6750 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
6751 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
6752 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6753 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
6754 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
6755 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6756 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
6757 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
6758 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6759 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
6760 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
6762 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6763 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
6764 * config/mips/mips.c (mips_compute_frame_info): Refer to
6765 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6766 (mips_starting_frame_offset): New function.
6767 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6768 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
6769 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
6770 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
6771 and return a HOST_WIDE_INT.
6772 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6773 (mmix_initial_elimination_offset): Refer to
6774 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
6775 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
6776 * config/pa/pa.c (pa_starting_frame_offset): New function.
6777 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
6778 (pa_expand_prologue): Likewise.
6779 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6780 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
6781 !FRAME_GROWS_DOWNWARD handling to...
6782 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6783 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
6784 !FRAME_GROWS_DOWNWARD handling to...
6785 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6786 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
6787 !FRAME_GROWS_DOWNWARD handling to...
6788 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6789 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
6791 (rs6000_starting_frame_offset): New function.
6792 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
6793 !FRAME_GROWS_DOWNWARD handling to...
6794 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6795 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
6796 !FRAME_GROWS_DOWNWARD handling to...
6797 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6798 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
6799 !FRAME_GROWS_DOWNWARD handling to...
6800 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
6801 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
6802 (rs6000_starting_frame_offset): New function.
6803 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
6804 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
6805 * config/vax/vax.c (vax_starting_frame_offset): New function.
6806 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
6807 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6808 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
6809 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
6810 (TARGET_STARTING_FRAME_OFFSET): Redefine.
6811 * system.h (STARTING_FRAME_OFFSET): Poison.
6813 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6815 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
6816 SCALAR_TYPE_MODE instead of TYPE_MODE.
6818 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6819 Alan Hayward <alan.hayward@arm.com>
6820 David Sherwood <david.sherwood@arm.com>
6822 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
6824 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6825 Alan Hayward <alan.hayward@arm.com>
6826 David Sherwood <david.sherwood@arm.com>
6828 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
6830 2017-10-23 Richard Biener <rguenther@suse.de>
6832 PR tree-optimization/82129
6833 * tree-ssa-pre.c (bitmap_set_and): Remove.
6834 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
6835 canonicalizing expressions in the set to those with lowest
6836 ID rather than taking that from the first edge.
6838 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
6840 * combine.c (rtx_equal_for_field_assignment_p): Use
6841 byte_lowpart_offset.
6843 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6844 Alan Hayward <alan.hayward@arm.com>
6845 David Sherwood <david.sherwood@arm.com>
6847 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
6848 to a SUBREG_PROMOTED_VAR.
6850 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6851 Alan Hayward <alan.hayward@arm.com>
6852 David Sherwood <david.sherwood@arm.com>
6854 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
6855 (expand_debug_source_expr): Likewise.
6856 * combine.c (combine_simplify_rtx): Likewise.
6857 * cse.c (fold_rtx): Likewise.
6858 * optabs.c (expand_float): Likewise.
6859 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6860 (simplify_binary_operation_1): Likewise.
6862 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6863 Alan Hayward <alan.hayward@arm.com>
6864 David Sherwood <david.sherwood@arm.com>
6866 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
6867 (record_promoted_value): Likewise.
6868 * expr.c (expand_expr_real_2): Likewise.
6869 * ree.c (update_reg_equal_equiv_notes): Likewise.
6870 (combine_set_extension): Likewise.
6871 * rtlanal.c (low_bitmask_len): Likewise.
6872 * simplify-rtx.c (neg_const_int): Likewise.
6873 (simplify_binary_operation_1): Likewise.
6875 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6876 Alan Hayward <alan.hayward@arm.com>
6877 David Sherwood <david.sherwood@arm.com>
6879 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
6880 * regcprop.c (maybe_mode_change): Likewise.
6881 * reload1.c (alter_reg): Likewise.
6883 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6885 * inchash.h (inchash::hash::add_wide_int): New function.
6886 * lto-streamer-out.c (hash_tree): Use it.
6888 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
6890 * inchash.h (inchash::hash::add_wide_int): Rename to...
6891 (inchash::hash::add_hwi): ...this.
6892 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
6893 (polymorphic_call_target_hasher::hash): Likewise.
6894 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
6895 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
6896 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
6897 * lto-streamer-out.c (hash_tree): Likewise.
6898 * optc-save-gen.awk: Likewise.
6899 * tree.c (add_expr): Likewise.
6901 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6904 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
6905 for ordered inequality comparisons even with TARGET_IEEE_FP.
6907 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
6910 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
6911 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
6912 Expand with cmp<dwi>_doubleword.
6914 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6916 * extend.texi: Add x86 specific to 'nocf_check' attribute.
6917 List CET intrinsics.
6918 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
6919 specific to -fcf-protection option.
6921 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6923 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
6924 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
6925 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
6926 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
6927 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
6928 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
6929 (extra_objs): Add cet.o for Linux/x86 targets.
6930 (tmake_file): Add i386/t-cet for Linux/x86 targets.
6931 * config/i386/cet.c: New file.
6932 * config/i386/cetintrin.h: Likewise.
6933 * config/i386/t-cet: Likewise.
6934 * config/i386/cpuid.h (bit_SHSTK): New.
6935 (bit_IBT): Likewise.
6936 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
6937 pass IBT and SHSTK bits.
6938 * config/i386/i386-builtin-types.def
6939 (VOID_FTYPE_UNSIGNED_PVOID): New.
6940 (VOID_FTYPE_UINT64_PVOID): Likewise.
6941 * config/i386/i386-builtin.def: Add CET intrinsics.
6942 * config/i386/i386-c.c (ix86_target_macros_internal): Add
6943 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
6944 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
6945 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
6947 * config/i386/i386.c (rest_of_insert_endbranch): New.
6948 (pass_data_insert_endbranch): Likewise.
6949 (pass_insert_endbranch): Likewise.
6950 (make_pass_insert_endbranch): Likewise.
6951 (ix86_notrack_prefixed_insn_p): Likewise.
6952 (ix86_target_string): Add -mibt, -mshstk flags.
6953 (ix86_option_override_internal): Add flag_cf_protection
6955 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
6956 (ix86_print_operand): Add 'notrack' prefix output.
6957 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
6958 (ix86_expand_builtin): Expand CET intrinsics.
6959 (x86_output_mi_thunk): Add 'endbranch' instruction.
6960 * config/i386/i386.h (TARGET_IBT): New.
6961 (TARGET_IBT_P): Likewise.
6962 (TARGET_SHSTK): Likewise.
6963 (TARGET_SHSTK_P): Likewise.
6964 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
6965 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
6966 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
6967 (builtin_setjmp_setup): New pattern.
6968 (builtin_longjmp): Likewise.
6969 (rdssp<mode>): Likewise.
6970 (incssp<mode>): Likewise.
6971 (saveprevssp): Likewise.
6972 (rstorssp): Likewise.
6973 (wrss<mode>): Likewise.
6974 (wruss<mode>): Likewise.
6975 (setssbsy): Likewise.
6976 (clrssbsy): Likewise.
6977 (nop_endbr): Likewise.
6978 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
6980 * config/i386/immintrin.h: Include <cetintrin.h>.
6981 * config/i386/linux-common.h
6982 (file_end_indicate_exec_stack_and_cet): New prototype.
6983 (TARGET_ASM_FILE_END): New.
6985 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
6987 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
6988 latencies instead of having separate table; make difference between
6989 integer and float costs.
6990 * i386.h (processor_costs): Remove scalar_stmt_cost,
6991 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
6992 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
6994 * x86-tune-costs.h: Remove entries which has been removed in
6995 procesor_costs from all tables; make cond_taken_branch_cost
6996 and cond_not_taken_branch_cost COST_N_INSNS based.
6998 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
7000 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
7002 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7004 * config/i386/i386.md (isa): Remove fma_avx512f.
7005 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
7006 <avx512>_fmadd_<mode>_mask3<round_name>,
7007 <avx512>_fmsub_<mode>_mask<round_name>,
7008 <avx512>_fmsub_<mode>_mask3<round_name>,
7009 <avx512>_fnmadd_<mode>_mask<round_name>,
7010 <avx512>_fnmadd_<mode>_mask3<round_name>,
7011 <avx512>_fnmsub_<mode>_mask<round_name>,
7012 <avx512>_fnmsub_<mode>_mask3<round_name>,
7013 <avx512>_fmaddsub_<mode>_mask<round_name>,
7014 <avx512>_fmaddsub_<mode>_mask3<round_name>,
7015 <avx512>_fmsubadd_<mode>_mask<round_name>,
7016 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
7017 (*vec_widen_umult_even_v16si<mask_name>,
7018 *vec_widen_smult_even_v16si<mask_name>): Likewise.
7019 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
7021 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7023 * extend.texi: Add 'nocf_check' documentation.
7024 * gimple.texi: Add second parameter to
7025 gimple_build_call_from_tree.
7026 * invoke.texi: Add -fcf-protection documentation.
7027 * rtl.texi: Add REG_CALL_NOTRACK documenation.
7029 2017-10-20 Richard Biener <rguenther@suse.de>
7031 PR tree-optimization/82473
7032 * tree-vect-loop.c (vectorizable_reduction): Properly get at
7033 the largest input type.
7035 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
7037 * c-attribs.c (handle_nocf_check_attribute): New function.
7038 (c_common_attribute_table): Add 'nocf_check' handling.
7039 * gimple-parser.c: Add second argument NULL to
7040 gimple_build_call_from_tree.
7041 * attrib.c (comp_type_attributes): Check nocf_check attribute.
7042 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
7044 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
7045 * common.opt: Add fcf-protection flag.
7046 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
7047 * flag-types.h: Add enum cf_protection_level.
7048 * gimple.c (gimple_build_call_from_tree): Add second parameter.
7049 Add 'nocf_check' attribute propagation to gimple call.
7050 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
7051 (gimple_build_call_from_tree): Update prototype.
7052 (gimple_call_nocf_check_p): New function.
7053 (gimple_call_set_nocf_check): Likewise.
7054 * gimplify.c: Add second argument to gimple_build_call_from_tree.
7055 * ipa-icf.c: Add nocf_check attribute in statement hash.
7056 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
7057 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
7058 * toplev.c (process_options): Add flag_cf_protection handling.
7060 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7062 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
7064 2017-10-20 Richard Biener <rguenther@suse.de>
7066 PR tree-optimization/82603
7067 * tree-if-conv.c (predicate_mem_writes): Make sure to only
7068 remove false predicated stores.
7070 2017-10-20 Richard Biener <rguenther@suse.de>
7072 * graphite-isl-ast-to-gimple.c
7073 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
7074 Remove return value and simplify, dump copied stmt after lhs
7076 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
7077 Reduce dump verbosity.
7078 (gsi_insert_earliest): Likewise.
7079 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
7080 * graphite.c (print_global_statistics): Adjust dumping.
7081 (print_graphite_scop_statistics): Likewise.
7082 (print_graphite_statistics): Do not dump loops here.
7083 (graphite_transform_loops): But here.
7085 2017-10-20 Nicolas Roche <roche@adacore.com>
7087 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
7088 * configure: Regenerate.
7090 2017-10-20 Jakub Jelinek <jakub@redhat.com>
7093 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
7094 functions when optimizing replace GIMPLE_RETURN stmts with
7095 calls to __builtin_unreachable ().
7098 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
7099 for -fsanitize=thread link of executables.
7100 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
7101 link of executables.
7104 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
7106 (<shift_insn><mode>3<mask_name>): Change the last of the 3
7107 define_insns for logical vector shifts to use VI248_AVX512BW
7108 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
7109 condition, useless isa and prefix attributes. Change the first
7110 2 of these define_insns to ...
7111 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
7112 define_insn for avx512vl.
7113 (<shift_insn><mode>3): ... and this, new define_insn without
7114 masking for non-avx512vl.
7117 * config/i386/sse.md (*andnot<mode>3,
7118 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
7119 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
7120 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
7121 not applied use empty suffix even for TARGET_AVX512VL.
7122 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
7123 is applied, supply evex,evex or evex,evex,evex instead of just
7126 2017-10-20 Julia Koval <julia.koval@intel.com>
7128 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
7129 (OPTION_MASK_ISA_GFNI_UNSET): New.
7130 (ix86_handle_option): Handle OPT_mgfni.
7131 * config/i386/cpuid.h (bit_GFNI): New.
7132 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
7133 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
7134 * config/i386/i386.c (ix86_target_string): Add -mgfni.
7135 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
7136 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
7137 * config/i386/i386.opt: Add mgfni.
7139 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
7141 * config/msp430/msp430.c (msp430_option_override): Disable
7142 -fdelete-null-pointer-checks.
7143 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
7145 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7147 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
7148 of x87 and SSE instructions.
7150 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
7152 * asan.c (create_cond_insert_point): Do not update edge count.
7153 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
7154 (afdo_propagate_circuit): Likewise.
7155 (afdo_calculate_branch_prob): Likewise.
7156 (afdo_annotate_cfg): Likewise.
7157 * basic-block.h (struct edge_def): Remove count.
7158 (edge_def::count): New accessor.
7159 * bb-reorder.c (rotate_loop): Update.
7160 (find_traces_1_round): Update.
7161 (connect_traces): Update.
7162 (sanitize_hot_paths): Update.
7163 * cfg.c (unchecked_make_edge): Update.
7164 (make_single_succ_edge): Update.
7165 (check_bb_profile): Update.
7166 (dump_edge_info): Update.
7167 (update_bb_profile_for_threading): Update.
7168 (scale_bbs_frequencies_int): Update.
7169 (scale_bbs_frequencies_gcov_type): Update.
7170 (scale_bbs_frequencies_profile_count): Update.
7171 (scale_bbs_frequencies): Update.
7172 * cfganal.c (connect_infinite_loops_to_exit): Update.
7173 * cfgbuild.c (compute_outgoing_frequencies): Update.
7174 (find_many_sub_basic_blocks): Update.
7175 * cfgcleanup.c (try_forward_edges): Update.
7176 (try_crossjump_to_edge): Update
7177 * cfgexpand.c (expand_gimple_cond): Update
7178 (expand_gimple_tailcall): Update
7179 (construct_exit_block): Update
7180 * cfghooks.c (verify_flow_info): Update
7181 (redirect_edge_succ_nodup): Update
7182 (split_edge): Update
7183 (make_forwarder_block): Update
7184 (duplicate_block): Update
7185 (account_profile_record): Update
7186 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
7187 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
7188 * cfgloopmanip.c (scale_loop_profile): Update.
7190 (lv_adjust_loop_entry_edge): Update.
7191 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7192 (force_nonfallthru_and_redirect): Update.
7193 (purge_dead_edges): Update.
7194 (rtl_flow_call_edges_add): Update.
7195 * cgraphunit.c (init_lowered_empty_function): Update.
7196 (cgraph_node::expand_thunk): Update.
7197 * gimple-pretty-print.c (dump_probability): Update.
7198 (dump_edge_probability): Update.
7199 * gimple-ssa-isolate-paths.c (isolate_path): Update.
7200 * haifa-sched.c (sched_create_recovery_edges): Update.
7201 * hsa-gen.c (convert_switch_statements): Update.
7202 * ifcvt.c (dead_or_predicable): Update.
7203 * ipa-inline-transform.c (inline_transform): Update.
7204 * ipa-split.c (split_function): Update.
7205 * ipa-utils.c (ipa_merge_profiles): Update.
7206 * loop-doloop.c (add_test): Update.
7207 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
7208 * lto-streamer-in.c (input_cfg): Update.
7209 (input_function): Update.
7210 * lto-streamer-out.c (output_cfg): Update.
7211 * modulo-sched.c (sms_schedule): Update.
7212 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
7213 * predict.c (maybe_hot_edge_p): Update.
7214 (unlikely_executed_edge_p): Update.
7215 (probably_never_executed_edge_p): Update.
7216 (dump_prediction): Update.
7217 (drop_profile): Update.
7218 (propagate_unlikely_bbs_forward): Update.
7219 (determine_unlikely_bbs): Update.
7220 (force_edge_cold): Update.
7221 * profile.c (compute_branch_probabilities): Update.
7222 * reg-stack.c (better_edge): Update.
7223 * shrink-wrap.c (handle_simple_exit): Update.
7224 * tracer.c (better_p): Update.
7225 * trans-mem.c (expand_transaction): Update.
7226 (split_bb_make_tm_edge): Update.
7227 * tree-call-cdce.c: Update.
7228 * tree-cfg.c (gimple_find_sub_bbs): Update.
7229 (gimple_split_edge): Update.
7230 (gimple_duplicate_sese_region): Update.
7231 (gimple_duplicate_sese_tail): Update.
7232 (gimple_flow_call_edges_add): Update.
7233 (insert_cond_bb): Update.
7234 (execute_fixup_cfg): Update.
7235 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
7236 * tree-complex.c (expand_complex_div_wide): Update.
7237 * tree-eh.c (lower_resx): Update.
7238 (unsplit_eh): Update.
7239 (cleanup_empty_eh_move_lp): Update.
7240 * tree-inline.c (copy_edges_for_bb): Update.
7241 (freqs_to_counts): Update.
7242 (copy_cfg_body): Update.
7243 * tree-ssa-dce.c (remove_dead_stmt): Update.
7244 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7245 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7246 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7247 (unloop_loops): Update.
7248 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7249 * tree-ssa-loop-split.c (connect_loops): Update.
7250 (split_loop): Update.
7251 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
7252 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7253 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7254 * tree-ssa-reassoc.c (branch_fixup): Update.
7255 * tree-ssa-tail-merge.c (replace_block_by): Update.
7256 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7257 (compute_path_counts): Update.
7258 (update_profile): Update.
7259 (recompute_probabilities): Update.
7260 (update_joiner_offpath_counts): Update.
7261 (estimated_freqs_path): Update.
7262 (freqs_to_counts_path): Update.
7263 (clear_counts_path): Update.
7264 (ssa_fix_duplicate_block_edges): Update.
7265 (duplicate_thread_path): Update.
7266 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7267 (case_bit_test_cmp): Update.
7268 (collect_switch_conv_info): Update.
7269 (gen_inbound_check): Update.
7270 (do_jump_if_equal): Update.
7271 (emit_cmp_and_jump_insns): Update.
7272 * tree-tailcall.c (decrease_profile): Update.
7273 (eliminate_tail_call): Update.
7274 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7275 (vect_do_peeling): Update.
7276 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7277 * ubsan.c (ubsan_expand_null_ifn): Update.
7278 (ubsan_expand_ptr_ifn): Update.
7279 * value-prof.c (gimple_divmod_fixed_value): Update.
7280 (gimple_mod_pow2): Update.
7281 (gimple_mod_subtract): Update.
7282 (gimple_ic): Update.
7283 (gimple_stringop_fixed_value): Update.
7285 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7288 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
7290 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
7292 PR rtl-optimization/82395
7293 * ira-color.c (allocno_priority_compare_func): Fix comparison step
7294 based on non_spilled_static_chain_regno_p.
7296 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
7298 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
7299 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
7300 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
7302 2017-10-19 Martin Sebor <msebor@redhat.com>
7304 PR tree-optimization/82596
7305 * tree.c (array_at_struct_end_p): Handle STRING_CST.
7307 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7309 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
7310 (get_mem_refs_of_builtin_call): Likewise.
7311 * builtins.c (expand_builtin_apply): Adjust call to
7312 allocate_dynamic_stack_space.
7313 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
7314 the third argument to allocate_dynamic_stack_space, otherwise -1.
7315 (expand_builtin): Deal with all alloca variants.
7316 (is_inexpensive_builtin): Likewise.
7317 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
7318 * calls.c (special_function_p): Deal with all alloca variants.
7319 (initialize_argument_information): Adjust call to
7320 allocate_dynamic_stack_space.
7321 (expand_call): Likewise.
7322 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
7323 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
7324 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
7325 use it for the stack usage computation.
7326 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
7327 * function.c (gimplify_parameters): Call build_alloca_call_expr.
7328 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
7329 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
7330 (in_loop_p): Remove first argument and useless check.
7331 (pass_walloca::execute): Remove useless test and adjust call to above.
7332 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
7333 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
7334 (gimplify_call_expr): Deal with all alloca variants.
7335 * hsa-gen.c (gen_hsa_alloca): Likewise.
7336 (gen_hsa_insns_for_call): Likewise.
7337 * ipa-pure-const.c (special_builtin_state): Likewise.
7338 * tree-chkp.c (chkp_build_returned_bound): Likewise.
7339 * tree-object-size.c (alloc_object_size): Likewise.
7340 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
7341 (call_may_clobber_ref_p_1): Likewise.
7342 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
7343 (ccp_fold_stmt): Likewise.
7344 (optimize_stack_restore): Likewise.
7345 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
7346 (mark_all_reaching_defs_necessary_1): Likewise.
7347 (propagate_necessity): Likewise.
7348 (eliminate_unnecessary_stmts): Likewise.
7349 * tree.c (build_common_builtin_nodes): Build
7350 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
7351 (build_alloca_call_expr): New function.
7352 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
7353 (CASE_BUILT_IN_ALLOCA): Likewise.
7354 (build_alloca_call_expr): Declare.
7355 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
7357 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
7360 * dwarf2out.c (new_die_raw): New static inline function.
7361 (new_die): Use it to create the DIE.
7362 (add_AT_external_die_ref): Likewise.
7363 (clone_die): Likewise.
7364 (clone_as_declaration): Likewise.
7365 (dwarf2out_vms_debug_main_pointer): Likewise.
7366 (base_type_die): Likewise. Remove early return for corner cases.
7367 Do not call add_pubtype on the DIE here.
7368 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
7369 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
7370 typedefs for base types with DW_AT_endianity. Make sure a DIE with
7371 native order exists for base types, attach the DIE manually and call
7372 add_pubtype on it. Do not equate a reverse order DIE to the type.
7374 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
7376 * config/arm/arm.c (align_ok_ldrd_strd): New function.
7377 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
7379 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
7381 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7383 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
7384 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
7385 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
7386 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
7387 * opts.c (sanitizer_opts): Add builtin.
7388 * ubsan.c (instrument_builtin): New function.
7389 (pass_ubsan::execute): Call it.
7390 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
7391 * doc/invoke.texi: Document -fsanitize=builtin.
7393 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
7394 builtins, store max (log2 (align), 0) into uchar field instead of
7395 align into uptr field.
7396 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
7397 store uchar 0 field instead of uptr 0 field.
7398 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
7399 instead of passing one address of struct with 2 locations pass
7400 two addresses of structs with 1 location each.
7401 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
7402 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
7403 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
7404 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
7405 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
7406 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
7407 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
7408 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
7410 2017-10-19 Martin Liska <mliska@suse.cz>
7413 * file-find.c (remove_prefix): Remove.
7414 * file-find.h (remove_prefix): Likewise.
7415 * gcc-ar.c: Remove smartness of lookup.
7417 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
7419 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
7420 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
7421 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
7423 2017-10-19 Jakub Jelinek <jakub@redhat.com>
7426 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
7427 (setcc + and to xor + setcc): New peephole2.
7429 2017-10-19 Tom de Vries <tom@codesourcery.com>
7431 * doc/sourcebuild.texi (Test Directives, Variants of
7432 dg-require-support): Add dg-require-stack-size.
7434 2017-10-19 Martin Liska <mliska@suse.cz>
7437 * gimplify.c (gimplify_decl_expr): Do not instrument variables
7438 that have a large alignment.
7439 (gimplify_target_expr): Likewise.
7441 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
7443 PR rtl-optimization/82602
7444 * ira.c (rtx_moveable_p): Return false for volatile asm.
7446 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
7449 * config/i386/i386-modes.def (CCGZ): New CC mode.
7450 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
7451 * config/i386/predicates.md (ix86_comparison_operator):
7453 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
7454 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
7455 with double-word subtraction.
7456 (put_condition_code): Handle CCGZmode.
7458 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
7460 * wide-int.cc (debug (const wide_int &)): New.
7461 (debug (const wide_int *)): New.
7462 (debug (const widest_int &)): New.
7463 (debug (const widest_int *)): New.
7465 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
7468 * lra-constraints.c (curr_insn_transform): Use non-input operand
7469 instead of output one for matched reload.
7471 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7473 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
7474 (tree-ssa-loop-ivopts.h): New header file.
7475 (struct builtin_info): New fields.
7476 (classify_builtin_1): Compute and record base and offset parts for
7477 memset builtin partition by calling strip_offset.
7478 (offset_cmp, fuse_memset_builtins): New functions.
7479 (finalize_partitions): Fuse adjacent memset partitions by calling
7481 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
7482 Expose the interface.
7483 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
7485 2017-10-18 Bin Cheng <bin.cheng@arm.com>
7487 PR tree-optimization/82574
7488 * tree-loop-distribution.c (find_single_drs): New parameter. Check
7489 that data reference must be executed exactly once per iteration
7490 against the outermost loop in nest.
7491 (classify_partition): Update call to above function.
7493 2017-10-18 Richard Biener <rguenther@suse.de>
7495 PR tree-optimization/82591
7496 * graphite.c (graphite_transform_loops): Move code gen message
7498 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7499 Here. Handle scop_to_isl_ast failing.
7500 (scop_to_isl_ast): Limit the number of ISL operations.
7502 2017-10-18 Richard Biener <rguenther@suse.de>
7504 * graphite-isl-ast-to-gimple.c
7505 (translate_isl_ast_to_gimple::set_rename): Simplify.
7506 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
7507 (graphite_copy_stmts_from_block): ... here.
7508 (copy_bb_and_scalar_dependences): Simplify.
7509 (add_parameters_to_ivs_params): Canonicalize.
7510 (generate_entry_out_of_ssa_copies): Simplify.
7511 * graphite-sese-to-poly.c (extract_affine_name): Simplify
7512 by passing in ISL dimension.
7513 (parameter_index_in_region_1): Rename to ...
7514 (parameter_index_in_region): ... this.
7515 (extract_affine): Adjust assert, pass down parameter index.
7516 (add_param_constraints): Use range-info when available.
7517 (build_scop_context): Adjust.
7518 * sese.c (new_sese_info): Adjust.
7519 (free_sese_info): Likewise.
7520 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
7521 Remove unused typedefs.
7522 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
7524 2017-10-18 Martin Liska <mliska@suse.cz>
7526 * combine.c (simplify_compare_const): Add gcc_fallthrough.
7528 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7530 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
7531 (s390_sched_init): Do not reset s390_sched_state if we entered the
7532 current basic block via a fallthru edge and all others are unlikely.
7534 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7536 * config/s390/s390.c (NUM_SIDES): New variable.
7537 (LONGRUNNING_THRESHOLD): New variable.
7538 (LATENCY_FACTOR): New variable.
7539 (s390_sched_score): Decrease score for long-running instructions on
7541 (s390_sched_variable_issue): Perform bookkeeping for long-running
7544 2017-10-18 Richard Biener <rguenther@suse.de>
7546 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
7547 Simplify with removal of the parameter rename map.
7548 (set_rename): Likewise.
7549 (should_copy_to_new_region): Likewise.
7550 (graphite_copy_stmts_from_block): Likewise.
7551 (copy_bb_and_scalar_dependences): Remove initialization of
7552 unused copied_bb_map.
7554 (copy_internal_parameters): Likewise.
7555 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
7556 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
7557 Use INTEGRAL_TYPE_P.
7558 (parameter_index_in_region_1): Rename to ...
7559 (assign_parameter_index_in_region): ... this. Assert we have
7560 a parameter we handle.
7561 (scan_tree_for_params): Adjust.
7562 * sese.h (parameter_rename_map_t): Remove.
7563 (struct sese_info_t): Remove unused parameter_rename_map and
7564 copied_bb_map members.
7565 * sese.c (new_sese_info): Adjust.
7566 (free_sese_info): Likewise.
7568 2017-10-18 Martin Liska <mliska@suse.cz>
7571 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
7572 on an abnormal edge.
7574 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
7576 * doc/invoke.texi (ffunction-sections and fdata-sections):
7579 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7581 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
7582 the use statement can throw internally.
7584 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
7586 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
7587 any RTX present on the RHS of a SET.
7588 * compare-elim.c (try_eliminate_compare): Restore comment.
7590 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7592 * langhooks.h (struct lang_hooks): Document that tree_size langhook
7593 may be also called on tcc_type nodes.
7594 * langhooks.c (lhd_tree_size): Likewise.
7596 2017-10-17 David Malcolm <dmalcolm@redhat.com>
7598 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
7599 format_warning_at_substring.
7600 (maybe_warn): Convert source_range * param to a location_t. Pass
7601 UNKNOWN_LOCATION rather than NULL to fmtwarn.
7602 (format_directive): Remove code to extract source_ranges and
7603 source_range * in favor of just a location_t.
7604 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
7606 * substring-locations.c (format_warning_va): Convert
7607 source_range * param to a location_t.
7608 (format_warning_at_substring): Likewise.
7609 * substring-locations.h (format_warning_va): Likewise.
7610 (format_warning_at_substring): Likewise.
7612 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
7614 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
7616 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
7617 and scatter/gather ops.
7619 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
7620 vec_gather_load and vec_scatter_store.
7621 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
7622 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
7623 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
7624 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
7625 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
7626 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
7628 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
7630 * reg-stack.c (compare_for_stack_reg): Add bool argument.
7631 Detect FTST instruction and handle its register pops. Only pop
7632 second operand if can_pop_second_op is true.
7633 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
7634 set can_pop_second_op to false in the compare_for_stack_reg call.
7636 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
7637 output_fp_compare for stack register operands.
7638 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
7639 instructions here. Do not emit stack register pops here. Assert
7640 that FCOMPP pops next to top stack register. Rewrite function.
7642 2017-10-17 Nathan Sidwell <nathan@acm.org>
7645 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
7646 use DECL_ASSEMBLER_NAME_RAW.
7649 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
7652 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
7653 Wilco Dijkstra <wilco.dijkstra@arm.com>
7655 * builtins.c (expand_builtin_update_setjmp_buf): Add a
7656 converstion to Pmode from the buf_addr.
7658 2017-10-17 Richard Biener <rguenther@suse.de>
7660 * graphite-dependences.c (scop_get_reads_and_writes): Change
7661 output parameters to references.
7663 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
7665 PR 71026/tree-optimization
7666 * fold-const.c (distribute_real_division): Removed.
7667 (fold_binary_loc): Remove calls to distribute_real_divison.
7669 2017-10-17 Richard Biener <rguenther@suse.de>
7671 * graphite-scop-detection.c
7672 (scop_detection::stmt_has_simple_data_refs_p): Always use
7673 the full nest as region.
7674 (try_generate_gimple_bb): Likewise.
7675 * sese.c (scalar_evolution_in_region): Simplify now that
7676 SCEV can handle instantiation in regions.
7677 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
7678 in the non-loop part of a function if requested.
7680 2017-10-17 Richard Biener <rguenther@suse.de>
7682 PR tree-optimization/82563
7683 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
7685 (graphite_regenerate_ast_isl): Call it.
7686 * graphite-scop-detection.c (build_scops): Remove entry edge split.
7688 2017-10-17 Jakub Jelinek <jakub@redhat.com>
7690 PR tree-optimization/82549
7691 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
7692 Formatting fixes. Instead of calling make_bit_field_ref with negative
7695 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
7697 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
7698 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
7699 _mm_maskz_reduce_ss): New.
7700 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
7701 __builtin_ia32_reducess_mask): Ditto..
7702 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
7703 * config/i386/sse.md (reduces<mode>): Renamed to ...
7704 (reduces<mode><mask_scalar_name>): ... this.
7705 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
7707 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
7708 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
7710 2017-10-16 David Malcolm <dmalcolm@redhat.com>
7712 * Makefile.in (OBJS): Add unique-ptr-tests.o.
7713 * selftest-run-tests.c (selftest::run_tests): Call
7714 selftest::unique_ptr_tests_cc_tests.
7715 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
7716 * unique-ptr-tests.cc: New file.
7718 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
7721 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
7723 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
7724 (make_hard_regno_born, make_hard_regno_dead): Update
7725 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
7727 2017-10-16 Jeff Law <law@redhat.com>
7729 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
7731 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
7733 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
7735 2017-10-16 Olivier Hainque <hainque@adacore.com>
7737 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
7738 with_cpu if we were configured for an e500v2 target cpu name.
7740 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
7742 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
7743 * doc/invoke.texi: Document +nodsp as a valid extension for
7746 2017-10-16 Martin Liska <mliska@suse.cz>
7748 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
7749 (test_set_range): Likewise.
7750 (test_range_functions): Rename to ...
7751 (test_bit_in_range): ... this.
7752 (sbitmap_c_tests): Add new test.
7754 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7756 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
7758 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
7759 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
7761 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7763 * config/aarch64/aarch64-builtins.c
7764 (aarch64_types_quadopu_lane_qualifiers): New.
7765 (TYPES_QUADOPU_LANE): New.
7766 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
7767 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
7768 (aarch64_<sur>dot_laneq<vsi2qi>): New.
7769 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
7770 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
7771 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
7772 (Vdottype, DOTPROD): New.
7773 (sur): Add SDOT and UDOT.
7775 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7777 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
7778 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
7779 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
7781 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
7782 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
7783 Enable TARGET_DOTPROD.
7784 (cortex-a75.cortex-a55): Likewise.
7785 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
7787 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7789 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
7790 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
7792 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
7794 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
7795 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
7796 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
7797 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
7798 * config/arm/types.md (neon_dot, neon_dot_q): New.
7799 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
7801 2017-10-16 Tamar Christina <tamar.christina@arm.com>
7803 * config/arm/arm.h (TARGET_DOTPROD): New.
7804 * config/arm/arm.c (arm_arch_dotprod): New.
7805 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
7806 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
7807 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
7808 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
7809 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
7810 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
7811 * doc/invoke.texi (armv8.2-a): Document dotprod
7813 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7815 * i386.c (ix86_vec_cost): New function.
7816 (ix86_rtx_costs): Handle vector operations better.
7817 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
7818 * x86-tune-costs.h: Add new costs to all tables.
7820 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
7822 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
7824 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
7825 divsd, sqrtss and sqrtsd
7826 * x86-tune-costs.h: Add new entries to all costs.
7827 (znver1_cost): Fix to match real instruction latencies.
7829 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7830 Michael Collison <michael.collison@arm.com>
7832 * compare-elim.c: Include emit-rtl.h.
7833 (can_merge_compare_into_arith): New function.
7834 (try_validate_parallel): Likewise.
7835 (try_merge_compare): Likewise.
7836 (try_eliminate_compare): Call the above when no previous clobber
7838 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
7841 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7845 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
7846 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
7850 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
7851 to be any operand_equal_p operands. For & (B - 1) require
7852 B to be power of 2. Recognize
7853 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
7855 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7858 * optabs.c (expand_memory_blockage): Fix call of
7859 targetm.have_memory_blockage.
7861 2017-10-14 Jakub Jelinek <jakub@redhat.com>
7864 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
7865 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
7866 objects to extra_objs instead of overwriting it.
7868 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7870 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
7871 Use any_fp_register_operand as operand[3] predicate. Simplify
7872 equality test for operands[2] and operands[4] memory location.
7873 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
7874 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
7875 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
7876 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
7877 any_fp_register_operand as operand[1] predicate. Simplify
7878 equality test for operands[0] and operands[3] memory location.
7879 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
7880 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
7881 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
7883 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
7885 * target-insns.def: Add memory_blockage.
7886 * optabs.c (expand_memory_blockage): New function.
7887 (expand_asm_memory_barrier): Rename ...
7888 (expand_asm_memory_blockage): ... to this.
7889 (expand_mem_thread_fence): Call expand_memory_blockage
7890 instead of expand_asm_memory_barrier.
7891 (expand_mem_singnal_fence): Ditto.
7892 (expand_atomic_load): Ditto.
7893 (expand_atomic_store): Ditto.
7894 * doc/md.texi (Standard Pattern Names For Generation):
7895 Document memory_blockage instruction pattern.
7897 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
7899 * config/rl78/rl78.c (rl78_emit_libcall): New function.
7900 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
7901 * config/rl78/rl78.md: New define_expand "adddi3".
7903 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7905 * cfghooks.c (verify_flow_info): Disable check that all probabilities
7908 2017-10-13 Jeff Law <law@redhat.com>
7910 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
7912 2017-10-13 Jakub Jelinek <jakub@redhat.com>
7915 * internal-fn.c (expand_mul_overflow): If both operands have
7916 the same highpart of -1 or 0 and the topmost bit of lowpart
7917 is different, overflow is if res <= 0 rather than res < 0.
7919 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
7921 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
7922 TARGET_P9_VECTOR code for unaligned_load case.
7924 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
7926 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
7928 2017-10-13 Nathan Sidwell <nathan@acm.org>
7930 * tree-core.h (tree_contains_struct): Make bool.
7931 * tree.c (tree_contains_struct): Likewise.
7932 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
7933 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
7934 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
7935 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
7936 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
7938 2017-10-13 Richard Biener <rguenther@suse.de>
7940 * graphite-isl-ast-to-gimple.c
7941 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
7942 parameters and dominance check.
7943 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
7944 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
7945 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
7946 Do not update SSA form here or do intermediate IL verification.
7947 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
7948 (graphite_initialize): Remove check on the number of loops in
7949 the function and inline into graphite_transform_loops.
7950 (graphite_finalize): Inline into graphite_transform_loops.
7951 (graphite_transform_loops): Perform SSA update and IL verification
7953 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
7955 2017-10-13 Richard Biener <rguenther@suse.de>
7957 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
7958 graphite_expression_type_precision): Avoid global constructor
7960 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
7961 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
7962 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
7963 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
7964 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
7965 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
7967 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
7970 * config/i386/i386.h (ix86_red_zone_size): New.
7971 * config/i386/i386.md (push peephole2s): Replace
7972 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
7974 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
7975 Alan Hayward <alan.hayward@arm.com>
7976 David Sherwood <david.sherwood@arm.com>
7978 * combine.c (can_change_dest_mode): Reject changes in
7979 REGMODE_NATURAL_SIZE.
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 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
7986 (expand_debug_source_expr): Likewise.
7987 * combine.c (combine_simplify_rtx): Likewise.
7988 * cse.c (fold_rtx): Likewise.
7989 * fwprop.c (canonicalize_address): Likewise.
7990 * targhooks.c (default_shift_truncation_mask): 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 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
7997 (widened_mode): Likewise.
7998 (expand_unop): Likewise.
7999 * ree.c (transform_ifelse): Likewise.
8000 (merge_def_and_ext): Likewise.
8001 (combine_reaching_defs): Likewise.
8002 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8004 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8005 Alan Hayward <alan.hayward@arm.com>
8006 David Sherwood <david.sherwood@arm.com>
8008 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
8009 * combine.c (gen_lowpart_for_combine): Likewise.
8010 * dwarf2out.c (rtl_for_decl_location): Likewise.
8011 * final.c (alter_subreg): Likewise.
8012 * rtlhooks.c (gen_lowpart_general): Likewise.
8013 (gen_lowpart_if_possible): Likewise.
8015 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
8016 Alan Hayward <alan.hayward@arm.com>
8017 David Sherwood <david.sherwood@arm.com>
8019 * calls.c (expand_call): Use subreg_lowpart_offset.
8020 * cse.c (cse_insn): Likewise.
8021 * regcprop.c (copy_value): Likewise.
8022 (copyprop_hardreg_forward_1): Likewise.
8024 2017-10-13 Jakub Jelinek <jakub@redhat.com>
8027 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
8028 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
8029 =Q constraints to +Q and into insn condition add check
8030 that operands[0] and operands[1] are equal.
8031 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
8032 =Q constraints to +Q and into insn condition add check
8033 that operands[0] is equal to either operands[1] or operands[2].
8036 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
8037 instead of handling MINUS_EXPR twice (once for each argument),
8038 canonicalize operand order and handle just once, use rtype where
8039 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
8042 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
8043 any values of __C while still being pattern recognizable as a simple
8046 2017-10-13 Richard Biener <rguenther@suse.de>
8048 PR tree-optimization/82451
8050 2017-10-02 Richard Biener <rguenther@suse.de>
8052 PR tree-optimization/82355
8053 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
8054 a mapping for the enclosing loop but avoid generating one for
8056 (copy_bb_and_scalar_dependences): Remove premature codegen
8057 error on PHIs in blocks duplicated into multiple places.
8058 * graphite-scop-detection.c
8059 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
8060 in the region use it as loop and nest to analyze the DR in.
8061 (try_generate_gimple_bb): Likewise.
8062 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
8063 (add_loop_constraints): For blocks in a loop not in the region
8064 create a dimension with a single iteration.
8065 * sese.h (gbb_loop_at_index): Remove assert.
8067 * cfgloop.c (loop_preheader_edge): For the loop tree root
8068 return the single successor of the entry block.
8069 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8070 Reset the SCEV hashtable and niters.
8071 * graphite-scop-detection.c
8072 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
8073 assert that we only have POLYNOMIAL_CHREC that vary in loops
8074 contained in the region.
8075 (scop_detection::graphite_can_represent_expr): Adjust.
8076 (scop_detection::stmt_has_simple_data_refs_p): For loops
8077 not in the region set loop to NULL. The nest is now the
8078 entry edge to the region.
8079 (try_generate_gimple_bb): Likewise.
8080 * sese.c (scalar_evolution_in_region): Adjust for
8081 instantiate_scev change.
8082 * tree-data-ref.h (graphite_find_data_references_in_stmt):
8083 Make nest parameter the edge into the region.
8084 (create_data_ref): Likewise.
8085 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
8086 entry edge into a region and adjust instantiate_scev calls.
8087 (create_data_ref): Likewise.
8088 (graphite_find_data_references_in_stmt): Likewise.
8089 (find_data_references_in_stmt): Pass the loop preheader edge
8090 from the nest argument.
8091 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
8092 parameter the edge into the region.
8093 (instantiate_parameters): Use the loop preheader edge as entry.
8094 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
8096 (get_instantiated_value_entry): Make instantiate_below parameter
8097 the edge into the region.
8098 (instantiate_scev_name): Likewise. Adjust dominance checks,
8099 when we cannot use loop-based instantiation instantiate by
8100 walking use-def chains.
8101 (instantiate_scev_poly): Adjust.
8102 (instantiate_scev_binary): Likewise.
8103 (instantiate_scev_convert): Likewise.
8104 (instantiate_scev_not): Likewise.
8105 (instantiate_array_ref): Remove.
8106 (instantiate_scev_3): Likewise.
8107 (instantiate_scev_2): Likewise.
8108 (instantiate_scev_1): Likewise.
8109 (instantiate_scev_r): Do not blindly handle N-operand trees.
8110 Do not instantiate array-refs. Handle all constants and invariants.
8111 (instantiate_scev): Make instantiate_below parameter
8112 the edge into the region.
8113 (resolve_mixers): Use the loop preheader edge for the region
8114 parameter to instantiate_scev_r.
8115 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
8117 2017-10-13 Richard Biener <rguenther@suse.de>
8119 PR tree-optimization/82525
8120 * graphite-isl-ast-to-gimple.c
8121 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
8123 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
8124 Fail code generation when we cannot represent the isl integer.
8125 (binary_op_to_tree): Elide modulo operations that are no-ops
8126 in the type we code generate. Remove now superfluous code
8129 2017-10-13 Richard Biener <rguenther@suse.de>
8131 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
8132 (scop_detection::harmful_loop_in_region): Remove premature
8133 IV type restriction.
8134 (scop_detection::graphite_can_represent_scev): We can handle
8135 pointer IVs just fine.
8137 2017-10-13 Alan Modra <amodra@gmail.com>
8139 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
8140 "Clobbers and Scratch Registers". Add paragraph on
8141 alternative to clobbers for scratch registers and OpenBLAS
8144 2017-10-13 Alan Modra <amodra@gmail.com>
8146 * doc/extend.texi (Clobbers): Correct vax example. Delete old
8147 example of a memory input for a string of known length. Move
8148 commentary out of table. Add a number of new examples
8149 covering array memory inputs.
8151 2017-10-12 Martin Liska <mliska@suse.cz>
8153 PR tree-optimization/82493
8154 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
8155 (test_range_functions): New function.
8156 (sbitmap_c_tests): Likewise.
8157 * selftest-run-tests.c (selftest::run_tests): Run new tests.
8158 * selftest.h (sbitmap_c_tests): New function.
8160 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
8162 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8164 * config/rs6000/amo.h: Fix spacing issue.
8166 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8169 * config/i386/i386.md (*ashl<mode>3_mask_1,
8170 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
8171 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
8174 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8176 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
8177 (profile_probability): Set max_probability
8178 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
8180 * profile-count.c (profile_probability::differs_from_p): Do not
8181 rely on max_probaiblity == 10000
8183 2017-10-12 Jeff Law <law@redhat.com>
8185 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
8188 2017-10-12 Martin Sebor <msebor@redhat.com>
8192 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
8193 (handle_alias_pairs): Call it.
8194 * common.opt (-Wattribute-alias): New option.
8195 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
8196 * doc/invoke.texi (-Wattribute-alias): Document.
8198 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
8201 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8203 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8205 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8206 (make_hard_regno_born, make_hard_regno_dead): Update
8207 bb_killed_pseudos and bb_gen_pseudos.
8209 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
8211 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
8213 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
8215 * config/alpha/alpha.c (alpha_split_conditional_move):
8216 Use std::swap instead of manually swapping.
8217 (alpha_stdarg_optimize_hook): Ditto.
8218 (alpha_canonicalize_comparison): Ditto.
8220 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8222 * tree-loop-distribution.c (struct builtin_info): New struct.
8223 (struct partition): Refactor fields into struct builtin_info.
8224 (partition_free): Free struct builtin_info.
8225 (build_size_arg_loc, build_addr_arg_loc): Delete.
8226 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
8227 information from struct builtin_info.
8228 (find_single_drs): New function refactored from classify_partition.
8229 Also moved builtin validity checks to this function.
8230 (compute_access_range, alloc_builtin): New functions.
8231 (classify_builtin_st, classify_builtin_ldst): New functions.
8232 (classify_partition): Refactor code into functions find_single_drs,
8233 classify_builtin_st and classify_builtin_ldst.
8234 (distribute_loop): Don't do runtime alias check when distributing
8236 (find_seed_stmts_for_distribution): New function.
8237 (pass_loop_distribution::execute): Refactor code finding seed
8238 stmts into above function. Support distribution for the innermost
8239 two-level loop nest. Adjust dump information.
8241 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8243 * tree-loop-distribution.c: Adjust the general comment.
8244 (NUM_PARTITION_THRESHOLD): New macro.
8245 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
8246 (classify_partition): Skip builtin pattern of loop nest's inner loop.
8247 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
8248 in call to build_partition_graph.
8249 (finalize_partitions): New parameter. Make loop distribution more
8250 conservative by fusing more partitions.
8251 (distribute_loop): Don't do runtime alias check in case of loop nest
8253 (find_seed_stmts_for_distribution): New function.
8254 (prepare_perfect_loop_nest): New function.
8255 (pass_loop_distribution::execute): Refactor code finding seed stmts
8256 and loop nest into above functions. Support loop nest distribution.
8257 Adjust dump information accordingly.
8259 2017-10-12 Bin Cheng <bin.cheng@arm.com>
8261 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
8262 and set PTYPE_SEQUENTIAL for merged partition.
8264 2017-10-12 Richard Biener <rguenther@suse.de>
8266 PR tree-optimization/69728
8268 2017-09-19 Richard Biener <rguenther@suse.de>
8270 PR tree-optimization/69728
8271 * graphite-sese-to-poly.c (schedule_error): New global.
8272 (add_loop_schedule): Handle empty domain by failing the
8274 (build_original_schedule): Handle schedule_error.
8276 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
8277 domain by returning an unchanged schedule.
8279 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8281 * genrecog.c (validate_pattern): For VEC_SELECT verify that
8282 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
8284 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
8286 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
8289 2017-10-12 Jakub Jelinek <jakub@redhat.com>
8292 * expr.c (store_field): Don't optimize away bitsize == 0 store
8293 from CALL_EXPR with addressable return type.
8295 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8297 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
8298 * config/rs6000/rs6000.md (sel): Delete mode attribute.
8299 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
8300 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
8301 TARGET_ISEL instead of TARGET_ISEL<sel>.
8303 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
8305 * config/rs6000/rs6000.c
8306 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
8308 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
8310 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
8312 (reg_or_cint_operand): Fix comment.
8313 (reg_or_zero_operand): New predicate.
8314 * config/rs6000/rs6000-protos.h (output_isel): Delete.
8315 * config/rs6000/rs6000.c (output_isel): Delete.
8316 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
8317 instead of reg_or_cint_operand. Output instruction directly (not via
8319 (isel_unsigned_<mode>): Ditto.
8320 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
8321 gpc_reg_operand. Add an instruction alternative for this. Output
8322 instruction directly.
8323 (*isel_reversed_unsigned_<mode>): Ditto.
8325 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8327 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
8328 (TARGET_CANONICALIZE_COMPARISON): Define.
8330 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
8333 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
8334 Check whether the dest is REG before adding REG_EQUIV note.
8336 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
8339 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
8341 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
8342 (make_hard_regno_born, make_hard_regno_dead): Update
8343 bb_killed_pseudos and bb_gen_pseudos.
8345 2017-10-11 Nathan Sidwell <nathan@acm.org>
8347 * incpath.h (enum incpath_kind): Name enum, prefix values.
8348 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
8349 * incpath.c (heads, tails): Use INC_MAX.
8350 (add_env_var_paths, add_standard_paths): Use incpath_kind.
8351 (merge_include_chains, split_quote_chain,
8352 register_include_chains): Update incpath_kind names.
8353 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
8354 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
8356 (add_framework_path, darwin_register_objc_includes): Likewise.
8357 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
8359 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8361 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
8362 Do not use float_operator operator predicate.
8363 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
8364 * config/i386/predicates.md (float_operator): Remove predicate.
8366 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
8368 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
8369 (*jccxf_i387): Ditto.
8370 (*jcc<mode>_i387): Ditto.
8371 (*jccu<mode>_i387): Ditto.
8372 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
8373 (*jcc_*_i387 splitters): Remove.
8374 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
8375 * config/i386/i386.c (ix86_split_fp_branch): Remove.
8376 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
8379 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8381 * profile-count.h (slow_safe_scale_64bit): New function.
8382 (safe_scale_64bit): New inline.
8383 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
8384 * profile-count.c: Include wide-int.h
8385 (slow_safe_scale_64bit): New.
8387 2017-10-11 Nathan Sidwell <nathan@acm.org>
8389 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
8390 HAS_DECL_ASSEMBLER_NAME_P.
8391 * gimple-expr.c (gimple_decl_printable_name: Check
8392 HAS_DECL_ASSEMBLER_NAME_P too.
8393 * ipa-utils.h (type_in_anonymous_namespace_p): Check
8394 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
8395 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
8396 * passes.c (rest_of_decl_compilation): Check
8397 HAS_DECL_ASSEMBLER_NAME_P too.
8398 * recog.c (verify_changes): Likewise.
8399 * tree-pretty-print.c (dump_decl_name): Likewise.
8400 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
8402 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
8403 (SET_DECL_ASSEMBLER_NAME): Use it.
8404 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
8405 (COPY_DECL_ASSEMBLER_NAME): Likewise.
8406 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
8408 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
8410 * config.gcc (i386, x86_64): Add extra objects.
8411 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
8412 (ix86_min_insn_size): Declare.
8413 (ix86_issue_rate): Declare.
8414 (ix86_adjust_cost): Declare.
8415 (ia32_multipass_dfa_lookahead): Declare.
8416 (ix86_macro_fusion_p): Declare.
8417 (ix86_macro_fusion_pair_p): Declare.
8418 (ix86_bd_has_dispatch): Declare.
8419 (ix86_bd_do_dispatch): Declare.
8420 (ix86_core2i7_init_hooks): Declare.
8421 (ix86_atom_sched_reorder): Declare.
8422 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
8423 (COSTS_N_BYTES): Move to x86-tune-costs.h.
8424 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
8425 (rip_relative_addr_p): Rename to ...
8426 (ix86_rip_relative_addr_p): ... this one; export.
8427 (memory_address_length): Update.
8428 (ix86_issue_rate): Move to x86-tune-sched.c.
8429 (ix86_flags_dependent): Move to x86-tune-sched.c.
8430 (ix86_agi_dependent): Move to x86-tune-sched.c.
8431 (exact_dependency_1): Move to x86-tune-sched.c.
8432 (exact_store_load_dependency): Move to x86-tune-sched.c.
8433 (ix86_adjust_cost): Move to x86-tune-sched.c.
8434 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
8435 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
8436 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
8437 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
8438 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
8439 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
8440 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
8441 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
8442 (min_insn_size): Rename to ...
8443 (ix86_min_insn_size): ... this one; export.
8444 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
8445 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
8446 (core2i7_first_cycle_multipass_backtrack): Move to
8447 x86-tune-sched-core.c.
8448 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
8449 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
8450 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
8451 (ix86_avoid_jump_mispredicts): Update.
8452 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
8453 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
8454 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
8455 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
8456 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
8457 (MAX_INSN): Move to ix86-tune-sched-bd.c.
8458 (MAX_IMM): Move to ix86-tune-sched-bd.c.
8459 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
8460 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
8461 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
8462 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
8463 (MAX_STORE): Move to ix86-tune-sched-bd.c.
8464 (BIG): Move to ix86-tune-sched-bd.c.
8465 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
8466 (enum insn_path): Move to ix86-tune-sched-bd.c.
8467 (get_mem_group): Move to ix86-tune-sched-bd.c.
8468 (is_cmp): Move to ix86-tune-sched-bd.c.
8469 (dispatch_violation): Move to ix86-tune-sched-bd.c.
8470 (is_branch): Move to ix86-tune-sched-bd.c.
8471 (is_prefetch): Move to ix86-tune-sched-bd.c.
8472 (init_window): Move to ix86-tune-sched-bd.c.
8473 (allocate_window): Move to ix86-tune-sched-bd.c.
8474 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
8475 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
8476 (process_end_window): Move to ix86-tune-sched-bd.c.
8477 (allocate_next_window): Move to ix86-tune-sched-bd.c.
8478 (find_constant): Move to ix86-tune-sched-bd.c.
8479 (get_num_immediates): Move to ix86-tune-sched-bd.c.
8480 (has_immediate): Move to ix86-tune-sched-bd.c.
8481 (get_insn_path): Move to ix86-tune-sched-bd.c.
8482 (get_insn_group): Move to ix86-tune-sched-bd.c.
8483 (count_num_restricted): Move to ix86-tune-sched-bd.c.
8484 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
8485 (add_insn_window): Move to ix86-tune-sched-bd.c.
8486 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
8487 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
8488 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
8489 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
8490 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
8491 (do_dispatch): Move to ix86-tune-sched-bd.c.
8492 (has_dispatch): Move to ix86-tune-sched-bd.c.
8493 * config/i386/t-i386: Add new object files.
8494 * config/i386/x86-tune-costs.h: New file.
8495 * config/i386/x86-tune-sched-atom.c: New file.
8496 * config/i386/x86-tune-sched-bd.c: New file.
8497 * config/i386/x86-tune-sched-core.c: New file.
8498 * config/i386/x86-tune-sched.c: New file.
8500 2017-10-11 Liu Hao <lh_mouse@126.com>
8502 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
8503 the generic version below instead.
8504 (should_colorize): Recognize Windows consoles as terminals
8506 * pretty-print.c [__MINGW32__] (write_all): New function.
8507 [__MINGW32__] (find_esc_head): Likewise.
8508 [__MINGW32__] (find_esc_terminator): Likewise.
8509 [__MINGW32__] (eat_esc_sequence): Likewise.
8510 [__MINGW32__] (mingw_ansi_fputs): New function that handles
8512 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
8515 2017-10-11 Richard Biener <rguenther@suse.de>
8517 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
8518 Properly call analyze_scalar_evolution with the loop of the stmt.
8520 2017-10-11 Richard Biener <rguenther@suse.de>
8522 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
8523 * tree-core.h (tree_base): Add chrec_var union member.
8524 * tree.h (CHREC_VAR): Remove.
8525 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
8526 * tree-chrec.h (build_polynomial_chrec): Adjust.
8527 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
8528 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
8530 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
8532 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
8533 * match.pd: ... here.
8534 ((T) X == (T) Y): Relax condition.
8536 2017-10-11 Bin Cheng <bin.cheng@arm.com>
8538 PR tree-optimization/82472
8539 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
8541 (break_alias_scc_partitions): Update postorder number.
8543 2017-10-11 Martin Liska <mliska@suse.cz>
8546 * opts.c (parse_no_sanitize_attribute): Do not use error_value
8548 * opts.h (parse_no_sanitize_attribute): Remove last argument.
8550 2017-10-11 Martin Liska <mliska@suse.cz>
8552 * print-rtl.c (print_insn): Move declaration of idbuf
8553 to same scope as name.
8555 2017-10-11 Martin Liska <mliska@suse.cz>
8560 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8563 2017-10-11 Martin Liska <mliska@suse.cz>
8566 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
8569 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
8571 * config.gcc (default_gnu_indirect_function): Default to yes for
8572 arm*-*-linux* with glibc.
8574 2017-10-11 Richard Biener <rguenther@suse.de>
8576 * tree-scalar-evolution.c (get_scalar_evolution): Handle
8577 default-defs and types we do not want to analyze.
8578 (interpret_loop_phi): Replace unreachable code with an assert.
8579 (compute_scalar_evolution_in_loop): Remove and inline ...
8580 (analyze_scalar_evolution_1): ... here, replacing condition with
8581 what makes the intent clearer. Remove handling of cases
8582 get_scalar_evolution now handles.
8584 2017-10-10 Jim Wilson <wilson@tuliptree.org>
8586 PR rtl-optimization/81434
8587 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
8588 comment for main loop. In sched_group_found if, also add checks for
8589 pass and min_cost_group.
8591 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
8593 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
8594 (rs6000_insn_cost): New function.
8595 * config/rs6000/rs6000.md (cost): New attribute.
8597 2017-10-10 Jakub Jelinek <jakub@redhat.com>
8598 H.J. Lu <hongjiu.lu@intel.com>
8602 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
8603 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
8604 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
8605 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
8606 to OPTION_MASK_ISA_AVX512VL - builtins that have both
8607 OPTION_MASK_ISA_MMX and some other bit set require both
8608 mmx and the ISAs without the mmx bit.
8609 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
8610 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
8611 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
8612 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
8613 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
8614 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
8615 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
8616 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
8617 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
8618 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
8619 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
8620 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
8621 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
8622 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
8623 Add OPTION_MASK_ISA_MMX.
8625 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
8627 * config.gcc (armv7*-*-freebsd*): New target.
8628 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
8630 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
8632 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
8633 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
8636 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
8638 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
8639 a template parameter.
8640 (WIDE_INT_REF_FOR): Update accordingly.
8641 * tree.h (wi::int_traits <const_tree>): Delete.
8642 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
8643 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
8644 (wi::tree_to_wide_ref): New typedef.
8645 (wi::to_wide): New function.
8646 * calls.c (get_size_range): Use wi::to_wide when operating on
8648 * cgraph.c (cgraph_node::create_thunk): Likewise.
8649 * config/i386/i386.c (ix86_data_alignment): Likewise.
8650 (ix86_local_alignment): Likewise.
8651 * dbxout.c (stabstr_O): Likewise.
8652 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
8653 * expr.c (const_vector_from_tree): Likewise.
8654 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
8655 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
8656 (fold_negate_expr_1, int_const_binop_1, const_binop)
8657 (fold_convert_const_int_from_real, optimize_bit_field_compare)
8658 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
8659 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
8660 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
8661 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
8662 (fold_not_const, round_up_loc): Likewise.
8663 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
8664 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
8665 (alloca_call_type): Likewise.
8666 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
8667 * godump.c (go_output_typedef): Likewise.
8668 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
8669 * internal-fn.c (get_min_precision): Likewise.
8670 * ipa-cp.c (ipcp_store_vr_results): Likewise.
8671 * ipa-polymorphic-call.c
8672 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
8673 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
8674 (ipa_modify_call_arguments): Likewise.
8675 * match.pd: Likewise.
8676 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
8677 * print-tree.c (print_node_brief, print_node): Likewise.
8678 * stmt.c (expand_case): Likewise.
8679 * stor-layout.c (layout_type): Likewise.
8680 * tree-affine.c (tree_to_aff_combination): Likewise.
8681 * tree-cfg.c (group_case_labels_stmt): Likewise.
8682 * tree-data-ref.c (dr_analyze_indices): Likewise.
8683 (prune_runtime_alias_test_list): Likewise.
8684 * tree-dump.c (dequeue_and_dump): Likewise.
8685 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
8686 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
8687 * tree-pretty-print.c (dump_generic_node): Likewise.
8688 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
8689 (simple_iv_with_niters): Likewise.
8690 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
8691 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
8692 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
8693 * tree-ssa-loop-niter.c (split_to_var_and_offset)
8694 (refine_value_range_using_guard, number_of_iterations_ne_max)
8695 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
8696 (get_cst_init_from_scev, record_nonwrapping_iv)
8697 (scev_var_range_cant_overflow): Likewise.
8698 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
8699 * tree-ssa-pre.c (compute_avail): Likewise.
8700 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
8701 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
8702 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
8703 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
8704 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
8705 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
8706 (dump_case_nodes, try_switch_expansion): Likewise.
8707 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
8708 (vect_do_peeling): Likewise.
8709 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
8710 * tree-vect-stmts.c (vectorizable_load): Likewise.
8711 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
8712 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
8713 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
8714 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
8715 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
8716 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
8717 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
8718 (evrp_dom_walker::before_dom_children): Likewise.
8719 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
8720 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
8721 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
8722 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
8723 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
8724 (get_range_pos_neg): Likewise.
8725 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
8726 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
8727 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
8728 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
8729 * config/avr/avr.c (avr_fold_builtin): Likewise.
8730 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
8731 * config/msp430/msp430.c (msp430_attr): Likewise.
8732 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
8733 * config/powerpcspe/powerpcspe-c.c
8734 (altivec_resolve_overloaded_builtin): Likewise.
8735 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
8736 (rs6000_expand_ternop_builtin): Likewise.
8737 * config/rs6000/rs6000-c.c
8738 (altivec_resolve_overloaded_builtin): Likewise.
8739 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
8740 (rs6000_expand_ternop_builtin): Likewise.
8741 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
8743 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8745 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
8746 when copying loop nest with only one inner loop.
8748 2017-10-10 Richard Biener <rguenther@suse.de>
8750 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
8751 blocks if SCEV is active.
8752 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
8754 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
8755 (scev_initialize): Assert we are not yet initialized.
8757 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8759 * tree-loop-distribution.c (generate_loops_for_partition): Remove
8760 inner loop's exit stmt by making it always exit the loop, otherwise
8761 we would generate an infinite empty loop.
8763 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8765 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
8766 renaming variables in new preheader if it's deleted.
8768 2017-10-10 Bin Cheng <bin.cheng@arm.com>
8770 * tree-loop-distribution.c (struct partition): Remove unused field
8771 loops of the structure.
8772 (partition_alloc, partition_free): Ditto.
8773 (build_rdg_partition_for_vertex): Ditto.
8775 2017-10-09 Jeff Law <law@redhat.com>
8777 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
8778 return type to match prototype and documentation.
8780 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8782 * config/rs6000/rs6000.c (processor_costs): Move to ...
8783 * config/rs6000/rs6000.h: ... here.
8784 (rs6000_cost): Declare.
8786 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
8788 * except.c (setjmp_fn): New global variable.
8789 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
8790 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
8791 if DONT_USE_BUILTIN_SETJMP is defined.
8793 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8795 * target.def (insn_cost): New hook.
8796 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
8797 * doc/tm.texi: Regenerate.
8798 * rtlanal.c (insn_cost): Use the new hook.
8800 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8802 * combine.c (combine_validate_cost): Compute the new insn_cost,
8803 not just pattern_cost.
8804 (try_combine): Adjust comment.
8806 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
8808 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
8810 * combine.c (uid_insn_cost): Adjust comment.
8811 (combine_validate_cost): Adjust comment. Use pattern_cost instead
8813 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
8814 * dse.c (find_shift_sequence): Ditto.
8815 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
8816 (bb_valid_for_noce_process_p): Use pattern_cost.
8817 * rtl.h (insn_rtx_cost): Delete.
8818 (pattern_cost): New prototype.
8819 (insn_cost): New prototype.
8820 * rtlanal.c (insn_rtx_cost): Rename to...
8821 (pattern_cost): ... this.
8824 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
8826 * config/i386/i386.md (*jcc_2): Remove insn pattern.
8827 (*jcc<mode>_0_r_i387): Ditto.
8828 (*jccxf_r_i387): Ditto.
8829 (*jcc<mode>_r_i387): Ditto.
8830 (*jccu<mode>_r_i387): Ditto.
8831 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
8832 (*jcc): Rename from *jcc_1.
8834 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8836 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
8837 deferred rescans after the lvx/stvx recombination pre-pass.
8839 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
8841 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
8842 memory operation instruction support.
8843 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
8844 (rs6000-ibm-aix[789]*): Likewise.
8845 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
8846 Document new functions.
8848 2017-10-09 Richard Biener <rguenther@suse.de>
8850 PR tree-optimization/82397
8851 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
8852 equality only for semantically equal trees.
8854 2017-10-09 Richard Biener <rguenther@suse.de>
8856 PR tree-optimization/82449
8857 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
8858 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
8859 allow constant addresses.
8860 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
8863 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8865 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
8868 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8871 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
8874 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8877 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
8879 2017-10-09 Jakub Jelinek <jakub@redhat.com>
8882 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
8883 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
8885 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
8887 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
8888 (WI_BINARY_PREDICATE_RESULT): Likewise.
8889 (wi::binary_traits::operator_result): New type.
8890 (wi::binary_traits::predicate_result): Likewise.
8891 (generic_wide_int::operator~, unary generic_wide_int::operator-)
8892 (generic_wide_int::operator==, generic_wide_int::operator!=)
8893 (generic_wide_int::operator&, generic_wide_int::and_not)
8894 (generic_wide_int::operator|, generic_wide_int::or_not)
8895 (generic_wide_int::operator^, generic_wide_int::operator+
8896 (binary generic_wide_int::operator-, generic_wide_int::operator*):
8898 (operator~, unary operator-, operator==, operator!=, operator&)
8899 (operator|, operator^, operator+, binary operator-, operator*): New
8901 * expr.c (get_inner_reference): Use wi::bit_and_not.
8902 * fold-const.c (fold_binary_loc): Likewise.
8903 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
8904 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
8905 (bit_value_binop): Likewise.
8906 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
8907 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
8908 (extract_range_from_binary_expr_1): Likewise.
8909 (masked_increment): Likewise.
8910 (simplify_bit_ops_using_ranges): Likewise.
8912 2017-10-09 Martin Jambor <mjambor@suse.cz>
8915 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
8916 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
8917 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
8919 (hsa_fixup_mov_insn_type): New function.
8920 (hsa_op_with_type::get_in_type): Use it.
8921 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
8922 immediates in an assert.
8923 (hsa_op_with_type::extend_int_to_32bit): New method.
8924 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
8925 types. Convert to dest type if necessary.
8926 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
8927 (reg_for_gimple_ssa): Pass false as min32int to
8928 hsa_type_for_scalar_tree_type.
8929 (gen_hsa_addr): Fixup type when creating addresable temporary.
8930 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
8931 (gen_hsa_unary_operation): Extend operands and convert to dest type if
8932 necessary. Call hsa_fixup_mov_insn_type.
8933 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
8934 extend operands and convert to dest type if necessary.
8935 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
8936 to dest type if necessary.
8937 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
8938 if conversion nt necessary and size matches.
8939 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
8940 to dest type if necessary.
8941 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
8942 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
8944 (gen_hsa_clrsb): Likewise.
8945 (gen_hsa_ffs): Likewise.
8946 (gen_hsa_divmod): Extend operands and convert to dest type if
8948 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
8950 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8952 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
8953 Remove empty default arguments. Use a brace block as output
8955 (conditional return): Ditto.
8957 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
8958 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
8960 (group_ending_nop): Ditto.
8961 (doloop_end): Ditto.
8962 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
8963 (splitters for those): Ditto.
8965 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
8967 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
8968 a conditional jump (and the compare for it) so that pc_rtx is the
8970 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
8971 for the deleted and renamed ctr<mode>_internal[234] patterns.
8972 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
8973 Delete second conditional return pattern.
8974 (ctr<mode>_internal2): Delete this second bdnz pattern.
8975 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
8976 (ctr<mode>_internal4): Delete this second bdz pattern.
8978 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
8980 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
8981 (always_initialized_rtx_for_ssa_name_p): New predicate.
8982 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
8983 (finish_out_of_ssa): Free new field of SA.
8984 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
8985 * tree-ssa-coalesce.c: Include tree-ssa.h.
8986 (get_parm_default_def_partitions): Remove extern keyword.
8987 (get_undefined_value_partitions): New function.
8988 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
8989 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
8991 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
8993 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
8995 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
8997 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
8999 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
9000 for targets that preffer 128bit.
9002 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
9004 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
9006 2017-10-08 Olivier Hainque <hainque@adacore.com>
9008 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
9009 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
9011 (thumb_set_return_address): Likewise.
9013 2017-10-08 Olivier Hainque <hainque@adacore.com>
9015 * common/config/arm/arm-common.c (arm_except_unwind_info):
9016 Handle DWARF2_UNWIND_INFO.
9018 2017-10-07 Michael Collison <michael.collison@arm.com>
9020 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
9023 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
9025 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
9026 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
9027 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
9028 defined, force the creation of a new block for a dispatch label.
9030 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
9032 * invoke.texi (Wsuggest-attribute=cold): Document.
9033 * common.opt (Wsuggest-attribute=cold): New
9034 * ipa-pure-const.c (warn_function_cold): New function.
9035 * predict.c (compute_function_frequency): Use it.
9036 * predict.h (warn_function_cold): Declare.
9038 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
9040 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
9043 2017-10-06 Martin Liska <mliska@suse.cz>
9045 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
9046 keyword for member functions.
9047 (struct sanopt_tree_couple): New struct.
9048 (struct sanopt_tree_couple_hash): New function.
9049 (struct sanopt_ctx): Add new hash_map.
9050 (has_dominating_ubsan_ptr_check): New function.
9051 (record_ubsan_ptr_check_stmt): Likewise.
9052 (maybe_optimize_ubsan_ptr_ifn): Likewise.
9053 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
9054 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
9056 2017-10-06 Sudakshina Das <sudi.das@arm.com>
9059 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
9060 aarch64_simd_valid_immediate on CONST_VECTORs.
9061 (aarch64_reg_or_bic_imm): Likewise.
9063 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
9065 PR rtl-optimization/82396
9066 * haifa-sched.c (ready_sort_real): Disable qsort checking.
9068 2017-10-06 Sebastian Pop <sebpop@gmail.com>
9070 * graphite-dependences.c (scop_get_reads): Move code to...
9071 (scop_get_must_writes): Move code to...
9072 (scop_get_may_writes): Move code to...
9073 (scop_get_reads_and_writes): ... here.
9074 (scop_get_dependences): Call scop_get_reads_and_writes.
9076 2017-10-06 Jakub Jelinek <jakub@redhat.com>
9078 PR tree-optimization/82434
9079 * fold-const.h (can_native_encode_type_p,
9080 can_native_encode_string_p): Remove.
9081 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
9082 don't encode anything, just return what would be otherwise returned.
9083 (native_encode_fixed, native_encode_complex, native_encode_vector):
9085 (native_encode_string): Likewise. Inline by hand
9086 can_native_encode_string_p.
9087 (can_native_encode_type_p): Remove.
9088 (can_native_encode_string_p): Remove.
9089 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
9090 STRING_CSTs using can_native_encode_string_p, test all
9091 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
9092 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
9093 argument from native_encode_expr.
9094 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
9095 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
9098 2017-10-06 Richard Biener <rguenther@suse.de>
9100 PR tree-optimization/82397
9101 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
9102 operand_equal_p but rely on data_ref_compare_tree for detecting
9104 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
9105 to match up with dr_group_sort_cmp.
9107 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9110 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
9112 * config/s390/s390-builtin-types.def: Regenerate.
9114 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9117 * config/s390/s390-builtin-types.def: Regenerate.
9118 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
9119 Change flag from B_VXE to B_VX.
9120 (s390_vec_min_dbl): Remove B_VXE flag.
9122 2017-10-06 Richard Biener <rguenther@suse.de>
9124 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
9125 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
9126 translate_isl_ast_to_gimple::is_valid_rename,
9127 translate_isl_ast_to_gimple::get_rename,
9128 translate_isl_ast_to_gimple::get_def_bb_for_const,
9129 translate_isl_ast_to_gimple::get_new_name,
9130 translate_isl_ast_to_gimple::collect_all_ssa_names,
9131 translate_isl_ast_to_gimple::copy_loop_phi_args,
9132 translate_isl_ast_to_gimple::collect_all_ssa_names,
9133 translate_isl_ast_to_gimple::copy_loop_phi_args,
9134 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
9135 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
9136 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
9137 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
9138 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
9139 translate_isl_ast_to_gimple::copy_cond_phi_args,
9140 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
9141 translate_isl_ast_to_gimple::edge_for_new_close_phis,
9142 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
9143 translate_isl_ast_to_gimple::rename_uses,
9144 translate_isl_ast_to_gimple::rename_all_uses): Remove.
9145 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
9146 (set_rename_for_each_def): Likewise.
9147 (graphite_copy_stmts_from_block): Handle debug stmt resetting
9148 here. Handle rewriting SCEV analyzable uses here.
9149 (copy_bb_and_scalar_dependences): Generate code for PHI
9151 (graphite_regenerate_ast_isl): Adjust.
9152 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
9153 (add_write, add_read): New functions.
9154 (build_cross_bb_scalars_def): Use it and simplify.
9155 (build_cross_bb_scalars_use): Likewise.
9156 (graphite_find_cross_bb_scalar_vars): Inline into...
9157 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
9158 simulating out-of-SSA. Compute liveout and add dependencies.
9159 (build_scops): Force an empty entry block.
9160 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
9162 (sese_build_liveouts): Declare.
9163 (sese_trivially_empty_bb_p): Likewise.
9164 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
9165 compute liveout and debug_liveout.
9166 (sese_bad_liveouts_use): Remove.
9167 (sese_reset_debug_liveouts_bb): Likewise.
9168 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
9169 (sese_build_liveouts): Build liveout and debug_liveout and store
9171 (new_sese_info): Adjust.
9172 (free_sese_info): Likewise.
9173 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
9174 do not build liveout here.
9175 (move_sese_in_condition): Adjust region entry.
9176 (scev_analyzable_p): Match up with chrec_apply requirements.
9177 (sese_trivially_empty_bb_p): New.
9178 * tree-into-ssa.c (get_reaching_def): Properly support generating
9179 default-defs for incremental rewrite of anonymous names.
9181 2017-10-06 Richard Biener <rguenther@suse.de>
9183 * graphite-sese-to-poly.c (extract_affine): For casts increasing
9184 precision do not perform modulo reduction.
9186 2017-10-06 Richard Biener <rguenther@suse.de>
9188 PR tree-optimization/82436
9189 * tree-vect-slp.c (vect_supported_load_permutation_p): More
9190 conservatively choose the vectorization factor when checking
9191 whether we can perform the required load permutation.
9192 (vect_transform_slp_perm_load): Assert when we may not fail.
9194 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
9196 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
9197 message for incompatible -msdata=* and -mcall-* options.
9199 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9201 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
9202 rate for post-reload scheduling.
9204 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9206 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
9208 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9210 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
9211 to improve monte carlo in scimark.
9213 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
9215 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
9216 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
9217 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
9218 pentium4_cost, nocona_cost): Set reassociation width to 1.
9219 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
9220 width to 2 for fp operations and 1 otherwise.
9221 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
9223 (atom_cost): Set reassociation width to 2.
9224 (slm_cost, generic_cost): Set fp reassociation width
9225 to 2 and 1 otherwise.
9226 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
9227 (core_cost): Set fp reassociation width to 4 and vector to 2.
9228 (ix86_reassociation_width): Rewrite using cost table; special case
9229 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
9230 and TARGET_AVX128_OPTIMAL.
9231 * config/i386/i386.h (processor_costs): Add
9232 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
9233 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
9234 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
9235 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
9236 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
9237 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
9239 2017-10-05 Nathan Sidwell <nathan@acm.org>
9241 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
9243 2017-10-05 Tamar Christina <tamar.christina@arm.com>
9245 * config/arm/arm.c (arm_test_fpu_data): New.
9246 (arm_run_selftests): Call arm_test_fpu_data.
9248 2017-10-04 Nathan Sidwell <nathan@acm.org>
9250 * toplev.c (toplev::main): Remove excess parens on pretty_printer
9252 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
9254 2017-10-04 Sudakshina Das <sudi.das@arm.com>
9256 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
9257 check type for aarch64_simd_valid_immediate.
9258 (aarch64_output_simd_mov_immediate): Update prototype.
9259 (aarch64_simd_valid_immediate): Update prototype.
9260 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
9261 support for ORR-immediate.
9262 (and<mode>3): modified pattern to add support for BIC-immediate.
9263 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
9264 now checks for valid immediate for BIC and ORR based on new enum
9266 (aarch64_output_simd_mov_immediate): Function now used to output
9267 BIC/ORR imm as well based on new enum argument.
9268 * config/aarch64/constraints.md (Do): New vector immediate constraint.
9270 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
9271 (aarch64_reg_or_bic_imm): Likewise.
9273 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9275 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
9276 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
9278 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9282 PR rtl-optimization/82396
9283 * haifa-sched.c (autopref_multipass_init): Simplify
9285 (autopref_rank_data): Simplify sort order.
9286 * sched-int.h (autopref_multipass_data_): Remove
9287 multi_mem_insn_p, min_offset and max_offset.
9289 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9291 * doc/sourcebuild.texi: Document vect_peeling_profitable.
9293 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9295 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
9298 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9300 * doc/sourcebuild.texi: Document vect_long_mult.
9302 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
9304 PR tree-optimization/82413
9305 * fold-const.c (build_range_check): Use widest_int when comparing
9306 the maximum ETYPE value with HIGH.
9308 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
9310 PR rtl-optimization/82396
9311 * haifa-sched.c (autopref_multipass_init): Simplify
9313 (autopref_rank_data): Simplify sort order.
9314 * sched-int.h (autopref_multipass_data_): Remove
9315 multi_mem_insn_p, min_offset and max_offset.
9317 2017-10-04 Jakub Jelinek <jakub@redhat.com>
9319 PR tree-optimization/82381
9320 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
9321 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
9324 PR tree-optimization/82374
9325 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
9326 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
9327 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
9328 current_function_decl to the new decl.
9330 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
9332 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
9333 helper macro for IEEE float128 hardware built-in functions.
9334 (SQRTF128_ODD): Add built-in functions with the round-to-odd
9336 (TRUNCF128_ODD): Likewise.
9337 (ADDF128_ODD): Likewise.
9338 (SUBF128_ODD): Likewise.
9339 (MULF128_ODD): Likewise.
9340 (DIVF128_ODD): Likewise.
9341 (FMAF128_ODD): Likewise.
9342 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
9343 UNSPEC_TRUNC_ROUND_TO_ODD.
9344 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
9345 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
9346 floating point round to odd instructions.
9347 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
9348 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
9349 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
9350 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
9351 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
9352 (trunc<mode>sf2_hw): Change the truncate with round to odd
9353 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
9354 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
9355 to odd hardware instructions.
9356 (sub<mode>3_odd): Likewise.
9357 (mul<mode>3_odd): Likewise.
9358 (div<mode>3_odd): Likewise.
9359 (sqrt<mode>2_odd): Likewise.
9360 (fma<mode>4_odd): Likewise.
9361 (fms<mode>4_odd): Likewise.
9362 (nfma<mode>4_odd): Likewise.
9363 (nfms<mode>4_odd): Likewise.
9364 (trunc<mode>df2_odd): Change the truncate with round to odd
9365 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
9367 * doc/extend.texi (PowerPC built-in functions): Update documentation
9368 for existing IEEE float128-bit built-in functions. Add built-in
9369 functions that generate the IEEE 128-bit floating point round to
9372 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
9374 PR rtl-optimization/77729
9375 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
9376 to (X&(C1&~C2))|C2 transformations.
9378 2017-10-03 Martin Jambor <mjambor@suse.cz>
9380 PR tree-optimization/82363
9381 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
9382 mismatch, mark lacc written regardless of racc.
9384 2017-10-03 Jakub Jelinek <jakub@redhat.com>
9386 PR tree-optimization/82381
9387 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
9388 stmt_to_insert nor wheather SSA_NAMEs are default defs.
9389 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
9390 fallthrough into reassoc_stmt_dominates_stmt_p.
9393 * combine.c (combine_instructions): Don't combine in unreachable
9396 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
9399 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
9400 function to not use the have_cpu variable. Do not set cpu_index,
9401 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
9403 (rs6000_valid_attribute_p): Remove duplicate initializations of
9404 old_optimize and func_optimize.
9405 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
9406 (rs6000_activate_target_options): Make global.
9407 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
9410 2017-10-02 Jakub Jelinek <jakub@redhat.com>
9412 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
9413 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
9414 Set *poffset to 0 and *psize and *pmax_size to -1 if
9415 *poffset + *psize overflows in HOST_WIDE_INT.
9417 PR tree-optimization/82387
9418 PR tree-optimization/82388
9419 PR tree-optimization/82389
9420 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
9421 instead of live_bytes non-NULL.
9423 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
9426 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
9429 2017-10-02 Richard Biener <rguenther@suse.de>
9431 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
9432 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
9433 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
9435 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9437 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
9438 requested precision matches the type's.
9439 * calls.c (alloc_max_size): Calculate the new candidate size as
9440 a widest_int and use wi::to_widest when comparing it with the
9441 current candidate size.
9442 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
9443 zero rather than integer_zero_node.
9444 * match.pd: Check for a no-op conversion before using wi::add
9445 rather than after. Use tree_to_uhwi when summing small shift
9446 counts into an unsigned int.
9448 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
9449 Alan Hayward <alan.hayward@arm.com>
9450 David Sherwood <david.sherwood@arm.com>
9453 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
9454 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
9455 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
9456 POINTER_AND_FP_REGS.
9458 2017-10-02 Richard Biener <rguenther@suse.de>
9460 PR tree-optimization/82355
9461 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9462 a mapping for the enclosing loop but avoid generating one for
9464 (copy_bb_and_scalar_dependences): Remove premature codegen
9465 error on PHIs in blocks duplicated into multiple places.
9466 * graphite-scop-detection.c
9467 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9468 in the region use it as loop and nest to analyze the DR in.
9469 (try_generate_gimple_bb): Likewise.
9470 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9471 (add_loop_constraints): For blocks in a loop not in the region
9472 create a dimension with a single iteration.
9473 * sese.h (gbb_loop_at_index): Remove assert.
9475 2017-10-01 Kevin Buettner <kevinb@redhat.com>
9477 * omp-expand.c (adjust_context_scope): New function.
9478 (expand_parallel_call): Call adjust_context_scope.
9480 2017-10-01 Jeff Law <law@redhat.com>
9482 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
9483 dom_opt_dom_walker class with direct access to private members.
9484 Add comments. Call test_for_singularity.
9485 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
9486 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
9487 m_dummy_cond anymore.
9488 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
9490 (pass_dominator:execute): Build the dummy_cond here and pass it
9491 to the dom_opt_dom_walker ctor.
9492 (test_for_singularity): New function.
9494 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
9495 Maya Rashish <coypu@sdf.org>
9497 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
9498 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
9499 (alpha*-*-netbsd*) Use nbsd_tm_file.
9500 (arm*-*-netbsdelf*) Likewise.
9501 (i[34567]86-*-netbsdelf*) Likewise.
9502 (x86_64-*-netbsd*) Likewise.
9503 (mips*-*-netbsd*) Likewise.
9504 (powerpc-*-netbsd*) Likewise.
9505 (sh*-*-netbsd*) Likewise.
9506 (sparc-*-netbsdelf*) Likewise.
9507 (sparc64-*-netbsd*) Likewise.
9508 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
9510 (vax-*-netbsdelf*) Likewise.
9511 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
9512 (UINT_FAST8_TYPE) Likewise.
9513 (INT_FAST16_TYPE) Check CHAR_FAST16.
9514 (UINT_FAST16_TYPE) Likewise.
9516 2017-09-30 Jakub Jelinek <jakub@redhat.com>
9519 * config/i386/i386.md
9520 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
9521 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
9522 *divmodsi4_zext_2): New define_insn_and_split.
9523 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
9524 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
9525 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
9526 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
9527 New define_insn_and_split.
9528 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
9529 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
9530 operands[1] having DImode when mode is SImode.
9532 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
9533 always SImode for DIV and MOD in REG_EQUAL notes.
9535 2017-09-29 Yury Gribov <tetra2005@gmail.com>
9538 * match.pd: Fix handling of NaNs in pattern.
9540 2017-09-29 Jeff Law <law@redhat.com>
9542 * sbitmap.c (bitmap_bit_in_range_p): New function.
9543 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
9544 * tree-ssa-dse.c (live_bytes_read): New function.
9545 (dse_classify_store): Ignore reads of dead bytes.
9547 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
9548 typos and whitespace errors.
9549 * config/i386/predicates.md (address_no_seg_operand): Likewise.
9550 * config/s390/s390.c (s390_emit_prologue): Likewise.
9552 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9555 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
9556 with a symbol for LRA.
9558 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
9560 PR rtl-optimization/82338
9561 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
9563 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
9565 * genmodes.c (calc_wider_mode): Suppress qsort macro.
9566 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
9567 (qsort_chk): Declare.
9568 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
9569 (qsort_chk): New function.
9571 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9573 PR tree-optimization/82337
9574 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
9575 phi definition if the PHI result appears in an abnormal PHI.
9576 (find_basis_for_base_expr): Don't record a basis if the LHS of the
9577 basis appears in an abnormal PHI.
9579 2017-09-29 Richard Biener <rguenther@suse.de>
9581 * graphite-isl-ast-to-gimple.c
9582 (translate_isl_ast_to_gimple::set_codegen_error): New function.
9583 (binary_op_to_tree): Use it.
9584 (get_rename_from_scev): Likewise.
9585 (copy_loop_phi_nodes): Likewise.
9586 (copy_bb_and_scalar_dependences): Likewise.
9587 (translate_pending_phi_nodes): Likewise.
9589 2017-09-29 Jakub Jelinek <jakub@redhat.com>
9592 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
9593 for movabsq $(i32 << shift), r64.
9595 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
9597 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
9598 index when encoding %esp as %rsp to avoid 0x67 prefix.
9600 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
9602 * config/i386/i386.md (*movsf_internal, *movdf_internal):
9603 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
9605 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
9607 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
9608 Extensions with more than 16 double VFP registers.
9609 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
9610 to_clear_mask and all code related to it. Replace the remaining
9611 entry by a sbitmap and adapt code accordingly.
9613 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
9615 * brig-builtins.def: Change pure attributes to const.
9617 2017-09-28 Joseph Myers <joseph@codesourcery.com>
9619 * config.gcc (default_gnu_indirect_function): Default to yes for
9620 sparc*-*-linux* with glibc.
9622 2017-09-28 Joseph Myers <joseph@codesourcery.com>
9624 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
9625 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
9626 when creating .init_array and .fini_array sections with priority
9629 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
9632 * config/aarch64/aarch64.c
9633 (aarch64_builtin_support_vector_misalignment): Always return false
9634 when misalignment is unknown.
9636 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
9638 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
9639 this function to return false if the definition used by the swap
9640 instruction is artificial, or if the memory address from which the
9641 constant value is loaded is not represented by a base address held
9642 in a register or if the base address register is a frame or stack
9643 pointer. Additionally, return false if the base address of the
9644 loaded constant is a SYMBOL_REF but is not considered to be a
9646 (replace_swapped_load_constant): New function.
9647 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
9648 loaded constant vector with a load of a swapped constant vector.
9650 2017-09-27 Carl Love <cel@us.ibm.com>
9652 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
9653 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
9654 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
9657 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
9659 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
9660 first, always call autopref_rank_data otherwise.
9662 2017-09-27 Richard Biener <rguenther@suse.de>
9664 * graphite-scop-detection.c (find_scop_parameters): Move
9665 loop bound handling ...
9666 (gather_bbs::before_dom_children): ... here, avoiding the need
9667 to build scop_info->loop_nest.
9668 (record_loop_in_sese): Remove.
9669 * sese.h (sese_info_t::loop_nest): Remove.
9670 * sese.c (new_sese_info): Do not allocate loop_nest.
9671 (free_sese_info): Do not free loop_nest.
9673 2017-09-27 Jakub Jelinek <jakub@redhat.com>
9676 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
9677 lhs from calls if the lhs has addressable type.
9679 2017-09-27 Richard Biener <rguenther@suse.de>
9681 * graphite.h (scop::max_alias_set): New member.
9682 * graphite-scop-detection.c: Remove references to non-existing
9683 --param in comments.
9684 (build_alias_sets): Record the maximum alias set used for drs.
9685 (build_scops): Support zero as unlimited for
9686 --param graphite-max-arrays-per-scop.
9687 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
9689 (build_poly_sr_1): ... here. Compute alias set based on the
9690 maximum alias set used for drs rather than
9691 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
9693 2017-09-27 Richard Biener <rguenther@suse.de>
9695 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
9696 --param loop-block-tile-size=0 to disable tiling.
9698 2017-09-27 Richard Biener <rguenther@suse.de>
9700 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
9701 (graphite-max-nb-scop-params): Document special value zero.
9702 * domwalk.h (dom_walker::STOP): New symbolical constant.
9703 (dom_walker::dom_walker): Add optional parameter for bb to
9705 (dom_walker::~dom_walker): Declare.
9706 (dom_walker::before_dom_children): Document STOP return value.
9707 (dom_walker::m_user_bb_to_rpo): New member.
9708 (dom_walker::m_bb_to_rpo): Likewise.
9709 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
9710 mapping here if not provided by the user.
9711 (dom_walker::~dom_walker): Free bb to RPO mapping if not
9712 provided by the user.
9713 (dom_walker::STOP): Define.
9714 (dom_walker::walk): Do not compute bb to RPO mapping here.
9715 Support STOP return value from before_dom_children to stop
9717 * graphite-optimize-isl.c (optimize_isl): If the schedule
9718 is the same still generate code if -fgraphite-identity
9719 or -floop-parallelize-all are given.
9720 * graphite-scop-detection.c: Include cfganal.h.
9721 (gather_bbs::gather_bbs): Get and pass through bb to RPO
9723 (gather_bbs::before_dom_children): Return STOP for BBs
9725 (build_scops): Compute bb to RPO mapping and pass it to
9726 the domwalk. Treat --param graphite-max-nb-scop-params=0
9727 as not limiting the number of params.
9728 * graphite.c (graphite_initialize): Remove limit on the
9729 number of basic-blocks in a function.
9730 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
9731 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
9732 default value of 10.
9734 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9736 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
9737 Adjust code to eliminate needing to do the shift right 32-bits
9738 operation after XSCVDPSPN.
9740 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9742 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
9743 ((X / Y) != 0 -> X >= Y): Likewise.
9745 2017-09-26 Carl Love <cel@us.ibm.com>
9747 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
9748 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
9749 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
9750 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
9751 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
9752 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
9753 definitions and overloading.
9754 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
9755 statement for P9V_BUILTIN_XST_LEN_R.
9756 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
9757 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
9758 define_expand and define_insn for the instructions and builtins.
9759 * doc/extend.texi: Update the built-in documentation file for the new
9761 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
9762 define_insn for the instructions
9764 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
9767 * gcc/config/netbsd-protos.h: New file.
9768 * gcc/config/netbsd.c: New file.
9769 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
9770 * gcc/config/t-netbsd: New file.
9771 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
9772 (tmake_file) Add t-netbsd.
9773 (extra_objs) Add netbsd.o.
9775 2017-09-26 Janus Weil <janus@gcc.gnu.org>
9779 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
9781 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
9783 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
9784 sign extension from a vector register to a GPR by doing a 32-bit
9785 direct move and then an EXTSW.
9786 (extendsi<mode>2 splitter): Likewise.
9787 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
9788 right or vector extract after doing XSCVDPSPN. Use
9789 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
9791 (movdi_from_sf_zero_ext): Likewise.
9792 (reload_gpr_from_vsxsf): Likewise.
9793 (p8_mfvsrd_4_disf): Delete, no longer used.
9794 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
9795 then needing to move the SFmode to a GPR to use the XSCVDPSP
9796 instruction instead of FRSP and XSCVDPSPN.
9797 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
9798 it is adjacent to the other XSCVSPDP insns.
9799 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
9800 SFmode to be in traditional Altivec registers.
9801 (vsx_xscvdpspn): Eliminate useless alternative constraint.
9802 (vsx_xscvspdpn): Likewise.
9803 (vsx_xscvspdpn_scalar): Likewise.
9805 2017-09-26 Martin Jambor <mjambor@suse.cz>
9807 * tree-sra.c (compare_access_positions): Put integral types first,
9808 stabilize sorting of integral types, remove conditions putting
9809 non-full-precision integers last.
9810 (sort_and_splice_var_accesses): Disable scalarization if a
9811 non-integert would be represented by a non-full-precision integer.
9813 2017-09-26 Joseph Myers <joseph@codesourcery.com>
9815 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
9816 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
9817 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
9818 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
9819 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
9820 conditionals inside the function instead of around it. Call
9821 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
9822 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
9824 2017-09-26 Richard Biener <rguenther@suse.de>
9826 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
9828 (scop_detection::build_scop_breadth): ... this. Removed.
9829 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
9830 (scop_detection::harmful_stmt_in_bb): Likewise.
9831 (scop_detection::graphite_can_represent_stmt): Likewise.
9832 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
9833 (scop_detection::can_represent_loop): Remove recursion, fold in ...
9834 (scop_detection::can_represent_loop_1): ... this. Removed.
9835 (scop_detection::harmful_loop_in_region): Simplify after inlining
9836 the above and remove more quadraticness.
9837 (build_scops): Adjust.
9838 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
9841 2017-09-26 Jakub Jelinek <jakub@redhat.com>
9844 * config/i386/i386.c (ix86_print_operand_address_as): Only test
9845 REGNO (base) == SP_REG if base is a REG.
9848 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
9849 if it is different SSA_NAME.
9850 (optimize_range_tests_cmp_bitwise): New function.
9851 (optimize_range_tests): Call it.
9853 2017-09-26 Richard Biener <rguenther@suse.de>
9855 PR tree-optimization/82321
9856 * graphite.c (canonicalize_loop_closed_ssa): Properly check
9857 for the def being inside the loop.
9859 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9861 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
9863 * config/s390/s390-builtins.def: Fix constraint on op4.
9865 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9867 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
9868 independent expanders.
9869 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
9870 ("vec_ordered", "vec_unordered"): New expanders.
9872 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9874 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
9877 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9879 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
9880 vec_unpacks_lo_v16qi.
9881 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
9883 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9885 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
9886 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
9887 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
9889 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9891 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
9893 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
9895 ("*vec_slb<mode>"): New insn pattern.
9896 ("vec_shr_<mode>"): New expander.
9897 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
9898 and force the shift count operand to V16QImode.
9899 ("vec_srb<mode>"): Set shift count mode to V16QI.
9901 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9903 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
9904 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
9905 ("vec_widen_smult_hi_<mode>"): New expander definitions.
9907 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
9910 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
9912 2017-09-26 Richard Biener <rguenther@suse.de>
9914 PR tree-optimization/82320
9915 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
9918 2017-09-25 Jeff Law <law@redhat.com>
9920 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
9921 prototype for new argument.
9922 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
9923 mostly extracted from rs6000_emit_allocate_stack.
9924 (rs6000_emit_probe_stack_range_stack_clash): New function.
9925 (rs6000_emit_allocate_stack): Call
9926 rs6000_emit_probe_stack_range_stack_clash as needed.
9927 (rs6000_emit_probe_stack_range): Add additional argument
9928 to call to gen_probe_stack_range{si,di}.
9929 (output_probe_stack_range): New.
9930 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
9931 (output_probe_stack_range_stack_clash): New.
9932 (rs6000_emit_prologue): Emit notes into dump file as requested.
9933 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
9934 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
9935 Add additional operand and pass it to output_probe_stack_range.
9937 2017-09-25 Bin Cheng <bin.cheng@arm.com>
9939 PR tree-optimization/82163
9940 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
9941 (checking_verify_loop_closed_ssa): New parameter.
9942 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
9943 (check_loop_closed_ssa_stmt): Delete.
9944 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
9945 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
9946 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
9949 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
9951 * brig-builtins.def: Treat HSAIL barrier builtins as
9952 setjmp/longjump style functions.
9954 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
9956 * target.def (constant_alignment): New hook.
9957 * defaults.h (CONSTANT_ALIGNMENT): Delete.
9958 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
9959 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
9960 * doc/tm.texi: Regenerate.
9961 * targhooks.h (default_constant_alignment): Declare.
9962 (constant_alignment_word_strings): Likewise.
9963 * targhooks.c (default_constant_alignment): New function.
9964 (constant_alignment_word_strings): Likewise.
9965 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
9966 instead of CONSTANT_ALIGNMENT.
9967 * varasm.c (align_variable, get_variable_align, build_constant_desc)
9968 (force_const_mem): Likewise.
9969 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
9970 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
9971 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
9972 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9973 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
9975 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
9976 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9977 constant_alignment_word_strings.
9978 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
9979 (CONSTANT_ALIGNMENT): Likewise.
9980 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9981 (arm_constant_alignment): New function.
9982 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
9983 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9984 constant_alignment_word_strings.
9985 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
9986 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9987 constant_alignment_word_strings.
9988 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
9989 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9990 (cris_constant_alignment): New function.
9991 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
9992 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
9993 (epiphany_constant_alignment): New function.
9994 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
9995 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9996 constant_alignment_word_strings.
9997 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
9998 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
9999 constant_alignment_word_strings.
10000 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
10001 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10002 constant_alignment_word_strings.
10003 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
10004 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
10005 * config/i386/i386.c (ix86_constant_alignment): Make static.
10006 Use the same interface as the target hook.
10007 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10008 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
10009 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10010 constant_alignment_word_strings.
10011 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
10012 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
10013 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10014 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
10015 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10016 constant_alignment_word_strings.
10017 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
10018 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10019 constant_alignment_word_strings.
10020 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
10021 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10022 constant_alignment_word_strings.
10023 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
10024 * config/microblaze/microblaze.c (microblaze_constant_alignment):
10026 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10027 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
10028 * config/mips/mips.c (mips_constant_alignment): New function.
10029 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10030 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
10031 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
10032 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10033 (mmix_constant_alignment): Make static. Use the same interface
10034 as the target hook.
10035 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
10036 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10037 constant_alignment_word_strings.
10038 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
10039 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10040 constant_alignment_word_strings.
10041 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
10042 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10043 constant_alignment_word_strings.
10044 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
10045 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10046 (rs6000_constant_alignment): New function.
10047 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
10048 * config/riscv/riscv.c (riscv_constant_alignment): New function.
10049 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10050 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
10051 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10052 (rs6000_constant_alignment): New function.
10053 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
10054 * config/s390/s390.c (s390_constant_alignment): New function.
10055 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10056 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
10057 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10058 constant_alignment_word_strings.
10059 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
10060 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10061 (sparc_constant_alignment): New function.
10062 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
10063 * config/spu/spu.c (spu_constant_alignment): New function.
10064 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10065 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
10066 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10067 constant_alignment_word_strings.
10068 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
10069 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10070 constant_alignment_word_strings.
10071 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
10072 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
10073 constant_alignment_word_strings.
10074 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
10075 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10076 (visium_constant_alignment): New function.
10077 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
10078 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
10079 (xtensa_constant_alignment): New function.
10080 * system.h (CONSTANT_ALIGNMENT): Poison.
10082 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
10084 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10085 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
10086 (rs6000_builtin_valid_without_lhs): New helper function.
10087 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10088 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
10090 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
10092 * target.h (vec_perm_indices): Use unsigned short rather than
10094 (auto_vec_perm_indices): Likewise.
10095 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
10096 Use unsigned int rather than unsigned char.
10097 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
10099 2017-09-25 Richard Biener <rguenther@suse.de>
10101 * cfgloop.h (sort_sibling_loops): Declare.
10102 * cfgloop.c (sort_sibling_loops_cmp): New helper.
10103 (sort_sibling_loops): New function sorting the sibling loop list
10105 * graphite.c (graphite_transform_loops): Sort sibling loops.
10107 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
10109 * target.def (vec_perm_const_ok): Change sel parameter to
10111 * optabs-query.c (can_vec_perm_p): Update accordingly.
10112 * doc/tm.texi: Regenerate.
10113 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
10114 auto_vec_perm_indices and remove separate nelt field.
10115 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
10116 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
10117 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
10118 (aarch64_expand_vec_perm_const): Update accordingly.
10119 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
10120 to vec_perm_indices.
10121 * config/arm/arm.c (expand_vec_perm_d): Change perm to
10122 auto_vec_perm_indices and remove separate nelt field.
10123 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
10124 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
10125 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
10127 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
10128 to vec_perm_indices.
10129 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
10130 sel to vec_perm_indices.
10131 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
10132 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
10133 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
10135 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
10138 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
10141 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
10142 on the FUNCTION_DECL function context if it has a DIE that is a
10145 2017-09-25 Richard Biener <rguenther@suse.de>
10147 PR tree-optimization/82285
10148 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
10151 2017-09-25 Tom de Vries <tom@codesourcery.com>
10155 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
10158 2017-09-25 Richard Biener <rguenther@suse.de>
10160 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
10161 ISL errors other than isl_error_quota happen. Dump if the
10162 schedule is the same.
10163 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
10164 errors instead of aborting inside ISL.
10166 2017-09-25 Iain Sandoe <iain@codesourcery.com>
10169 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
10170 of libgcc_eh for m64.
10171 * config/i386/darwin64.h: Likewise.
10173 2017-09-25 Richard Biener <rguenther@suse.de>
10175 PR middle-end/82144
10176 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
10177 attribute for incomplete types nor twice for complete ones.
10179 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
10182 * config/i386/i386.c (ix86_print_operand_address_as): Encode
10183 %esp as %rsp to avoid 0x67 prefix if there is no index or base
10186 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
10189 * config/i386/i386.opt (mprefer-avx256): Use
10190 ix86_target_flags variable.
10191 * config/i386/i386.c (ix86_target_string): Move
10192 -mprefer-avx256 to flag2_opts.
10194 2017-09-22 Jakub Jelinek <jakub@redhat.com>
10196 PR middle-end/35691
10197 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
10198 and x != -1 | y != -1 into (x & y) != -1.
10200 2017-09-22 Steve Ellcey <sellcey@cavium.com>
10202 * config.gcc: Add new case statement to set
10203 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
10204 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
10205 s390x-*-linux* case statements. Added aarch64 to the list of
10206 supported architectures.
10208 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10210 PR tree-optimization/82289
10211 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
10212 STMT_VINFO_RELEVANT_P.
10214 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10215 Alan Hayward <alan.hayward@arm.com>
10216 David Sherwood <david.sherwood@arm.com>
10218 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
10219 for VR_RANGE only; don't allow VR_ANTI_RANGE.
10220 (extract_range_from_binary_expr_1): Don't call
10221 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
10223 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10224 Alan Hayward <alan.hayward@arm.com>
10225 David Sherwood <david.sherwood@arm.com>
10227 * target.def (preferred_vector_alignment): New hook.
10228 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
10230 * doc/tm.texi: Regenerate.
10231 * targhooks.h (default_preferred_vector_alignment): Declare.
10232 * targhooks.c (default_preferred_vector_alignment): New function.
10233 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
10235 (DR_TARGET_ALIGNMENT): New macro.
10236 (aligned_access_p): Update commentary.
10237 (vect_known_alignment_in_bytes): New function.
10238 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
10240 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
10241 Calculate the misalignment based on the target alignment rather than
10243 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
10244 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
10245 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
10246 the target alignment, rather than masking the element misalignment
10247 with the number of elements in a vector. Also use the target
10248 alignment when calculating the maximum number of peels.
10249 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
10250 instead of TYPE_ALIGN_UNIT.
10251 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
10252 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
10253 (vect_create_addr_base_for_vector_ref): Update call accordingly.
10254 (vect_create_data_ref_ptr): Likewise.
10255 (vect_setup_realignment): Realign by ANDing with
10256 -DR_TARGET_MISALIGNMENT.
10257 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
10258 the number of peels based on DR_TARGET_ALIGNMENT.
10259 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
10260 with the guaranteed alignment boundary when deciding whether
10262 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
10263 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
10264 (ensure_base_align): Remove stmt_info parameter. Get the
10265 target base alignment from DR_TARGET_ALIGNMENT.
10266 (vectorizable_store): Update call accordingly. Interpret
10267 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
10269 (vectorizable_load): Likewise.
10271 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
10272 Alan Hayward <alan.hayward@arm.com>
10273 David Sherwood <david.sherwood@arm.com>
10275 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
10276 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
10277 (vect_enhance_data_refs_alignment): Likewise.
10279 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10281 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10282 error. Only quit immediately if parsing is complete.
10283 (BEGIN): Initialize fatal_err and parse_done.
10284 (begin fpu, end fpu): Check number of arguments.
10285 (begin arch, end arch): Likewise.
10286 (begin cpu, end cpu): Likewise.
10287 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10288 (optalias): Likewise.
10290 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
10292 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10293 * config/arm/arm-isa.h: Delete. Move definitions to ...
10294 * arm-cpus.in: ... here. Use new feature and fgroup values.
10295 * config/arm/arm.c (arm_option_override): Use lower case for feature
10297 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10298 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10299 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10301 (print_isa_bits_for): New function.
10302 (gen_isa): New function.
10303 (gen_comm_data): Use print_isa_bits_for.
10304 (define feature): New keyword.
10305 (define fgroup): New keyword.
10306 * config/arm/t-arm (TM_H): Remove.
10307 (GTM_H): Add arm-isa.h.
10308 (arm-isa.h): Add rule to generate file.
10309 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10310 case for feature bit names.
10312 2017-09-22 Richard Biener <rguenther@suse.de>
10314 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
10316 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
10317 print of no dependency loops ...
10318 * graphite.c (graphite_transform_loops): ... here.
10319 (canonicalize_loop_closed_ssa_form): Work from inner to outer
10321 (same_close_phi_node, remove_duplicate_close_phi,
10322 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
10323 (canonicalize_loop_closed_ssa): ... here and simplify.
10324 * graphite-optimize-isl.c: Include tree-vectorizer.h.
10325 (optimize_isl): Use dump_printf_loc to tell when we stopped
10326 optimizing because of an ISL timeout.
10328 2017-09-22 Richard Biener <rguenther@suse.de>
10330 PR tree-optimization/82291
10331 * tree-if-conv.c (predicate_mem_writes): Make sure to
10332 remove writes in blocks predicated with false.
10334 2017-09-22 Richard Biener <rguenther@suse.de>
10336 * sese.c: Include cfganal.h.
10337 (if_region_set_false_region): Remove.
10338 (create_if_region_on_edge): Likewise.
10339 (move_sese_in_condition): Re-implement without destroying
10342 2017-09-22 Richard Biener <rguenther@suse.de>
10344 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
10345 Verify both BBs contain loop PHI nodes before dispatching to
10346 copy_loop_phi_args.
10347 (graphite_regenerate_ast_isl): Do not recompute dominators,
10348 do not verify three times. Restructure for clarity.
10349 * graphite-scop-detection.c (same_close_phi_node,
10350 remove_duplicate_close_phi, make_close_phi_nodes_unique,
10351 defined_in_loop_p, canonicalize_loop_closed_ssa,
10352 canonicalize_loop_closed_ssa_form): Simplify, remove excess
10353 checking and SSA rewrite, move to ...
10354 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
10355 (graphite_initialize): Do not pass in ctx, do not reset the
10356 SCEV cache, compute only dominators.
10357 (graphite_transform_loops): Allocate ISL ctx after
10358 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
10359 Maintain post-dominators only around build_scops.
10360 * sese.c (if_region_set_false_region): Make static. Free
10361 and recompute dominators.
10362 (move_sese_in_condition): Assert we don't get called with
10363 post-dominators computed.
10364 * sese.h (if_region_set_false_region): Remove.
10366 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
10368 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
10369 mode attribute for TARGET_AVX512VL.
10371 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
10373 * config/i386/i386.opt (mprefer-avx256): New option.
10374 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
10376 (ix86_preferred_simd_mode): Return 256-bit AVX modes
10377 for TARGET_PREFER_AVX256.
10378 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
10380 2017-09-21 Jeff Law <law@redhat.com>
10382 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
10383 Fix dump output if the only stack space is for pushed registers.
10385 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10387 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
10390 2017-09-21 Martin Sebor <msebor@redhat.com>
10393 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
10394 code (in C++) or code that triggers warnings.
10396 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
10398 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
10400 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
10402 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
10403 * sched-rgn.c: Ditto.
10404 * sel-sched-ir.c: Ditto.
10406 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
10408 * toplev.h (set_random_seed): Adjust return type.
10409 * toplev.c (init_local_tick): Move eager initialization of random_seed
10410 to get_random_seed. Adjust comment.
10411 (init_random_seed): Inline to get_random_seed, delete.
10412 (get_random_seed): Initialize random_seed lazily.
10413 (set_random_seed): Do not return previous value.
10414 (print_switch_value): Do not call get_random_seed.
10416 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
10418 * cgraph.c (delete_function_version): New, broken out from...
10419 (cgraph_node::delete_function_version): ...here. Rename to
10420 cgraph_node::delete_function_version_by_decl. Update all uses.
10421 (cgraph_node::remove): Call delete_function_version.
10423 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10426 * tree-inline.c (expand_call_inline): Emit clobber stmts for
10427 VAR_DECLs to which addressable non-volatile parameters are mapped
10428 and for id->retvar after the return value assignment. Clear
10429 id->retval and id->retbnd after inlining.
10431 2017-09-21 Richard Biener <rguenther@suse.de>
10433 PR tree-optimization/82276
10434 PR tree-optimization/82244
10435 * tree-vrp.c (build_assert_expr_for): Set
10436 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
10438 (remove_range_assertions): Revert earlier change.
10440 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
10443 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
10445 2017-09-21 Richard Biener <rguenther@suse.de>
10447 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10448 Restore valid IL after code generation errors.
10449 * graphite.c (graphite_transform_loops): Diagnose code
10450 generation issues as MSG_MISSED_OPTIMIZATION and continue
10451 with processing SCOPs.
10453 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10454 Alan Hayward <alan.hayward@arm.com>
10455 David Sherwood <david.sherwood@arm.com>
10457 * calls.c (compute_argument_addresses): Use simplify_gen_binary
10458 rather than choosing between plus_constant and gen_rtx_<CODE>.
10459 * expr.c (emit_push_insn): Likewise.
10460 (expand_expr_real_2): Likewise.
10462 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10463 Alan Hayward <alan.hayward@arm.com>
10464 David Sherwood <david.sherwood@arm.com>
10466 * loop-unroll.c (split_iv): Call copy_rtx on the step.
10468 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10469 Alan Hayward <alan.hayward@arm.com>
10470 David Sherwood <david.sherwood@arm.com>
10472 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
10473 calling tree_to_uhwi.
10475 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10476 Alan Hayward <alan.hayward@arm.com>
10477 David Sherwood <david.sherwood@arm.com>
10479 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
10480 INTEGER_CST rather than a negative test for ADDR_EXPR.
10482 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
10483 Alan Hayward <alan.hayward@arm.com>
10484 David Sherwood <david.sherwood@arm.com>
10486 * tree-vrp.c (extract_range_from_binary_expr_1): Check
10487 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
10489 2017-09-21 Richard Biener <rguenther@suse.de>
10491 PR tree-optimization/71351
10492 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
10493 graphite_create_new_loop_guard): Remove, fold remaining parts
10495 (translate_isl_ast_node_for): ... here and simplify.
10497 2017-09-21 Jakub Jelinek <jakub@redhat.com>
10500 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
10501 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
10502 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
10503 alternative always use QI mode, for -Os imov (=R,R) alternative
10504 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
10507 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10508 Jeff Law <law@redhat.com>
10510 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
10511 (allocate_stack_space): New function, partially extracted from
10512 s390_emit_prologue.
10513 (s390_emit_prologue): Track offset to most recent stack probe.
10514 Code to allocate space moved into allocate_stack_space.
10515 Dump actions when no stack is allocated.
10516 (s390_prologue_plus_offset): New function.
10517 (s390_emit_stack_probe): Likewise.
10519 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
10521 * common.opt (Wa, Wl, Wp, g, gz=): Add
10523 (gno-column-info): Remove.
10524 (gcolumn-info): Drop RejectNegative.
10525 (gno-): New prefix.
10526 (gno-record-gcc-switches): Remove.
10527 (grecord-gcc-switches): Drop RejectNegative.
10528 (gno-split-dwarf): Remove.
10529 (gsplit-dwarf): Drop RejectNegative.
10530 (gno-strict-dwarf): Remove.
10531 (gstrict-dwarf): Drop RejectNegative.
10532 * config/darwin.opt (gfull, gused): Add RejectNegative.
10533 * dwarf2out.c (gen_producer_string): Drop
10534 gno-record-gcc-switches handler.
10535 * optc-gen.awk: Add g to prefixes with negative forms.
10536 * opts-common.c (remapping_prefix_p): New.
10537 (find_opt): Check it.
10538 (generate_canonical_option): Test g prefix.
10539 (option_map): Add -gno- mapping.
10540 (add_misspelling_candidates): Check remapping_prefix_p.
10542 2017-09-20 Jeff Law <law@redhat.com>
10544 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
10545 thinko in stack clash protection support.
10547 * explow.c (compute_stack_clash_protection_loop_data): Use
10548 CONST_INT_P instead of explicit test. Verify object is a
10549 CONST_INT_P before looking at INTVAL.
10550 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
10551 instead of explicit test.
10553 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
10556 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
10557 address instead of to r1 and r11.
10559 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
10561 * config.gcc: Support "knm".
10562 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
10563 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
10565 * config/i386/i386.c (m_KNM): Define.
10566 (processor_target_table): Add "knm".
10568 (ix86_option_override_internal): Add "knm".
10569 (ix86_issue_rate): Add PROCESSOR_KNM.
10570 (ix86_adjust_cost): Ditto.
10571 (ia32_multipass_dfa_lookahead): Ditto.
10572 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
10573 (fold_builtin_cpu): Add M_INTEL_KNM.
10574 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
10575 (processor_type): Add PROCESSOR_KNM.
10576 * config/i386/x86-tune.def: Add m_KNM.
10577 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
10579 2017-09-20 Richard Biener <rguenther@suse.de>
10581 PR tree-optimization/80213
10582 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
10583 are allowed in empty BBs as well.
10584 (canonicalize_loop_closed_ssa): Also look for other complex
10586 (scop_detection::get_sese): Include the loop-closed PHI block
10588 (scop_detection::merge_sese): Remove code adding extra blocks.
10589 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
10590 (build_scops): Assert the final returned scop is invalid.
10592 2017-09-20 Richard Biener <rguenther@suse.de>
10594 PR tree-optimization/82264
10595 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
10597 (vn_phi_lookup): Likewise.
10598 (vn_phi_insert): Likewise.
10600 2017-09-20 Jakub Jelinek <jakub@redhat.com>
10602 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
10603 that fits into uhwi or shwi, add DW_AT_const_value regardless
10604 of early_dwarf without going through RTL, using add_AT_unsigned
10607 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
10608 (DEBUG_LTO_ABBREV_SECTION): Likewise.
10609 (DEBUG_LTO_MACINFO_SECTION): Likewise.
10610 (DEBUG_MACRO_SECTION): Likewise.
10611 (DEBUG_LTO_MACRO_SECTION): Likewise.
10612 (DEBUG_STR_DWO_SECTION): Likewise.
10613 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
10614 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
10615 (DEBUG_LTO_DWO_LINE_SECTION): Define.
10616 (DEBUG_LTO_LINE_STR_SECTION): Define.
10617 (init_sections_and_labels): Initialize debug_line_str_section
10618 variable. Initialize debug_loc_section for -gdwarf-5 to
10619 DEBUG_LOCLISTS_SECTION. Formatting fixes.
10621 2017-09-20 Richard Biener <rguenther@suse.de>
10623 * graphite-sese-to-poly.c (extract_affine): Properly handle
10624 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
10626 2017-09-20 Richard Biener <rguenther@suse.de>
10628 PR tree-optimization/81373
10629 * graphite-scop-detection.c (build_cross_bb_scalars_def):
10630 Force SESE live-out defs to be handled even if they are
10633 2017-09-19 Jeff Law <law@redhat.com>
10635 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
10636 nothing for stack adjustments with REG_STACK_CHECK.
10637 * sched-deps.c (parse_add_or_inc): Reject insns with
10638 REG_STACK_CHECK from dependency breaking.
10639 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
10640 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
10641 * reg-notes.def (STACK_CHECK): New note.
10643 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
10644 (ix86_expand_prologue): Dump stack clash info as needed.
10645 Call ix86_adjust_stack_and_probe_stack_clash as needed.
10647 * function.c (dump_stack_clash_frame_info): New function.
10648 * function.h (dump_stack_clash_frame_info): Prototype.
10649 (enum stack_clash_probes): New enum.
10651 * config/alpha/alpha.c (alpha_expand_prologue): Also check
10652 flag_stack_clash_protection.
10653 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
10654 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
10655 (arm_frame_pointer_required): Likewise.
10656 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
10657 (ia64_expand_prologue): Likewise.
10658 * config/mips/mips.c (mips_expand_prologue): Likewise.
10659 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
10660 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10661 (sparc_flat_expand_prologue): Likewise.
10662 * config/spu/spu.c (spu_expand_prologue): Likewise.
10664 * explow.c: Include "params.h".
10665 (anti_adjust_stack_and_probe_stack_clash): New function.
10666 (get_stack_check_protect): Likewise.
10667 (compute_stack_clash_protection_loop_data): Likewise.
10668 (emit_stack_clash_protection_loop_start): Likewise.
10669 (emit_stack_clash_protection_loop_end): Likewise.
10670 (allocate_dynamic_stack_space): Use get_stack_check_protect.
10671 Use anti_adjust_stack_and_probe_stack_clash.
10672 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
10673 (emit_stack_clash_protection_loop_start): Likewise.
10674 (emit_stack_clash_protection_loop_end): Likewise.
10675 * rtl.h (get_stack_check_protect): Prototype.
10676 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
10677 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
10678 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
10680 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
10682 * doc/tm.texi: Rebuilt.
10683 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
10684 get_stack_check_protect.
10685 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
10686 * config/arm/arm.c (arm_expand_prologue): Likewise.
10687 (arm_frame_pointer_required): Likewise.
10688 * config/i386/i386.c (ix86_expand_prologue): Likewise.
10689 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
10690 * config/mips/mips.c (mips_expand_prologue): Likewise.
10691 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
10692 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
10693 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
10694 (sparc_flat_expand_prologue): Likewise.
10696 * common.opt (-fstack-clash-protection): New option.
10697 * flag-types.h (enum stack_check_type): Note difference between
10698 -fstack-check= and -fstack-clash-protection.
10699 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
10700 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
10701 * toplev.c (process_options): Issue warnings/errors for cases
10702 not handled with -fstack-clash-protection.
10703 * doc/invoke.texi (-fstack-clash-protection): Document new option.
10704 (-fstack-check): Note additional problem with -fstack-check=generic.
10705 Note that -fstack-check is primarily for Ada and refer users
10706 to -fstack-clash-protection for stack-clash-protection.
10707 Document new params for stack clash protection.
10709 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
10711 * config/i386/i386.c (ix86_split_long_move): Do not handle
10712 address used for LEA in a special way.
10714 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
10716 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
10718 2017-09-19 Martin Sebor <msebor@redhat.com>
10721 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
10722 of incompatible types.
10724 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
10726 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10727 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
10728 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10729 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
10731 2017-09-19 Richard Biener <rguenther@suse.de>
10733 PR tree-optimization/82244
10734 * tree-vrp.c (remove_range_assertions): Do not propagate
10735 a constant to abnormals but replace the assert with a copy.
10737 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
10739 PR rtl-optimization/57878
10740 PR rtl-optimization/68988
10741 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
10742 avoidance test involving non_reload_pseudos. Move frequency test
10743 below the general fragmentation avoidance test.
10745 2017-09-19 Richard Biener <rguenther@suse.de>
10747 PR tree-optimization/69728
10748 * graphite-sese-to-poly.c (schedule_error): New global.
10749 (add_loop_schedule): Handle empty domain by failing the
10751 (build_original_schedule): Handle schedule_error.
10753 2017-09-19 Richard Biener <rguenther@suse.de>
10755 * graphite-scop-detection.c (scop_detection::can_represent_loop):
10756 Do not iterate to sibling loops but only to siblings of inner
10759 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
10762 * config/m68k/m68k.md (moveq feeding equality comparison): Check
10763 that the registers are different.
10765 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
10767 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
10768 to processor_model and "amdfam17h" to arch_names_table.
10769 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
10771 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10774 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
10776 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10777 Alan Hayward <alan.hayward@arm.com>
10778 David Sherwood <david.sherwood@arm.com>
10780 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
10782 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
10784 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
10785 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
10786 vect_schedule_slp_instance.
10787 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
10788 Update call to vect_slp_analyze_node_operations. Simplify return
10790 (vect_slp_analyze_bb_1): Update call accordingly.
10791 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
10792 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
10793 (vect_schedule_slp): Update call accordingly.
10795 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10796 Alan Hayward <alan.hayward@arm.com>
10797 David Sherwood <david.sherwood@arm.com>
10799 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
10800 with types that aren't in fact scalar.
10802 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10804 * tree-vect-slp.c (vect_record_max_nunits): New function,
10806 (vect_build_slp_tree_1): ...here.
10807 (vect_build_slp_tree_2): Call it for phis too.
10809 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10811 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
10812 to vect_get_vec_def_for_operand when getting the mask operand.
10814 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10815 Alan Hayward <alan.hayward@arm.com>
10816 David Sherwood <david.sherwood@arm.com>
10818 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
10821 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10822 Alan Hayward <alan.hayward@arm.com>
10823 David Sherwood <david.sherwood@arm.com>
10825 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
10826 calculation for vector booleans.
10828 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
10829 Alan Hayward <alan.hayward@arm.com>
10830 David Sherwood <david.sherwood@arm.com>
10832 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
10834 (vect_transform_stmt): ...here.
10835 (vect_analyze_stmt): Use it instead of calling
10836 vectorizable_live_operation directly.
10838 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
10840 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
10841 non-SIMT targets in acc vector loops.
10843 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
10845 * configure.ac: Add arc and check if assembler supports gdwarf2.
10846 * configure: Regenerate.
10848 2017-09-18 Richard Biener <rguenther@suse.de>
10850 PR tree-optimization/82220
10851 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
10852 epilogue niters from the min_profitable_iters compute.
10854 2017-09-18 Jakub Jelinek <jakub@redhat.com>
10857 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
10858 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
10859 (ix86_init_pic_reg): Revert 2017-09-01 changes.
10861 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
10864 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
10865 switching to a new text section.
10867 2017-09-18 Richard Biener <rguenther@suse.de>
10869 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
10871 (build_alias_set): Reject aliases with no access function.
10873 2017-09-18 Richard Biener <rguenther@suse.de>
10875 PR tree-optimization/79622
10876 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
10878 (build_cross_bb_scalars_use): Likewise.
10880 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
10882 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
10884 2017-09-18 Alan Modra <amodra@gmail.com>
10887 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
10888 stack_pointer_rtx for count 0. Update comments. Break up
10889 large rtl expression.
10891 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
10893 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
10894 Increase to 20 bytes.
10895 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
10896 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
10897 or avx version of the stub.
10899 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
10902 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
10903 compute the minimum stack alignment. Also update preferred stack
10904 boundary for leaf functions.
10906 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
10908 PR tree-optimization/82228
10909 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
10912 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10914 * common/config/nds32/nds32-common.c
10915 (nds32_option_optimization_table): Refine formatting.
10916 (nds32_option_optimization_table): Use -fsched-pressure and
10917 -fomit-frame-pointer for specific optimization level.
10919 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
10921 * config/nds32/nds32.c: Refine formatting and comments.
10922 * config/nds32/nds32.h: Likewise.
10923 * config/nds32/nds32.md: Likewise.
10924 * config/nds32/nds32-cost.c: Likewise.
10925 * config/nds32/nds32-isr.c: Likewise.
10926 * config/nds32/nds32-md-auxiliary.c: Likewise.
10927 * config/nds32/nds32-multiple.md: Likewise.
10928 * config/nds32/nds32-predicates.c: Likewise.
10930 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
10931 Jakub Jelinek <jakub@redhat.com>
10933 Add support for -std=c++2a.
10934 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
10936 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
10938 2017-09-15 Steve Ellcey <sellcey@cavium.com>
10941 * doc/extend.texi (Common Function Attributes): Add
10942 references to ARM, AArch64, and S/390 specific attributes.
10943 (Function Specific Option Pragmas): Add AArch64 and S/390
10944 to list of back ends that support the target pragma.
10946 2017-09-15 Nathan Sidwell <nathan@acm.org>
10948 * doc/standards.texi: Fix C++17 description. Update URLs for
10951 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
10953 * common.opt (Wcast-align=strict): New warning option.
10954 * doc/invoke.texi: Document -Wcast-align=strict.
10956 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
10958 * cgraph.h (cgraph_thunk_info): Add comments.
10959 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
10960 assert for VIRTUAL_* arguments stricter.
10962 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
10964 PR tree-optimization/71026
10965 * match.pd: Move RDIV patterns from fold-const.c
10966 * fold-const.c (distribute_real_division): Removed.
10967 (fold_binary_loc): Remove calls to distribute_real_divison.
10969 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10971 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
10972 c++1z and gnu++1z as deprecated. Change other references to
10973 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
10974 Change -Wc++1z-compat to -Wc++17-compat.
10975 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
10976 * dwarf2out.c (highest_c_language): Handle C++17.
10977 (gen_compile_unit_die): Likewise.
10979 2017-09-15 Jakub Jelinek <jakub@redhat.com>
10981 PR rtl-optimization/82192
10982 * combine.c (make_extraction): Don't look through non-paradoxical
10983 SUBREGs or TRUNCATE if pos + len is or might be bigger than
10986 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
10987 Alan Hayward <alan.hayward@arm.com>
10988 David Sherwood <david.sherwood@arm.com>
10990 * target.def (function_arg_offset): New hook.
10991 * targhooks.h (default_function_arg_offset): Declare.
10992 * targhooks.c (default_function_arg_offset): New function.
10993 * function.c (locate_and_pad_parm): Use
10994 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
10995 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
10996 (TARGET_FUNCTION_ARG_OFFSET): ...this.
10997 * doc/tm.texi: Regenerate.
10998 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
10999 * config/spu/spu.c (spu_function_arg_offset): New function.
11000 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
11001 * system.h (FUNCTION_ARG_OFFSET): Poison.
11003 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11004 Alan Hayard <alan.hayward@arm.com>
11005 David Sherwood <david.sherwood@arm.com>
11007 * target.def (truly_noop_truncation): New hook.
11008 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
11009 than TRULY_NOOP_TRUNCATION.
11010 * hooks.h (hook_bool_uint_uint_true): Declare.
11011 * hooks.c (hook_bool_uint_uint_true): New function.
11012 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
11013 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
11014 * doc/tm.texi: Regenerate.
11015 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
11016 rather than TRULY_NOOP_TRUNCATION in comments.
11017 (simplify_comparison): Likewise.
11018 (record_truncated_value): Likewise.
11019 * expmed.c (extract_bit_field_1): Likewise.
11020 (extract_split_bit_field): Likewise.
11021 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
11022 instead of TRULY_NOOP_TRUNCATION.
11023 * function.c (assign_parm_setup_block): Likewise.
11024 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
11025 * rtlhooks.c: Include target.h.
11026 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
11027 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
11028 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
11029 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
11030 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
11031 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
11032 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
11033 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
11034 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
11035 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
11036 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
11037 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
11038 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
11039 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
11040 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
11041 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
11042 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
11043 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
11044 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
11045 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
11046 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
11047 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
11048 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
11049 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
11050 * config/mips/mips.c (mips_truly_noop_truncation): New function.
11051 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11052 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11053 rather than TRULY_NOOP_TRUNCATION in comments.
11054 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
11055 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
11056 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
11057 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
11058 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
11059 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
11060 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
11061 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
11062 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
11063 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
11064 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
11065 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11066 rather than TRULY_NOOP_TRUNCATION in comments.
11067 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
11068 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
11069 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
11070 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
11071 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
11072 TRULY_NOOP_TRUNCATION condition.
11073 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
11074 (TRULY_NOOP_TRUNCATION): Delete.
11075 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
11076 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
11077 * config/spu/spu.c (spu_truly_noop_truncation): New function.
11078 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11079 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
11080 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
11081 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
11082 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
11083 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
11084 rather than TRULY_NOOP_TRUNCATION in comments.
11085 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
11086 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
11087 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
11088 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
11089 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
11090 * system.h (TRULY_NOOP_TRUNCATION): Poison.
11092 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
11095 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
11096 (*cmp_ior): Likewise.
11097 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
11098 (*ior_scc_scc_cmp): Likewise.
11099 (*and_scc_scc): Likewise.
11100 (*and_scc_scc_cmp): Likewise.
11102 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
11103 Alan Hayard <alan.hayward@arm.com>
11104 David Sherwood <david.sherwood@arm.com>
11106 * target.def (can_change_mode_class): New hook.
11107 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
11108 (hard_regno_nregs): Likewise.
11109 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
11110 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
11111 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
11112 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
11113 (LOAD_EXTEND_OP): Update accordingly.
11114 * doc/tm.texi: Regenerate.
11115 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
11116 CANNOT_CHANGE_MODE_CLASS.
11117 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
11118 (REG_CAN_CHANGE_MODE_P): ...this new macro.
11119 * combine.c (simplify_set): Update accordingly.
11120 * emit-rtl.c (validate_subreg): Likewise.
11121 * recog.c (general_operand): Likewise.
11122 * regcprop.c (mode_change_ok): Likewise.
11123 * reload1.c (choose_reload_regs): Likewise.
11124 (inherit_piecemeal_p): Likewise.
11125 * rtlanal.c (simplify_subreg_regno): Likewise.
11126 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
11127 instead of CANNOT_CHANGE_MODE_CLASS.
11128 (reload_cse_simplify_operands): Likewise.
11129 * reload.c (push_reload): Use targetm.can_change_mode_class
11130 instead of CANNOT_CHANGE_MODE_CLASS.
11131 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
11132 REG_CANNOT_CHANGE_MODE_P.
11133 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11134 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
11135 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11136 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11137 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11138 (arm_can_change_mode_class): New function.
11139 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11140 than CANNOT_CHANGE_MODE_CLASS in comments.
11141 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11142 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
11143 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
11144 (ix86_can_change_mode_class): ...this new function, inverting the
11145 sense of the return value.
11146 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11147 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11148 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11149 (ia64_can_change_mode_class): New function.
11150 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11151 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
11152 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
11153 (m32c_can_change_mode_class): ...this new function, inverting the
11154 sense of the return value.
11155 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11156 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11157 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
11158 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
11159 (mips_can_change_mode_class): ...this new function, inverting the
11160 sense of the return value.
11161 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11162 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11163 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11164 (msp430_can_change_mode_class): New function.
11165 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11166 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
11167 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11168 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11169 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11170 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
11171 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11172 (pa_cannot_change_mode_class): Replace with...
11173 (pa_can_change_mode_class): ...this new function, inverting the
11174 sense of the return value.
11175 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
11176 than CANNOT_CHANGE_MODE_CLASS in comments.
11177 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11178 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
11179 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11180 (pdp11_cannot_change_mode_class): Replace with...
11181 (pdp11_can_change_mode_class): ...this new function, inverting the
11182 sense of the return value.
11183 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11184 * config/powerpcspe/powerpcspe-protos.h
11185 (rs6000_cannot_change_mode_class_ptr): Delete.
11186 * config/powerpcspe/powerpcspe.c
11187 (rs6000_cannot_change_mode_class_ptr): Delete.
11188 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11189 (rs6000_option_override_internal): Assign to
11190 targetm.can_change_mode_class instead of
11191 rs6000_cannot_change_mode_class_ptr.
11192 (rs6000_cannot_change_mode_class): Replace with...
11193 (rs6000_can_change_mode_class): ...this new function, inverting the
11194 sense of the return value.
11195 (rs6000_debug_cannot_change_mode_class): Replace with...
11196 (rs6000_debug_can_change_mode_class): ...this new function.
11197 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11198 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
11199 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11200 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11201 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
11203 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
11204 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11205 (rs6000_option_override_internal): Assign to
11206 targetm.can_change_mode_class instead of
11207 rs6000_cannot_change_mode_class_ptr.
11208 (rs6000_cannot_change_mode_class): Replace with...
11209 (rs6000_can_change_mode_class): ...this new function, inverting the
11210 sense of the return value.
11211 (rs6000_debug_cannot_change_mode_class): Replace with...
11212 (rs6000_debug_can_change_mode_class): ...this new function.
11213 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11214 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
11215 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
11216 (s390_can_change_mode_class): ...this new function, inverting the
11217 sense of the return value.
11218 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11219 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11220 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
11221 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11222 (sh_cannot_change_mode_class): Replace with...
11223 (sh_can_change_mode_class): ...this new function, inverting the
11224 sense of the return value.
11225 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11226 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11227 (sparc_can_change_mode_class): New function.
11228 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11229 * config/spu/spu.c (spu_can_change_mode_class): New function.
11230 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11231 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
11232 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
11233 (visium_can_change_mode_class): New function.
11234 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
11236 2017-09-15 Richard Biener <rguenther@suse.de>
11238 PR tree-optimization/82217
11239 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
11240 but not undefined case.
11242 2017-09-15 Jakub Jelinek <jakub@redhat.com>
11245 * postreload.c (reload_cse_simplify_operands): Skip
11246 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
11248 2017-09-15 Richard Biener <rguenther@suse.de>
11250 PR tree-optimization/68823
11251 * graphite-scop-detection.c (build_alias_set): If we have a
11252 possible dependence check whether we can handle them by just
11253 looking at the DRs DR_ACCESS_FNs.
11254 (build_scops): If build_alias_set fails, fail the SCOP.
11256 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11258 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
11259 to support float128 built-in functions that require the ISA 3.0
11261 (BU_FLOAT128_3_HW): Likewise.
11262 (SQRTF128): Add support for the IEEE 128-bit square root and fma
11263 built-in functions.
11264 (FMAF128): Likewise.
11266 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
11267 support for built-in functions that need the ISA 3.0 IEEE 128-bit
11268 floating point instructions.
11269 (rs6000_invalid_builtin): Likewise.
11270 (rs6000_builtin_mask_names): Likewise.
11271 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
11272 (RS6000_BTM_FLOAT128_HW): Likewise.
11273 (RS6000_BTM_COMMON): Likewise.
11274 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
11276 * doc/extend.texi (RS/6000 built-in functions): Document the
11277 IEEE 128-bit floating point square root and fused multiply-add
11278 built-in functions.
11280 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
11282 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
11283 reg (r2) isn't in the set of registers defined in the prologue.
11285 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11286 Alan Hayward <alan.hayward@arm.com>
11287 David Sherwood <david.sherwood@arm.com>
11289 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
11290 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
11291 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
11292 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
11293 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
11295 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
11296 max_vectorization_factor.
11297 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
11299 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11300 Alan Hayward <alan.hayward@arm.com>
11301 David Sherwood <david.sherwood@arm.com>
11303 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
11304 (vect_worthwhile_without_simd_p): Declare.
11305 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
11306 (vectorizable_reduction): Use it.
11307 * tree-vect-stmts.c (vectorizable_shift): Likewise.
11308 (vectorizable_operation): Likewise.
11310 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11311 Alan Hayward <alan.hayward@arm.com>
11312 David Sherwood <david.sherwood@arm.com>
11314 * tree-vectorizer.h (vect_get_num_copies): New function.
11315 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
11316 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11317 (vectorizable_induction): Likewise.
11318 (vectorizable_live_operation): Likewise.
11319 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11320 (vectorizable_bswap): Likewise.
11321 (vectorizable_call): Likewise.
11322 (vectorizable_conversion): Likewise.
11323 (vectorizable_assignment): Likewise.
11324 (vectorizable_shift): Likewise.
11325 (vectorizable_operation): Likewise.
11326 (vectorizable_store): Likewise.
11327 (vectorizable_load): Likewise.
11328 (vectorizable_condition): Likewise.
11329 (vectorizable_comparison): Likewise.
11330 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
11332 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11333 Alan Hayward <alan.hayward@arm.com>
11334 David Sherwood <david.sherwood@arm.com>
11336 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
11337 of vect_init_vector.
11339 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11340 Alan Hayward <alan.hayward@arm.com>
11341 David Sherwood <david.sherwood@arm.com>
11343 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
11344 an inline wrapper that provides a location.
11345 (gimple_build_vector): Likewise.
11346 * gimple-fold.c (gimple_build_vector_from_val): New function.
11347 (gimple_build_vector): Likewise.
11348 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
11349 functions to build the initial value. Always return a gimple value.
11350 (get_initial_defs_for_reduction): Likewise. Only compute
11352 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
11353 vect_init_vector on the results from get_initial_def(s)_for_reduction.
11354 (vectorizable_induction): Use gimple_build_vector rather than
11357 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11358 Alan Hayward <alan.hayward@arm.com>
11359 David Sherwood <david.sherwood@arm.com>
11361 * target.h (vec_perm_indices): New typedef.
11362 (auto_vec_perm_indices): Likewise.
11363 * optabs-query.h: Include target.h
11364 (can_vec_perm_p): Take a vec_perm_indices *.
11365 * optabs-query.c (can_vec_perm_p): Likewise.
11366 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
11367 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11368 * tree-vect-generic.c (lower_vec_perm): Likewise.
11369 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
11370 (vect_grouped_load_supported): Likewise.
11371 (vect_shift_permute_load_chain): Likewise.
11372 (vect_permute_store_chain): Use auto_vec_perm_indices.
11373 (vect_permute_load_chain): Likewise.
11374 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
11375 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
11376 Update uses of can_vec_perm_p.
11377 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
11378 mode with a number of elements. Take a vec_perm_indices *.
11379 (vect_create_epilog_for_reduction): Update accordingly.
11380 Use auto_vec_perm_indices.
11381 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
11382 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
11383 (vect_transform_slp_perm_load): Likewise.
11384 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
11385 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
11386 (vect_gen_perm_mask_checked): Likewise.
11387 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
11388 (vect_gen_perm_mask_checked): Likewise.
11389 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
11390 (vectorizable_store): Likewise.
11391 (vectorizable_load): Likewise.
11392 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
11393 (vectorizable_bswap): Likewise.
11395 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11396 Alan Hayward <alan.hayward@arm.com>
11397 David Sherwood <david.sherwood@arm.com>
11399 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
11400 * tree.c (build_vector): Likewise.
11401 (build_vector_from_ctor): Update accordingly.
11402 (build_vector_from_val): Likewise.
11403 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
11404 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
11405 * tree-vect-generic.c (add_rshift): Likewise.
11406 (expand_vector_divmod): Likewise.
11407 (optimize_vector_constructor): Likewise.
11408 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
11409 (vect_transform_slp_perm_load): Likewise.
11410 (vect_schedule_slp_instance): Likewise.
11411 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
11412 (vectorizable_call): Likewise.
11413 (vect_gen_perm_mask_any): Likewise. Add elements in order.
11414 * expmed.c (make_tree): Likewise.
11415 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
11416 a vector passed to build_vector.
11417 (fold_convert_const): Likewise.
11418 (exact_inverse): Likewise.
11419 (fold_ternary_loc): Likewise.
11420 (fold_relational_const): Likewise.
11421 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
11422 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
11423 (const_unop): Likewise. Store the reduction accumulator in a
11424 variable rather than an array.
11425 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
11426 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
11427 the new vector, rather than constructing it after the input arrays.
11428 (native_interpret_vector): Use auto_vec<tree> when building
11429 a vector passed to build_vector. Add elements in order.
11430 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
11431 auto_vec<tree> when building a vector passed to build_vector.
11432 (vect_create_epilog_for_reduction): Likewise.
11433 (vectorizable_induction): Likewise.
11434 (get_initial_def_for_reduction): Likewise. Fix indentation of
11436 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
11438 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
11439 passed to build_vector.
11441 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
11442 Alan Hayward <alan.hayward@arm.com>
11443 David Sherwood <david.sherwood@arm.com>
11445 * tree-core.h (tree_base::u): Add an "nelts" field.
11446 (tree_vector): Use VECTOR_CST_NELTS as the length.
11447 * tree.c (tree_size): Likewise.
11448 (make_vector): Initialize VECTOR_CST_NELTS.
11449 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
11450 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
11451 TYPE_VECTOR_SUBPARTS.
11452 * expr.c (const_vector_mask_from_tree): Consistently use "units"
11453 as the number of units, setting it from VECTOR_CST_NELTS.
11454 (const_vector_from_tree): Likewise.
11455 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
11456 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
11457 (fold_negate_expr_1): Likewise.
11458 (fold_convert_const): Likewise.
11459 (const_binop): Likewise. Differentiate the number of output and
11461 (const_unop): Likewise.
11462 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
11463 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
11464 in cases that did the opposite.
11466 2017-09-14 Richard Biener <rguenther@suse.de>
11468 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
11471 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
11473 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
11475 2017-09-14 Jakub Jelinek <jakub@redhat.com>
11478 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
11479 if and where to split a bb, except for splitting before debug insn
11480 sequences followed by non-label real insn. Delete debug insns
11481 in between basic blocks.
11483 * combine.c (make_compound_operation_int): Formatting fixes.
11485 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
11486 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11487 * config/netbsd.h (LINK_EH_SPEC): Likewise.
11488 * config/sol2.h (LINK_EH_SPEC): Likewise.
11489 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11490 * config/s390/linux.h (LINK_SPEC): Likewise.
11491 * config/freebsd.h (LINK_EH_SPEC): Likewise.
11492 * config/openbsd.h (LINK_EH_SPEC): Likewise.
11493 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11494 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
11495 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
11496 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11497 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
11498 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
11500 2017-09-13 Jakub Jelinek <jakub@redhat.com>
11502 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
11504 (ENDFILE_LINUX_SPEC): Likewise.
11505 (LINK_EH_SPEC): Likewise.
11506 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
11507 (LINK_OS_LINUX_SPEC32): Likewise.
11508 (LINK_OS_LINUX_SPEC64): Likewise.
11509 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
11510 (LINK_OS_LINUX_SPEC): Likewise.
11512 2017-09-13 Martin Liska <mliska@suse.cz>
11514 PR middle-end/82154
11515 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
11516 CASE_HIGH is NULL_TREE.
11518 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11519 Alan Hayward <alan.hayward@arm.com>
11520 David Sherwood <david.sherwood@arm.com>
11522 * target.def (secondary_memory_needed): New hook.
11523 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
11524 instead of SECONDARY_MEMORY_NEEDED.
11525 (secondary_memory_needed_mode): Likewise.
11526 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
11527 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
11528 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
11529 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
11530 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
11531 * doc/tm.texi: Regenerate.
11532 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
11533 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
11534 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11535 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
11536 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
11537 * config/i386/i386.c (inline_secondary_memory_needed): Put the
11538 mode argument first and change the reg_class arguments to reg_class_t.
11539 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
11540 Make static. Update the call to inline_secondary_memory_needed.
11541 (ix86_register_move_cost): Update the call to
11542 inline_secondary_memory_needed.
11543 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11544 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
11546 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
11547 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11549 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
11550 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
11551 * config/mips/mips.c (mips_secondary_memory_needed): Make static
11552 and match hook interface. Add comment from mips.h.
11553 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11554 * config/mmix/mmix.md (truncdfsf2): Refer to
11555 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11557 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
11558 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
11559 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11560 (pa_secondary_memory_needed): New function.
11561 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
11562 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
11563 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11564 (pdp11_secondary_memory_needed): Make static and match hook interface.
11565 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
11566 * config/powerpcspe/powerpcspe-protos.h
11567 (rs6000_secondary_memory_needed_ptr): Delete.
11568 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
11570 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11571 (rs6000_option_override_internal): Assign to
11572 targetm.secondary_memory_needed rather than
11573 rs6000_secondary_memory_needed_ptr.
11574 (rs6000_secondary_memory_needed): Match hook interface.
11575 (rs6000_debug_secondary_memory_needed): Likewise.
11576 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
11577 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
11578 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
11579 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11580 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
11581 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
11583 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
11584 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11585 (rs6000_option_override_internal): Assign to
11586 targetm.secondary_memory_needed rather than
11587 rs6000_secondary_memory_needed_ptr.
11588 (rs6000_secondary_memory_needed): Match hook interface.
11589 (rs6000_debug_secondary_memory_needed): Likewise.
11590 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
11591 * config/s390/s390.c (s390_secondary_memory_needed): New function.
11592 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11593 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
11594 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
11595 (sparc_secondary_memory_needed): New function.
11596 * lra-constraints.c (check_and_process_move): Refer to
11597 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
11599 (curr_insn_transform): Likewise.
11600 (process_alt_operands): Use targetm.secondary_memory_needed
11601 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11602 (check_secondary_memory_needed_p): Likewise.
11603 (choose_split_class): Likewise.
11604 * reload.c: Unconditionally include code that was previously
11605 conditional on SECONDARY_MEMORY_NEEDED.
11606 (push_secondary_reload): Use targetm.secondary_memory_needed
11607 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11608 (push_reload): Likewise.
11609 * reload1.c: Unconditionally include code that was previously
11610 conditional on SECONDARY_MEMORY_NEEDED.
11611 (choose_reload_regs): Use targetm.secondary_memory_needed
11612 instead of TARGET_SECONDARY_MEMORY_NEEDED.
11613 (gen_reload): Likewise.
11614 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
11616 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
11617 Alan Hayward <alan.hayward@arm.com>
11618 David Sherwood <david.sherwood@arm.com>
11620 * target.def (secondary_memory_needed_mode): New hook:
11621 * targhooks.c (default_secondary_memory_needed_mode): Declare.
11622 * targhooks.h (default_secondary_memory_needed_mode): New function.
11623 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
11624 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
11625 * doc/tm.texi: Regenerate.
11626 * lra-constraints.c (check_and_process_move): Use
11627 targetm.secondary_memory_needed_mode instead of
11628 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
11629 (curr_insn_transform): Likewise.
11630 * reload.c (get_secondary_mem): Likewise.
11631 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11632 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
11634 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11635 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11636 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
11637 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11638 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
11640 * config/powerpcspe/powerpcspe-protos.h
11641 (rs6000_secondary_memory_needed_mode): Delete.
11642 * config/powerpcspe/powerpcspe.c
11643 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11644 (rs6000_secondary_memory_needed_mode): Make static.
11645 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11646 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
11648 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11650 (rs6000_secondary_memory_needed_mode): Make static.
11651 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11652 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
11653 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
11654 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
11655 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
11657 (sparc_secondary_memory_needed_mode): New function.
11658 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
11660 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
11662 * config/aarch64/constraints.md (Umq): New constraint.
11663 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
11665 (mov<mode>): Update condition.
11667 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11669 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
11670 when bitposition is the same.
11672 2017-09-13 Richard Biener <rguenther@suse.de>
11674 * dwarf2out.c (output_die_symbol): Remove.
11675 (output_die): Do not output a DIEs symbol.
11677 2017-09-13 Richard Biener <rguenther@suse.de>
11679 PR middle-end/82128
11680 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
11681 default-def to avoid breaking iterator update with the weird
11682 interaction with cgraph_update_edges_for_call_stmt_node.
11684 2017-09-13 Richard Biener <rguenther@suse.de>
11686 * tree-cfg.c (verify_gimple_assign_binary): Add verification
11687 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
11688 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
11689 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
11691 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
11693 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
11694 Disable pc relative literal load irrespective of
11695 TARGET_FIX_ERR_A53_84341 for default.
11697 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
11699 * config/sparc/sparc.c (output_return): Output the source location of
11700 the insn in the delay slot, if any.
11701 (output_sibcall): Likewise.
11703 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
11706 * common.opt (-static-pie): New alias.
11707 (shared): Negate static-pie.
11708 (-no-pie): Update help text.
11710 (static-pie): New option.
11711 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
11712 -static-pie support.
11713 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
11714 (LINK_EH_SPEC): Likewise.
11715 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
11716 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11717 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
11718 * gcc.c (LINK_COMMAND_SPEC): Likewise.
11719 (init_gcc_specs): Likewise.
11720 (init_spec): Likewise.
11721 (display_help): Update help message for -pie.
11722 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
11725 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
11727 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
11728 (movdi_aarch64): Likewise.
11729 (movti_aarch64): Likewise.
11731 2017-09-12 Simon Wright <simon@pushface.org>
11734 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
11735 calculation of the minor version, always output as 0.
11737 2017-09-12 Jakub Jelinek <jakub@redhat.com>
11740 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11741 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
11742 on it early, rather than manual conversion late. For
11743 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
11744 instead of performing manual conversion.
11746 2017-09-12 Carl Love <cel@us.ibm.com>
11748 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
11749 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
11751 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11752 VMULOSW): Add definitions.
11753 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11754 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11755 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11756 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11757 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11759 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11761 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
11763 (movti_aarch64): Likewise.
11764 (movdf_aarch64): Likewise.
11765 (movtf_aarch64): Likewise.
11766 (load_pairdi): Likewise.
11767 (store_pairdi): Likewise.
11768 (load_pairdf): Likewise.
11769 (store_pairdf): Likewise.
11770 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
11771 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
11772 (ldr_got_small_<mode>): Likewise.
11773 (ldr_got_small_28k_<mode>): Likewise.
11774 (ldr_got_tiny): Likewise.
11775 * config/aarch64/iterators.md (ldst_sz): New.
11776 (ldpstp_sz): Likewise.
11777 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
11779 (thunderx_load): Split load_8 to load_16.
11780 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
11782 (thunderx2t99_storepair_basic): Split store_8 to store_16.
11783 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
11784 (xgene1_store_pair): Split store_8 to store_16.
11785 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
11786 (falkor_st_0_st_sd): Split store_8 to store_16.
11788 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
11790 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
11791 and store1/2/3/4 to store_4/8/12/16.
11792 * config/aarch64/aarch64.md: Update for rename.
11793 * config/arm/arm.md: Likewise.: Likewise.
11794 * config/arm/arm.c: Likewise.
11795 * config/arm/thumb1.md: Likewise.
11796 * config/arm/thumb2.md: Likewise.
11797 * config/arm/vfp.md: Likewise.
11798 * config/arm/arm-generic.md: Likewise.
11799 * config/arm/arm1020e.md: Likewise.
11800 * config/arm/arm1026ejs.md: Likewise.
11801 * config/arm/arm1136jfs.md: Likewise.
11802 * config/arm/arm926ejs.md: Likewise.
11803 * config/arm/cortex-a15.md: Likewise.
11804 * config/arm/cortex-a17.md: Likewise.
11805 * config/arm/cortex-a5.md: Likewise.
11806 * config/arm/cortex-a53.md: Likewise.
11807 * config/arm/cortex-a57.md: Likewise.
11808 * config/arm/cortex-a7.md: Likewise.
11809 * config/arm/cortex-a8.md: Likewise.
11810 * config/arm/cortex-a9.md: Likewise.
11811 * config/arm/cortex-m4.md: Likewise.
11812 * config/arm/cortex-m7.md: Likewise.
11813 * config/arm/cortex-r4.md: Likewise.
11814 * config/arm/exynos-m1.md: Likewise.
11815 * config/arm/fa526.md: Likewise.
11816 * config/arm/fa606te.md: Likewise.
11817 * config/arm/fa626te.md: Likewise.
11818 * config/arm/fa726te.md: Likewise.
11819 * config/arm/fmp626.md: Likewise.
11820 * config/arm/iwmmxt.md: Likewise.
11821 * config/arm/ldmstm.md: Likewise.
11822 * config/arm/marvell-pj4.md: Likewise.
11823 * config/arm/xgene1.md: Likewise.
11824 * config/aarch64/thunderx.md: Likewise.
11825 * config/aarch64/thunderx2t99.md: Likewise.
11826 * config/aarch64/falkor.md: Likewise.
11828 2017-09-12 Martin Liska <mliska@suse.cz>
11830 * attribs.c (private_lookup_attribute): New function.
11831 * attribs.h (private_lookup_attribute): Declared here.
11832 (lookup_attribute): Called from this place.
11834 2017-09-12 Richard Biener <rguenther@suse.de>
11836 PR tree-optimization/82157
11837 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
11838 stmts with side-effects.
11840 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
11841 Alan Hayward <alan.hayward@arm.com>
11842 David Sherwood <david.sherwood@arm.com>
11844 * target.def (hard_regno_nregs): New hook.
11845 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
11846 * targhooks.h (default_hard_regno_nregs): Declare.
11847 * targhooks.c (default_hard_regno_nregs): New function.
11848 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
11849 (TARGET_HARD_REGNO_NREGS): ...this hook.
11850 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
11851 (CLASS_MAX_NREGS): Likewise.
11852 * doc/tm.texi: Regenerate.
11853 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
11854 instead of HARD_REGNO_NREGS.
11855 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
11856 HARD_REGNO_NREGS in the comment.
11857 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
11858 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
11859 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
11860 Return an unsigned int.
11861 (TARGET_HARD_REGNO_NREGS): Redefine.
11862 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
11863 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
11864 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11865 (arc_hard_regno_nregs): New function.
11866 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
11867 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
11868 (arm_hard_regno_nregs): New function.
11869 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
11870 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
11871 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
11872 (TARGET_HARD_REGNO_NREGS): Redefine.
11873 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
11874 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
11875 (HARD_REGNO_NREGS): Delete.
11876 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
11877 (cr16_hard_regno_nregs): New function.
11878 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
11879 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
11880 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
11881 (cris_hard_regno_nregs): New function.
11882 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
11883 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
11884 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11885 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
11886 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
11887 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
11888 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
11889 (frv_hard_regno_nregs): Make static. Take and return an
11891 (frv_class_max_nregs): Remove outdated copy of documentation.
11892 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
11893 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
11894 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
11895 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
11896 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
11897 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
11898 (TARGET_HARD_REGNO_NREGS): Redefine.
11899 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
11900 (CLASS_MAX_NREGS): Update comment.
11901 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
11902 (ia64_hard_regno_nregs): New function.
11903 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
11904 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
11905 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
11906 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
11907 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
11909 (m32c_hard_regno_nregs): Likewise. Make static.
11910 (TARGET_HARD_REGNO_NREGS): Redefine.
11911 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
11912 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
11913 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
11914 (m68k_hard_regno_nregs): New function.
11915 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
11916 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
11917 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
11918 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
11919 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
11920 Take and return an unsigned int.
11921 (TARGET_HARD_REGNO_NREGS): Redefine.
11922 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
11923 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
11924 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
11925 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
11926 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
11927 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
11928 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
11929 (msp430_hard_regno_nregs): Make static. Take and return an
11931 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
11932 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
11933 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
11934 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
11935 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
11936 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
11937 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
11938 (TARGET_HARD_REGNO_NREGS): Redefine.
11939 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
11940 (PA_HARD_REGNO_NREGS): ...this.
11941 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
11942 (PA_HARD_REGNO_NREGS): ...this.
11943 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11944 (pa_hard_regno_nregs): New function.
11945 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
11946 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
11947 (pdp11_hard_regno_nregs): New function.
11948 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
11949 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
11950 (rs6000_hard_regno_nregs_hook): New function.
11951 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
11952 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
11953 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
11954 Take and return an unsigned int. Move earlier in file.
11955 (TARGET_HARD_REGNO_NREGS): Redefine.
11956 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
11957 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
11958 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
11959 (rl78_hard_regno_nregs): Make static. Take and return an
11961 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
11962 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
11963 (rs6000_hard_regno_nregs_hook): New function.
11964 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
11965 * config/rx/rx.c (rx_hard_regno_nregs): New function.
11966 (TARGET_HARD_REGNO_NREGS): Redefine.
11967 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
11968 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
11969 instead of HARD_REGNO_NREGS.
11970 (s390_hard_regno_nregs): New function.
11971 (s390_hard_regno_mode_ok): Add comment from s390.h.
11972 (TARGET_HARD_REGNO_NREGS): Redefine.
11973 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
11974 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
11975 (sh_hard_regno_nregs): New function.
11976 (sh_pass_in_reg_p): Use it.
11977 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
11978 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
11979 (sparc_hard_regno_nregs): New function.
11980 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
11981 * config/spu/spu.c (spu_hard_regno_nregs): New function.
11982 (spu_function_arg_advance): Use it, supplying a valid register number.
11983 (TARGET_HARD_REGNO_NREGS): Redefine.
11984 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
11985 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
11986 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
11987 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
11988 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
11989 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
11990 (CLASS_MAX_NREGS): Remove copy of old documentation.
11991 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
11992 (visium_hard_regno_nregs): New function.
11993 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
11994 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
11995 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
11996 xtensa_hard_regno_nregs): New function.
11997 * system.h (HARD_REGNO_NREGS): Poison.
11999 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12001 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
12002 hard_regno_nregs instead of HARD_REGNO_NREGS.
12003 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
12004 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
12005 (c6x_expand_epilogue): Likewise.
12006 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
12007 (frv_read_iacc_argument): Likewise.
12008 * config/sh/sh.c: Include regs.h.
12009 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12010 (regs_used): Likewise.
12011 (output_stack_adjust): Likewise.
12012 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
12013 * expmed.c: Include regs.h.
12014 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
12015 * ree.c: Include regs.h.
12016 (combine_reaching_defs): Use hard_regno_nregs instead of
12018 (add_removable_extension): Likewise.
12020 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12022 * regs.h (hard_regno_nregs): Turn into a function.
12023 (end_hard_regno): Update accordingly.
12024 * caller-save.c (setup_save_areas): Likewise.
12025 (save_call_clobbered_regs): Likewise.
12026 (replace_reg_with_saved_mem): Likewise.
12027 (insert_restore): Likewise.
12028 (insert_save): Likewise.
12029 * combine.c (can_change_dest_mode): Likewise.
12030 (move_deaths): Likewise.
12031 (distribute_notes): Likewise.
12032 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
12033 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
12034 (rs6000_split_multireg_move): Likewise.
12035 (rs6000_register_move_cost): Likewise.
12036 (rs6000_memory_move_cost): Likewise.
12037 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
12038 (rs6000_split_multireg_move): Likewise.
12039 (rs6000_register_move_cost): Likewise.
12040 (rs6000_memory_move_cost): Likewise.
12041 * cselib.c (cselib_reset_table): Likewise.
12042 (cselib_lookup_1): Likewise.
12043 * emit-rtl.c (set_mode_and_regno): Likewise.
12044 * function.c (aggregate_value_p): Likewise.
12045 * ira-color.c (setup_profitable_hard_regs): Likewise.
12046 (check_hard_reg_p): Likewise.
12047 (calculate_saved_nregs): Likewise.
12048 (assign_hard_reg): Likewise.
12049 (improve_allocation): Likewise.
12050 (calculate_spill_cost): Likewise.
12051 * ira-emit.c (modify_move_list): Likewise.
12052 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
12053 (ira_hard_reg_in_set_p): Likewise.
12054 * ira.c (setup_reg_mode_hard_regset): Likewise.
12055 (clarify_prohibited_class_mode_regs): Likewise.
12056 (check_allocation): Likewise.
12057 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12058 (lra_setup_reg_renumber): Likewise.
12059 (setup_try_hard_regno_pseudos): Likewise.
12060 (spill_for): Likewise.
12061 (assign_hard_regno): Likewise.
12062 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
12063 * lra-constraints.c (in_class_p): Likewise.
12064 (lra_constraint_offset): Likewise.
12065 (simplify_operand_subreg): Likewise.
12066 (lra_constraints): Likewise.
12067 (split_reg): Likewise.
12068 (split_if_necessary): Likewise.
12069 (invariant_p): Likewise.
12070 (inherit_in_ebb): Likewise.
12071 * lra-lives.c (process_bb_lives): Likewise.
12072 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
12073 (get_hard_regs): Likewise.
12074 (do_remat): Likewise.
12075 * lra-spills.c (assign_spill_hard_regs): Likewise.
12076 * mode-switching.c (create_pre_exit): Likewise.
12077 * postreload.c (reload_combine_recognize_pattern): Likewise.
12078 * recog.c (peep2_find_free_register): Likewise.
12079 * regcprop.c (kill_value_regno): Likewise.
12080 (set_value_regno): Likewise.
12081 (copy_value): Likewise.
12082 (maybe_mode_change): Likewise.
12083 (find_oldest_value_reg): Likewise.
12084 (copyprop_hardreg_forward_1): Likewise.
12085 * regrename.c (check_new_reg_p): Likewise.
12086 (regrename_do_replace): Likewise.
12087 * reload.c (push_reload): Likewise.
12088 (combine_reloads): Likewise.
12089 (find_dummy_reload): Likewise.
12090 (operands_match_p): Likewise.
12091 (find_reloads): Likewise.
12092 (find_equiv_reg): Likewise.
12093 (reload_adjust_reg_for_mode): Likewise.
12094 * reload1.c (count_pseudo): Likewise.
12095 (count_spilled_pseudo): Likewise.
12096 (find_reg): Likewise.
12097 (clear_reload_reg_in_use): Likewise.
12098 (free_for_value_p): Likewise.
12099 (allocate_reload_reg): Likewise.
12100 (choose_reload_regs): Likewise.
12101 (reload_adjust_reg_for_temp): Likewise.
12102 (emit_reload_insns): Likewise.
12103 (delete_output_reload): Likewise.
12104 * rtlanal.c (subreg_get_info): Likewise.
12105 * sched-deps.c (sched_analyze_reg): Likewise.
12106 * sel-sched.c (init_regs_for_mode): Likewise.
12107 (mark_unavailable_hard_regs): Likewise.
12108 (choose_best_reg_1): Likewise.
12109 (verify_target_availability): Likewise.
12110 * valtrack.c (dead_debug_insert_temp): Likewise.
12111 * var-tracking.c (track_loc_p): Likewise.
12112 (emit_note_insn_var_location): Likewise.
12113 * varasm.c (make_decl_rtl): Likewise.
12114 * reginfo.c (choose_hard_reg_mode): Likewise.
12115 (init_reg_modes_target): Refer directly to
12116 this_target_regs->x_hard_regno_nregs.
12118 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12120 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
12121 instead of hard_regno_nregs.
12123 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12125 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
12126 end_hard_regno instead of hard_regno_nregs.
12127 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
12128 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
12129 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
12130 * ira-color.c (improve_allocation): Likewise.
12131 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12132 * lra-lives.c (mark_regno_live): Likewise.
12133 (mark_regno_dead): Likewise.
12134 * lra-remat.c (operand_to_remat): Likewise.
12135 * lra.c (collect_non_operand_hard_regs): Likewise.
12136 * postreload.c (reload_combine_note_store): Likewise.
12137 (move2add_valid_value_p): Likewise.
12138 * reload.c (regno_clobbered_p): Likewise.
12140 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12142 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
12144 * config/v850/v850.c (v850_reorg): Likewise.
12145 * reload.c (refers_to_regno_for_reload_p): Likewise.
12146 (find_equiv_reg): Likewise.
12147 * reload1.c (reload_reg_reaches_end_p): Likewise.
12149 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12151 * caller-save.c (add_used_regs): Use REG_NREGS instead of
12153 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
12154 * config/arm/arm.c (output_move_neon): Likewise.
12155 (arm_attr_length_move_neon): Likewise.
12156 (neon_split_vcombine): Likewise.
12157 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
12158 (c6x_mark_reg_written): Likewise.
12159 (c6x_dwarf_register_span): Likewise.
12160 * config/i386/i386.c (ix86_save_reg): Likewise.
12161 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
12162 (rws_access_reg): Likewise.
12163 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
12164 * mode-switching.c (create_pre_exit): Likewise.
12165 * ree.c (combine_reaching_defs): Likewise.
12166 (add_removable_extension): Likewise.
12167 * regcprop.c (find_oldest_value_reg): Likewise.
12168 (copyprop_hardreg_forward_1): Likewise.
12169 * reload.c (reload_inner_reg_of_subreg): Likewise.
12170 (push_reload): Likewise.
12171 (combine_reloads): Likewise.
12172 (find_dummy_reload): Likewise.
12173 (reload_adjust_reg_for_mode): Likewise.
12174 * reload1.c (find_reload_regs): Likewise.
12175 (forget_old_reloads_1): Likewise.
12176 (reload_reg_free_for_value_p): Likewise.
12177 (reload_adjust_reg_for_temp): Likewise.
12178 (emit_reload_insns): Likewise.
12179 (delete_output_reload): Likewise.
12180 * sel-sched.c (choose_best_reg_1): Likewise.
12181 (choose_best_pseudo_reg): Likewise.
12183 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12184 Alan Hayward <alan.hayward@arm.com>
12185 David Sherwood <david.sherwood@arm.com>
12187 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
12188 * target.def (slow_unaligned_access): New hook.
12189 * targhooks.h (default_slow_unaligned_access): Declare.
12190 * targhooks.c (default_slow_unaligned_access): New function.
12191 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
12192 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
12193 * doc/tm.texi: Regenerate.
12194 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
12195 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
12196 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
12198 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
12199 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
12201 (rs6000_slow_unaligned_access): New function.
12202 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12203 (expand_block_compare): Likewise.
12204 (expand_strn_compare): Likewise.
12205 (rs6000_rtx_costs): Likewise.
12206 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
12207 (riscv_slow_unaligned_access): Likewise.
12208 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
12209 (riscv_slow_unaligned_access_p): ...this and make static.
12210 (riscv_option_override): Update accordingly.
12211 (riscv_slow_unaligned_access): New function.
12212 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12213 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
12214 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
12215 (rs6000_slow_unaligned_access): New function.
12216 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
12217 (rs6000_rtx_costs): Likewise.
12218 * config/rs6000/rs6000-string.c (expand_block_compare)
12219 (expand_strn_compare): Use targetm.slow_unaligned_access instead
12220 of SLOW_UNALIGNED_ACCESS.
12221 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
12222 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
12223 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
12224 of SLOW_UNALIGNED_ACCESS.
12225 * expmed.c (simple_mem_bitfield_p): Likewise.
12226 * expr.c (alignment_for_piecewise_move): Likewise.
12227 (emit_group_load_1): Likewise.
12228 (emit_group_store): Likewise.
12229 (copy_blkmode_from_reg): Likewise.
12230 (emit_push_insn): Likewise.
12231 (expand_assignment): Likewise.
12232 (store_field): Likewise.
12233 (expand_expr_real_1): Likewise.
12234 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
12235 * lra-constraints.c (simplify_operand_subreg): Likewise.
12236 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12237 * gimple-ssa-store-merging.c: Likewise in block comment at start
12239 * tree-ssa-strlen.c: Include target.h.
12240 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
12241 of SLOW_UNALIGNED_ACCESS.
12242 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
12244 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
12246 PR rtl-optimization/82185
12247 * expmed.c (emit_store_flag_int): Only test tem if it has been
12250 2017-09-12 Richard Biener <rguenther@suse.de>
12252 PR middle-end/82149
12253 * match.pd ((FTYPE) N CMP CST): Fix typo.
12255 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
12257 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
12259 (mips_near_type_p): Add 'short_call' attribute as a synonym
12261 * doc/extend.texi (short_call): Document new function attribute.
12263 2017-09-12 Jakub Jelinek <jakub@redhat.com>
12266 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
12267 assertion check that in the condition.
12268 (get_atomic_generic_size): Likewise. Before testing if parameter
12269 has pointer type, if it has array type, call for C++
12270 default_conversion to perform array-to-pointer conversion.
12272 2017-09-12 Richard Biener <rguenther@suse.de>
12274 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
12275 for operations we cannot scalarize.
12277 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
12279 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
12280 vectors heap vectors. Clean up comments.
12281 Make visited_bbs a reference.
12282 (profitable_jump_thread_path): Make GC
12283 vectors heap vectors. Clean up comments.
12285 (convert_and_register_jump_thread_path): Make GC vectors heap
12287 (check_subpath_and_update_thread_path): Same. Clean up comments.
12288 Make visited_bbs a reference.
12289 (handle_phi): Abstract common code to to
12290 register_jump_thread_path_if_profitable.
12291 Rename VAR_BB to DEF_BB.
12293 Make GC vectors heap vectors.
12294 Make visited_bbs a reference.
12295 (handle_assignment): Same.
12296 (register_jump_thread_path_if_profitable): New.
12297 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
12299 Make GC vectors heap vectors. Clean up comments.
12300 Make visited_bbs a reference.
12301 (find_jump_threads_backwards): Make visited_bbs live in the stack.
12302 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
12305 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
12308 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
12309 words of E_DImode object are reachable by xtensa_uimm8x4 access.
12311 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
12313 Revert r251800 and r251799.
12315 2017-09-11 Martin Jambor <mjambor@suse.cz>
12318 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
12319 arguments in advance.
12320 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
12321 use it to find predecessor edges.
12322 (naive_outof_ssa): Collect vector of predecessors.
12324 2017-09-08 Jason Merrill <jason@redhat.com>
12326 PR c++/70029 - ICE with ref-qualifier and -flto
12327 * langhooks.h (struct lang_hooks_for_types): Add
12328 copy_lang_qualifiers.
12329 * attribs.c (build_type_attribute_qual_variant): Use it.
12330 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
12332 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
12333 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
12335 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
12338 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
12339 (*mulsi3_sp64): New instruction.
12340 (mulsi3): New expander.
12342 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
12344 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
12346 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12348 * sancov.c: Include memmodel.h.
12350 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
12353 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
12356 2017-09-07 Carl Love <cel@us.ibm.com>
12358 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
12359 the sldi instruction.
12361 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
12363 * sancov.c: Include tm_p.h.
12365 2017-09-07 Jakub Jelinek <jakub@redhat.com>
12368 * output.h (switch_to_other_text_partition): New declaration.
12369 * varasm.c (switch_to_other_text_partition): New function.
12370 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
12371 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
12372 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
12373 to the other text partition before emitting LCL label and switch back
12374 after emitting the word after it.
12376 2017-09-07 Richard Biener <rguenther@suse.de>
12378 * passes.def (pass_split_crit_edges): Remove instance before PRE.
12379 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
12380 critical edges here, after loop init.
12381 (pass_data_pre): Remove PROP_no_crit_edges flags.
12382 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
12383 for valueization of call args to avoid leaking VN_TOP.
12384 (visit_use): Assert we do not visit default defs.
12385 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
12386 Use error_mark_node to more easily detect leaking VN_TOP.
12387 All default-defs are varying, not VN_TOP. Mark them visited.
12388 (run_scc_vn): Make code match comment.
12390 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12392 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
12393 OPTION_MASK_FLOAT128_KEYWORD.
12394 (POWERPC_MASKS): Likewise.
12395 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
12396 support for the -mfloat128-type option, and make -mfloat128
12397 default on PowerPC Linux systems. Define or undefine
12398 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
12399 Define __float128 to be __ieee128 if IEEE 128-bit support is
12400 enabled, or undefine it.
12401 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
12402 Delete defining __FLOAT128_TYPE__.
12403 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
12404 -mfloat128-type option and make -mfloat128 default on PowerPC
12406 (TARGET_FLOAT128_TYPE): Likewise.
12407 (-mfloat128-type): Likewise.
12408 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12409 Delete the -mfloat128-type option and make -mfloat128 default on
12410 PowerPC Linux systems. Always use __ieee128 to be the keyword for
12411 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
12412 128-bit floating point is enabled. Change tests from using
12413 -mfloat128-type to -mfloat128.
12414 (rs6000_mangle_type): Use the correct mangling for the __float128
12415 type even if normal long double is restricted to 64-bits.
12416 (floatn_mode): Enable the _Float128 type by default on VSX Linux
12418 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
12419 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
12420 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
12422 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
12423 documentation for -mfloat128.
12425 2017-09-06 Olivier Hainque <hainque@adacore.com>
12427 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
12429 2017-09-06 Wish Wu <wishwu007@gmail.com>
12430 Jakub Jelinek <jakub@redhat.com>
12432 * asan.c (initialize_sanitizer_builtins): Add
12433 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
12434 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
12435 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
12436 BT_FN_VOID_UINT64_PTR variables.
12437 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
12438 (BT_FN_VOID_UINT16_UINT16): Likewise.
12439 (BT_FN_VOID_UINT32_UINT32): Likewise.
12440 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
12441 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
12442 (BT_FN_VOID_UINT64_PTR): Likewise.
12443 * common.opt (flag_sanitize_coverage): New variable.
12444 (fsanitize-coverage=trace-pc): Remove.
12445 (fsanitize-coverage=): Add.
12446 * flag-types.h (enum sanitize_coverage_code): New enum.
12447 * fold-const.c (fold_range_test): Disable non-short-circuit
12448 optimization if flag_sanitize_coverage.
12449 (fold_truth_andor): Likewise.
12450 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
12451 * opts.c (COVERAGE_SANITIZER_OPT): Define.
12452 (coverage_sanitizer_opts): New array.
12453 (get_closest_sanitizer_option): Add OPTS argument, handle also
12454 OPT_fsanitize_coverage_.
12455 (parse_sanitizer_options): Adjusted to also handle
12456 OPT_fsanitize_coverage_.
12457 (common_handle_option): Add OPT_fsanitize_coverage_.
12458 * sancov.c (instrument_comparison, instrument_switch): New function.
12459 (sancov_pass): Add trace-cmp support.
12460 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
12461 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
12462 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
12463 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
12464 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
12465 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
12466 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
12467 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
12468 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
12469 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
12471 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12473 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12474 error. Only quit immediately if parsing is complete.
12475 (BEGIN): Initialize fatal_err and parse_done.
12476 (begin fpu, end fpu): Check number of arguments.
12477 (begin arch, end arch): Likewise.
12478 (begin cpu, end cpu): Likewise.
12479 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12480 (optalias): Likewise.
12482 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
12484 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12485 * config/arm/arm-isa.h: Delete. Move definitions to ...
12486 * arm-cpus.in: ... here. Use new feature and fgroup values.
12487 * config/arm/arm.c (arm_option_override): Use lower case for feature
12489 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12490 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12491 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12493 (print_isa_bits_for): New function.
12494 (gen_isa): New function.
12495 (gen_comm_data): Use print_isa_bits_for.
12496 (define feature): New keyword.
12497 (define fgroup): New keyword.
12498 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
12499 (arm-isa.h): Add rule to generate file.
12500 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12501 case for feature bit names.
12503 2017-09-06 Richard Biener <rguenther@suse.de>
12505 * tree-ssa-pre.c (NECESSARY): Remove.
12506 (create_expression_by_pieces): Do not touch pass-local flags.
12507 (insert_into_preds_of_block): Likewise.
12508 (do_pre_regular_insertion): Likewise.
12509 (eliminate_insert): Likewise.
12510 (eliminate_dom_walker::before_dom_children): Likewise.
12511 (fini_eliminate): Do not look at inserted_exprs.
12512 (mark_operand_necessary): Remove.
12513 (remove_dead_inserted_code): Replace with simple work-list
12514 algorithm based on inserted_exprs and SSA uses.
12515 (pass_pre::execute): Re-order fini_eliminate and
12516 remove_dead_inserted_code.
12518 2017-09-06 Olivier Hainque <hainque@adacore.com>
12520 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
12521 for VxWorks 7. Adjust surrounding comments.
12523 2017-09-06 Richard Biener <rguenther@suse.de>
12525 * gimple-ssa-strength-reduction.c
12526 (find_candidates_dom_walker::before_dom_children): Also allow
12529 2017-09-06 Richard Biener <rguenther@suse.de>
12531 PR tree-optimization/82108
12532 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
12533 for gap in the non-permutation SLP case.
12535 2017-09-06 Martin Jambor <mjambor@suse.cz>
12537 PR tree-optimization/82078
12538 * tree-sra.c (sort_and_splice_var_accesses): Move call to
12539 add_access_to_work_queue...
12540 (build_accesses_from_assign): ...here.
12541 (propagate_all_subaccesses): Make sure racc is the group
12542 representative, if there is one.
12544 2017-09-06 Jakub Jelinek <jakub@redhat.com>
12546 PR middle-end/82095
12547 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
12550 2017-09-06 Richard Biener <rguenther@suse.de>
12552 * gimple-ssa-strength-reduction.c
12553 (find_candidates_dom_walker::before_doom_children): Use a
12554 type and not a mode check.
12556 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
12559 * config/arm/predicates.md (arm_general_adddi_operand): Create new
12561 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
12563 2017-09-05 Jeff Law <law@redhat.com>
12565 PR tree-optimization/64910
12566 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
12567 cases where we have 3 or more operands.
12569 2017-09-05 Jakub Jelinek <jakub@redhat.com>
12571 PR middle-end/81768
12572 * omp-low.c (lower_omp_for): Recompute tree invariant if
12573 gimple_omp_for_initial/final is ADDR_EXPR.
12575 PR middle-end/81768
12576 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
12577 into gimple val before gimplification fo the COND_EXPR.
12579 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
12581 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
12582 REGION_COPY argument.
12583 (thread_through_all_blocks): Remove unused argument to
12584 duplicate_thread_path.
12586 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12587 Alan Hayward <alan.hayward@arm.com>
12588 David Sherwood <david.sherwood@arm.com>
12590 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
12591 Take a scalar_mode rather than a machine_mode.
12592 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12593 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
12594 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12595 (aarch64_gen_adjusted_ldpstp): Likewise.
12596 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
12598 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12599 Alan Hayward <alan.hayward@arm.com>
12600 David Sherwood <david.sherwood@arm.com>
12602 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
12603 Take a scalar_int_mode instead of a machine_mode.
12604 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
12605 (aarch64_output_scalar_simd_mov_immediate): Likewise.
12606 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
12607 (aarch64_simd_attr_length_rglist): Delete.
12608 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
12609 a scalar_int_mode instead of a machine_mode.
12610 (aarch64_add_offset): Likewise.
12611 (aarch64_internal_mov_immediate): Likewise
12612 (aarch64_add_constant_internal): Likewise.
12613 (aarch64_add_constant): Likewise.
12614 (aarch64_movw_imm): Likewise.
12615 (aarch64_rtx_arith_op_extract_p): Likewise.
12616 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
12617 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
12618 Remove assert that the mode isn't a vector.
12619 (aarch64_output_scalar_simd_mov_immediate): Likewise.
12620 (aarch64_expand_mov_immediate): Update calls after above changes.
12621 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
12622 (aarch64_and_bitmask_imm): Check for scalar integer modes.
12623 (aarch64_move_imm): Likewise.
12624 (aarch64_can_const_movi_rtx_p): Likewise.
12625 (aarch64_strip_extend): Likewise.
12626 (aarch64_extr_rtx_p): Likewise.
12627 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
12628 a CONST_INT when the mode parameter is VOIDmode.
12629 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
12631 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12633 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
12634 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12635 (bitwise_type_for_mode): Update accordingly.
12637 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12639 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
12640 * stor-layout.c (mode_for_size_tree): Likewise.
12641 (mode_for_array): Update accordingly.
12642 (layout_decl): Likewise.
12643 (compute_record_mode): Likewise. Only set the mode once.
12645 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12647 * target.def (get_mask_mode): Change return type to opt_mode.
12649 * doc/tm.texi: Regenerate.
12650 * targhooks.h (default_get_mask_mode): Return an opt_mode.
12651 * targhooks.c (default_get_mask_mode): Likewise.
12652 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12653 * optabs-query.c (can_vec_mask_load_store_p): Update use of
12654 targetm.get_mask_mode.
12655 * tree.c (build_truth_vector_type): Likewise.
12657 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12659 * machmode.h (mode_for_vector): Return an opt_mode.
12660 * stor-layout.c (mode_for_vector): Likewise.
12661 (mode_for_int_vector): Update accordingly.
12662 (layout_type): Likewise.
12663 * config/i386/i386.c (emit_memmov): Likewise.
12664 (ix86_expand_set_or_movmem): Likewise.
12665 (ix86_expand_vector_init): Likewise.
12666 (ix86_get_mask_mode): Likewise.
12667 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
12669 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
12670 * expmed.c (extract_bit_field_1): Likewise.
12671 * expr.c (expand_expr_real_2): Likewise.
12672 * optabs-query.c (can_vec_perm_p): Likewise.
12673 (can_vec_mask_load_store_p): Likewise.
12674 * optabs.c (expand_vec_perm): Likewise.
12675 * targhooks.c (default_get_mask_mode): Likewise.
12676 * tree-vect-stmts.c (vectorizable_store): Likewise.
12677 (vectorizable_load): Likewise.
12678 (get_vectype_for_scalar_type_and_size): Likewise.
12680 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12682 * machmode.h (mode_for_int_vector): New function.
12683 * stor-layout.c (mode_for_int_vector): Likewise.
12684 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
12685 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
12686 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12687 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12688 (s390_expand_vcond): Likewise.
12690 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12692 * machmode.h (opt_machine_mode): New type.
12693 (opt_mode<T>): Allow construction from anything that can be
12695 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
12696 (mode_for_size): Return an opt_machine_mode.
12697 * stor-layout.c (mode_for_size): Likewise.
12698 (mode_for_size_tree): Update call accordingly.
12699 (bitwise_mode_for_mode): Likewise.
12700 (make_fract_type): Likewise.
12701 (make_accum_type): Likewise.
12702 * caller-save.c (replace_reg_with_saved_mem): Update call
12704 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12705 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12706 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12707 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
12708 * expmed.c (extract_bit_field_1): Likewise.
12709 * reload.c (get_secondary_mem): Likewise.
12710 * varasm.c (assemble_integer): Likewise.
12711 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
12714 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12716 * machmode.h (decimal_float_mode_for_size): New function.
12717 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
12718 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
12719 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
12720 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
12721 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
12722 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
12724 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12726 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
12727 (get_builtin_sync_mode): Likewise.
12728 (expand_ifn_atomic_compare_exchange): Likewise.
12729 (expand_builtin_atomic_clear): Likewise.
12730 (expand_builtin_atomic_test_and_set): Likewise.
12731 (fold_builtin_atomic_always_lock_free): Likewise.
12732 * calls.c (compute_argument_addresses): Likewise.
12733 (emit_library_call_value_1): Likewise.
12734 (store_one_arg): Likewise.
12735 * combine.c (combine_instructions): Likewise.
12736 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
12737 * config/arm/arm.c (arm_function_value): Likewise.
12738 (aapcs_allocate_return_reg): Likewise.
12739 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
12740 * config/i386/i386.c (construct_container): Likewise.
12741 (ix86_gimplify_va_arg): Likewise.
12742 (ix86_expand_sse_cmp): Likewise.
12743 (emit_memmov): Likewise.
12744 (emit_memset): Likewise.
12745 (expand_small_movmem_or_setmem): Likewise.
12746 (ix86_expand_pextr): Likewise.
12747 (ix86_expand_pinsr): Likewise.
12748 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
12749 * config/microblaze/microblaze.c (microblaze_block_move_straight):
12751 * config/mips/mips.c (mips_function_value_1) Likewise.
12752 (mips_block_move_straight): Likewise.
12753 (mips_expand_ins_as_unaligned_store): Likewise.
12754 * config/powerpcspe/powerpcspe.c
12755 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12756 (rs6000_darwin64_record_arg_flush): Likewise.
12757 * config/rs6000/rs6000.c
12758 (rs6000_darwin64_record_arg_advance_flush): Likewise.
12759 (rs6000_darwin64_record_arg_flush): Likewise.
12760 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
12761 (sparc_function_value_1): Likewise.
12762 * config/spu/spu.c (adjust_operand): Likewise.
12763 (spu_emit_branch_or_set): Likewise.
12764 (arith_immediate_p): Likewise.
12765 * emit-rtl.c (gen_lowpart_common): Likewise.
12766 * expr.c (expand_expr_real_1): Likewise.
12767 * function.c (assign_parm_setup_block): Likewise.
12768 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
12769 * reload1.c (alter_reg): Likewise.
12770 * stor-layout.c (mode_for_vector): Likewise.
12771 (layout_type): Likewise.
12773 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
12775 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
12776 (spu_convert_move): Likewise.
12777 * lower-subreg.c (resolve_simple_move): Likewise.
12779 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12782 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
12783 define_insn to a define_expand.
12784 (altivec_vsum2sws_direct): New define_insn.
12785 (altivec_vsumsws): Convert from a define_insn to a define_expand.
12787 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12789 * config/arm/arm.c (arm_option_params_internal): Improve setting of
12792 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
12797 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
12798 realign stack if stack alignment needed is less than incoming
12801 2017-09-05 Marek Polacek <polacek@redhat.com>
12804 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
12807 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
12809 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
12811 2017-09-05 Richard Biener <rguenther@suse.de>
12813 PR tree-optimization/82084
12814 * fold-const.c (can_native_encode_string_p): Handle wide characters.
12816 2017-09-05 Richard Biener <rguenther@suse.de>
12818 PR tree-optimization/82102
12819 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
12821 2017-09-05 Martin Liska <mliska@suse.cz>
12823 PR tree-optimization/82032
12824 * tree-cfg.c (generate_range_test): New function.
12825 * tree-cfg.h (generate_range_test): Declared here.
12826 * tree-cfgcleanup.c (convert_single_case_switch): New function.
12827 (cleanup_control_expr_graph): Use it.
12828 * tree-switch-conversion.c (try_switch_expansion): Remove
12830 (emit_case_nodes): Use generate_range_test.
12832 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12835 * config/i386/i386.md (*<btsc><mode>_mask): Add
12836 TARGET_USE_BT to insn constraint.
12837 (*btr<mode>_mask): Ditto.
12839 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
12841 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
12842 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
12844 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
12847 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
12848 TARGET_NEON and TARGET_IWMMXT.
12849 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
12850 TARGET_NEON and TARGET_IWMMXT.
12851 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
12853 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
12855 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
12856 (ix86_rewrite_tls_address): Ditto.
12857 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
12858 (ix86_rewrite_tls_address_1): Ditto.
12859 (ix86_rewrite_tls_address): Ditto.
12860 * config/i386/predicates.md (tls_address_pattern): New predicate.
12861 * config/i386/i386.md (TLS address splitter): New splitter.
12863 2017-09-04 Richard Biener <rguenther@suse.de>
12865 PR tree-optimization/82084
12866 * fold-const.h (can_native_encode_string_p): Declare.
12867 * fold-const.c (can_native_encode_string_p): Factor out from ...
12868 (native_encode_string): ... here.
12869 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
12870 vectorizing stores from constants we later cannot handle.
12872 2017-09-04 Marek Polacek <polacek@redhat.com>
12875 * doc/invoke.texi: Update -Wtautological-compare documentation.
12877 2017-09-04 Jeff Law <law@redhat.com>
12879 PR tree-optimization/64910
12880 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
12881 swap the first and last operand if the last is a constant.
12883 2017-09-04 Marek Polacek <polacek@redhat.com>
12886 * convert.c (do_narrow): When sanitizing signed integer overflows,
12887 bail out for signed types.
12888 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
12890 2017-09-04 Richard Biener <rguenther@suse.de>
12892 PR tree-optimization/82060
12893 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
12894 Move devirtualization after stmt folding and before EH/AB/noreturn
12895 cleanup to get the stmt refs canonicalized. Use a bool instead
12896 of gimple_modified_p since that doesn't work for NOPs. Schedule
12897 NOPs generated by folding for removal.
12899 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
12900 Alan Hayward <alan.hayward@arm.com>
12901 David Sherwood <david.sherwood@arm.com>
12903 * coretypes.h (pad_direction): New enum.
12904 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
12905 (FUNCTION_ARG_PADDING): Likewise.
12906 * target.def (function_arg_padding): New hook.
12907 * targhooks.h (default_function_arg_padding): Declare.
12908 * targhooks.c (default_function_arg_padding): New function.
12909 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
12910 (TARGET_FUNCTION_ARG_PADDING): ...this.
12911 * doc/tm.texi: Regenerate.
12912 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
12913 instead of direction.
12914 (compute_argument_addresses): Likewise.
12915 (load_register_parameters): Likewise.
12916 (emit_library_call_value_1): Likewise.
12917 (store_one_arg): Use targetm.calls.function_arg_padding instead
12918 of FUNCTION_ARG_PADDING.
12919 (must_pass_in_stack_var_size_or_pad): Likewise.
12920 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
12921 (emit_group_store): Likewise.
12922 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
12923 instead of FUNCTION_ARG_PADDING.
12924 (emit_push_insn): Likewise, and propagate enum change throughout
12926 * function.h (direction): Delete.
12927 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
12929 * function.c (assign_parm_find_stack_rtl): Likewise.
12930 (assign_parm_setup_block_p): Likewise.
12931 (assign_parm_setup_block): Likewise.
12932 (gimplify_parameters): Likewise.
12933 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
12934 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
12936 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
12937 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12938 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
12939 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
12940 (aarch64_function_arg_padding): ...this new function.
12941 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
12942 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12943 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
12944 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12945 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
12946 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12947 (arm_pad_arg_upward): Replace with...
12948 (arm_function_arg_padding): ...this new function.
12949 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
12951 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
12952 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
12953 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12954 (ia64_hpux_function_arg_padding): Replace with...
12955 (ia64_function_arg_padding): ...this new function. Use pad_direction
12956 instead of direction. Check for TARGET_HPUX.
12957 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
12958 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12959 (iq2000_function_arg_padding): New function.
12960 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
12961 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
12962 (mips_function_arg_padding): ...this new function.
12963 (mips_pad_reg_upward): Update accordingly.
12964 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12965 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
12966 targetm.calls.function_arg_padding.
12967 (FUNCTION_ARG_PADDING): Delete.
12968 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
12969 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
12970 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12971 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
12972 (nios2_block_reg_padding): Return pad_direction instead of direction.
12973 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
12974 instead of direction.
12975 (nios2_function_arg_padding): Likewise. Make static.
12976 (TARGET_FUNCTION_ARG_PADDING): Redefine.
12977 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
12978 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
12979 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
12980 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
12981 (pa_function_arg_padding): Make static. Return pad_direction instead
12983 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
12984 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12985 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
12986 instead of direction. Use targetm.calls.function_arg_padding.
12987 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
12988 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
12989 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
12990 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
12991 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
12993 (function_arg_padding): Rename to...
12994 (rs6000_function_arg_padding): ...this. Make static. Return
12995 pad_direction instead of direction.
12996 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
12997 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
12998 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
12999 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
13000 instead of direction. Use targetm.calls.function_arg_padding.
13001 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
13002 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
13003 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
13004 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
13005 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13006 (function_arg_padding): Rename to...
13007 (rs6000_function_arg_padding): ...this. Make static. Return
13008 pad_direction instead of direction.
13009 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
13010 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
13011 * config/s390/s390.c (s390_function_arg_padding): New function.
13012 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13013 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
13014 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
13015 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
13016 (function_arg_padding): Rename to...
13017 (sparc_function_arg_padding): ...this. Make static. Return
13018 pad_direction instead of direction.
13019 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
13020 * config/spu/spu.c (spu_function_arg_padding): New function.
13021 (TARGET_FUNCTION_ARG_PADDING): Redefine.
13022 * system.h (FUNCTION_ARG_PADDING): Poison.
13024 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13025 Alan Hayward <alan.hayward@arm.com>
13026 David Sherwood <david.sherwood@arm.com>
13028 * target.def (modes_tieable_p): New hook.
13029 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
13030 (TARGET_MODES_TIEABLE_P): ...this.
13031 * doc/tm.texi.in: Regenerate.
13032 * hooks.h (hook_bool_mode_mode_true): Declare.
13033 * hooks.c (hook_bool_mode_mode_true): New function.
13034 * combine.c (subst): Use targetm.modes_tieable_p instead of
13036 * dse.c (find_shift_sequence): Likewise.
13037 * expmed.c (extract_low_bits): Likewise.
13038 * lower-subreg.c: Include target.h.
13039 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
13041 * rtlanal.c (rtx_cost): Likewise.
13042 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
13043 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
13044 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
13045 (TARGET_MODES_TIEABLE_P): Redefine.
13046 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
13047 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
13048 (TARGET_MODES_TIEABLE_P): Redefine.
13049 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
13050 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
13051 (arc_modes_tieable_p): New function.
13052 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
13053 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
13054 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
13055 (arm_modes_tieable_p): Make static.
13056 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
13057 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
13058 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
13059 (TARGET_MODES_TIEABLE_P): Redefine.
13060 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
13061 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
13062 (TARGET_MODES_TIEABLE_P): Redefine.
13063 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
13064 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
13065 (cr16_modes_tieable_p): New function.
13066 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
13067 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
13068 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
13069 (TRULY_NOOP_TRUNCATION): Update comment.
13070 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
13071 (TRULY_NOOP_TRUNCATION): Update comment.
13072 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
13073 (frv_modes_tieable_p): New function.
13074 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
13075 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
13076 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
13077 (TARGET_MODES_TIEABLE_P): Redefine.
13078 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
13079 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
13080 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
13081 (TARGET_MODES_TIEABLE_P): Redefine.
13082 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
13083 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
13084 (ia64_modes_tieable_p): New function.
13085 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
13086 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
13087 (iq2000_modes_tieable_p): New function.
13088 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
13089 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
13090 (lm32_modes_tieable_p): New function.
13091 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
13092 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
13093 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
13094 (TARGET_MODES_TIEABLE_P): Redefine.
13095 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
13096 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
13097 (m32r_modes_tieable_p): New function.
13098 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
13099 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
13100 (m68k_modes_tieable_p): New function.
13101 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
13102 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
13103 (mcore_modes_tieable_p): New function.
13104 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
13105 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
13107 (TARGET_MODES_TIEABLE_P): Redefine.
13108 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
13109 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
13110 * config/mips/mips.c (mips_modes_tieable_p): Make static.
13111 (TARGET_MODES_TIEABLE_P): Redefine.
13112 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
13113 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
13114 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
13115 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
13116 (mn10300_modes_tieable_p): ...this and make static.
13117 (TARGET_MODES_TIEABLE_P): Redefine.
13118 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
13119 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
13120 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
13121 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
13122 (msp430_modes_tieable_p): Make static.
13123 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
13124 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
13125 (TARGET_MODES_TIEABLE_P): Redefine.
13126 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
13127 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
13128 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
13129 (TARGET_MODES_TIEABLE_P): Redefine.
13130 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
13131 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
13132 * config/pa/pa.c (pa_modes_tieable_p): Make static.
13133 (TARGET_MODES_TIEABLE_P): Redefine.
13134 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
13135 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
13136 (pdp11_modes_tieable_p): New function.
13137 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
13138 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
13139 (rs6000_modes_tieable_p): New function.
13140 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13141 * config/powerpcspe/powerpcspe.md: Update comment.
13142 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
13143 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
13144 (TARGET_MODES_TIEABLE_P): Redefine.
13145 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
13146 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
13147 (rl78_modes_tieable_p): New function.
13148 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
13149 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
13150 (rs6000_modes_tieable_p): New function.
13151 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
13152 * config/rs6000/rs6000.md: Update comment.
13153 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
13154 * config/rx/rx.c (rx_modes_tieable_p): New function.
13155 (TARGET_MODES_TIEABLE_P): Redefine.
13156 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
13157 * config/s390/s390.c (s390_modes_tieable_p): New function.
13158 (TARGET_MODES_TIEABLE_P): Redefine.
13159 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
13160 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
13161 (sh_modes_tieable_p): New function.
13162 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
13163 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
13164 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
13165 (sparc_modes_tieable_p): Make static.
13166 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
13167 * config/spu/spu.c (spu_modes_tieable_p): New function.
13168 (TARGET_MODES_TIEABLE_P): Redefine.
13169 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
13170 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
13171 (TARGET_MODES_TIEABLE_P): Redefine.
13172 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
13173 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
13174 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
13175 * config/v850/v850.c (v850_modes_tieable_p): New function.
13176 (TARGET_MODES_TIEABLE_P): Redefine.
13177 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
13178 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
13179 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
13180 (visium_modes_tieable_p): New function.
13181 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
13182 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
13183 (xtensa_modes_tieable_p): New function.
13184 * system.h (MODES_TIEABLE_P): Poison.
13186 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13187 Alan Hayward <alan.hayward@arm.com>
13188 David Sherwood <david.sherwood@arm.com>
13190 * target.def (hard_regno_mode_ok): New hook.
13191 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
13192 (TARGET_HARD_REGNO_MODE_OK): ...this.
13193 * doc/tm.texi.in: Regenerate.
13194 * hooks.h (hook_bool_uint_mode_true): Declare.
13195 * hooks.c (hook_bool_uint_mode_true): New function.
13196 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
13197 HARD_REGNO_MODE_OK.
13198 * genpreds.c (write_insn_preds_c): Add an include of target.h.
13199 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
13200 instead of HARD_REGNO_MODE_OK.
13201 * caller-save.c: Include target.h.
13202 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
13203 HARD_REGNO_MODE_OK.
13204 * combine.c (can_combine_p): Likewise.
13205 (combinable_i3pat): Likewise.
13206 (can_change_dest_mode): Likewise.
13207 * expr.c (init_expr_target): Likewise.
13208 (convert_move): Likewise.
13209 (convert_modes): Likewise.
13210 * ira.c (setup_prohibited_class_mode_regs): Likewise.
13211 (setup_prohibited_mode_move_regs): Likewise.
13212 * ira.h (target_ira): Likewise.
13213 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13214 * lra-constraints.c (process_alt_operands): Likewise.
13215 (split_reg): Likewise.
13216 * recog.c (peep2_find_free_register): Likewise.
13217 * ree.c (combine_reaching_defs): Likewise.
13218 * regcprop.c (maybe_mode_change): Likewise.
13219 * reginfo.c (init_reg_sets_1): Likewise.
13220 (choose_hard_reg_mode): Likewise.
13221 (simplifiable_subregs): Likewise.
13222 * regrename.c (check_new_reg_p): Likewise.
13223 * reload.c (find_valid_class): Likewise.
13224 (find_valid_class_1): Likewise.
13225 (reload_inner_reg_of_subreg): Likewise.
13226 (push_reload): Likewise.
13227 (combine_reloads): Likewise.
13228 (find_dummy_reload): Likewise.
13229 (find_reloads): Likewise.
13230 * reload1.c (find_reg): Likewise.
13231 (set_reload_reg): Likewise.
13232 (allocate_reload_reg): Likewise.
13233 (choose_reload_regs): Likewise.
13234 (reload_adjust_reg_for_temp): Likewise.
13235 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
13236 (simplify_subreg_regno): Likewise.
13237 * sel-sched.c (init_regs_for_mode): Likewise.
13238 * varasm.c (make_decl_rtl): Likewise.
13239 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
13240 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
13241 HARD_REGNO_MODE_OK.
13242 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
13243 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
13244 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13245 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
13246 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
13247 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13248 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
13249 (arc_mode_class): Delete.
13250 (HARD_REGNO_MODE_OK): Delete.
13251 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13252 (arc_hard_regno_mode_ok): Rename old array to...
13253 (arc_hard_regno_mode_ok_modes): ...this.
13254 (arc_conditional_register_usage): Update accordingly.
13255 (arc_mode_class): Make static.
13256 (arc_hard_regno_mode_ok): New function.
13257 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
13258 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
13259 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13260 (arm_hard_regno_mode_ok): Make static.
13261 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
13262 HARD_REGNO_MODE_OK.
13263 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
13264 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
13265 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
13267 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13268 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
13269 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
13270 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
13271 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
13272 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13273 * config/bfin/predicates.md (valid_reg_operand): Use
13274 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
13275 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
13276 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
13277 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13278 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
13279 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
13280 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13281 (cr16_hard_regno_mode_ok): Make static and return a bool.
13282 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
13283 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13284 (cris_hard_regno_mode_ok): New function.
13285 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
13286 (epiphany_mode_class): Delete.
13287 (HARD_REGNO_MODE_OK): Delete.
13288 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
13289 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13290 (hard_regno_mode_ok): Rename to...
13291 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
13292 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
13293 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
13294 HARD_REGNO_MODE_OK.
13295 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
13296 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
13297 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13298 (frv_hard_regno_mode_ok): Make static and return a bool.
13299 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
13300 HARD_REGNO_MODE_OK.
13301 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
13302 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
13303 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
13304 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
13306 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13307 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
13308 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
13309 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
13311 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13312 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
13313 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13314 (ia64_hard_regno_mode_ok): New function.
13315 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
13316 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13317 (iq2000_hard_regno_mode_ok): New function.
13318 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
13319 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13320 (lm32_hard_regno_mode_ok): New function.
13321 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
13322 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
13323 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
13324 instead of HARD_REGNO_MODE_OK.
13325 (m32c_hard_regno_ok): Rename to...
13326 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
13327 (m32c_cannot_change_mode_class): Update accordingly.
13328 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13329 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
13330 (m32r_mode_class): Delete.
13331 (HARD_REGNO_MODE_OK): Delete.
13332 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13333 (m32r_hard_regno_mode_ok): Rename to...
13334 (m32r_hard_regno_modes): ...this.
13335 (m32r_mode_class): Make static.
13336 (m32r_hard_regno_mode_ok): New function.
13337 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
13338 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
13339 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13340 (m68k_hard_regno_mode_ok): Make static.
13341 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
13342 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13343 (mcore_hard_regno_mode_ok): New function.
13344 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
13345 (HARD_REGNO_MODE_OK): Delete.
13346 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
13348 (microblaze_hard_regno_mode_ok_p): ...this and make static.
13349 (microblaze_hard_regno_mode_ok): New function.
13350 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13351 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
13352 (mips_hard_regno_mode_ok): Delete.
13353 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
13354 (mips_hard_regno_mode_ok_p): ...this and make static.
13355 (mips_hard_regno_mode_ok_p): Rename to...
13356 (mips_hard_regno_mode_ok_uncached): ...this.
13357 (mips_hard_regno_mode_ok): New function.
13358 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
13359 of HARD_REGNO_MODE_OK.
13360 (mips_option_override): Update after above name changes.
13361 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13362 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
13363 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
13364 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
13365 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
13366 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13367 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
13368 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
13369 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
13370 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13371 (msp430_hard_regno_mode_ok): Make static and return a bool.
13372 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
13373 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
13374 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
13376 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13377 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
13378 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
13379 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
13380 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
13381 (PA_HARD_REGNO_MODE_OK): ...this
13382 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
13383 (PA_HARD_REGNO_MODE_OK): ...this.
13384 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13385 (pa_hard_regno_mode_ok): New function.
13386 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
13387 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13388 (pdp11_hard_regno_mode_ok): New function.
13389 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
13390 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
13392 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
13394 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13395 (rs6000_hard_regno_mode_ok): Rename to...
13396 (rs6000_hard_regno_mode_ok_uncached): ...this.
13397 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13398 (rs6000_hard_regno_mode_ok): New function.
13399 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
13400 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
13401 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
13402 (riscv_hard_regno_mode_ok): ...this and make static.
13403 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13404 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
13405 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
13406 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13407 (rl78_hard_regno_mode_ok): Make static and return bool.
13408 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
13409 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
13411 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
13412 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13413 (rs6000_hard_regno_mode_ok): Rename to...
13414 (rs6000_hard_regno_mode_ok_uncached): ...this.
13415 (rs6000_init_hard_regno_mode_ok): Update accordingly.
13416 (rs6000_hard_regno_mode_ok): New function.
13417 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
13418 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
13419 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13420 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
13421 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
13422 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
13423 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13424 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
13425 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
13426 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13427 (sh_hard_regno_mode_ok): Make static.
13428 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
13429 instead of HARD_REGNO_MODE_OK.
13430 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
13431 (sparc_mode_class): Delete.
13432 (HARD_REGNO_MODE_OK): Delete.
13433 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13434 (hard_regno_mode_classes): Make static.
13435 (sparc_mode_class): Likewise.
13436 (sparc_hard_regno_mode_ok): New function.
13437 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
13438 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
13439 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
13441 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13442 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
13443 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
13444 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
13445 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
13446 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13447 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
13448 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
13449 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
13450 (visium_hard_regno_mode_ok): New function.
13451 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
13452 instead of HARD_REGNO_MODE_OK.
13453 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
13454 (HARD_REGNO_MODE_OK): Delete.
13455 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
13456 (xtensa_hard_regno_mode_ok_p): ...this and make static.
13457 (xtensa_option_override): Update accordingly.
13458 (TARGET_HARD_REGNO_MODE_OK): Redefine.
13459 (xtensa_hard_regno_mode_ok): New function.
13460 * system.h (HARD_REGNO_MODE_OK): Poison.
13462 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13463 Alan Hayward <alan.hayward@arm.com>
13464 David Sherwood <david.sherwood@arm.com>
13466 * target.def (hard_regno_call_part_clobbered): New hook.
13467 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
13468 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
13469 * doc/tm.texi: Regenerate.
13470 * hooks.h (hook_bool_uint_mode_false): Declare.
13471 * hooks.c (hook_bool_uint_mode_false): New function.
13472 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13473 * cselib.c (cselib_process_insn): Use
13474 targetm.hard_regno_call_part_clobbered instead of
13475 HARD_REGNO_CALL_PART_CLOBBERED.
13476 * ira-conflicts.c (ira_build_conflicts): Likewise.
13477 * ira-costs.c (ira_tune_allocno_costs): Likewise.
13478 * lra-constraints.c (need_for_call_save_p): Likewise.
13479 * lra-lives.c: Include target.h.
13480 (check_pseudos_live_through_calls): Use
13481 targetm.hard_regno_call_part_clobbered instead of
13482 HARD_REGNO_CALL_PART_CLOBBERED.
13483 * regcprop.c: Include target.h.
13484 (copyprop_hardreg_forward_1): Use
13485 targetm.hard_regno_call_part_clobbered instead of
13486 HARD_REGNO_CALL_PART_CLOBBERED.
13487 * reginfo.c (choose_hard_reg_mode): Likewise.
13488 * regrename.c (check_new_reg_p): Likewise.
13489 * reload.c (find_equiv_reg): Likewise.
13490 * reload1.c (emit_reload_insns): Likewise.
13491 * sched-deps.c (deps_analyze_insn): Likewise.
13492 * sel-sched.c (init_regs_for_mode): Likewise.
13493 (mark_unavailable_hard_regs): Likewise.
13494 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
13495 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13496 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
13498 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13499 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13500 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
13502 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
13504 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13505 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13506 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
13508 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13509 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13510 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
13512 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13513 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
13515 * config/powerpcspe/powerpcspe.c
13516 (rs6000_hard_regno_call_part_clobbered): New function.
13517 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13518 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13519 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
13521 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13522 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13523 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
13525 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
13526 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
13527 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
13529 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13530 Alan Hayward <alan.hayward@arm.com>
13531 David Sherwood <david.sherwood@arm.com>
13533 * rtl.h (subreg_memory_offset): Declare.
13534 * emit-rtl.c (subreg_memory_offset): New function.
13535 * expmed.c (store_bit_field_1): Use it.
13536 * expr.c (undefined_operand_subword_p): Likewise.
13537 * simplify-rtx.c (simplify_subreg): Likewise.
13539 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
13541 PR rtl-optimization/57448
13544 * optabs.c (expand_atomic_load): Place compiler memory barriers if
13545 using atomic_load pattern.
13546 (expand_atomic_store): Likewise.
13548 2017-09-04 Jakub Jelinek <jakub@redhat.com>
13551 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
13552 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
13553 handling. Use replace_call_with_value with NULL instead of
13554 gsi_replace, unlink_stmt_vdef and release_defs.
13556 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
13559 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
13561 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
13564 * rtl.h (emit_library_call_value_1): Declare.
13565 (emit_library_call): Replace declaration with a series of overloads.
13566 Remove the parameter count argument.
13567 (emit_library_call_value): Likewise.
13568 * calls.c (emit_library_call_value_1): Make global. Replace varargs
13569 with an "rtx_mode_t *".
13570 (emit_library_call_value): Delete.
13571 (emit_library_call): Likewise.
13572 * asan.c (asan_emit_stack_protection): Update calls accordingly.
13573 (asan_emit_allocas_unpoison): Likewise.
13574 * builtins.c (expand_builtin_powi): Likewise.
13575 (expand_asan_emit_allocas_unpoison): Likewise.
13576 * cfgexpand.c (expand_main_function): Likewise.
13577 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
13578 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
13579 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
13580 * config/arm/arm.c (arm_trampoline_init): Likewise.
13581 (arm_call_tls_get_addr): Likewise.
13582 (arm_expand_divmod_libfunc): Likewise.
13583 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
13584 (smulsi3_highpart): Likewise.
13585 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
13586 (c6x_expand_compare): Likewise.
13587 (c6x_expand_movmem): Likewise.
13588 * config/frv/frv.c (frv_trampoline_init): Likewise.
13589 * config/i386/i386.c (ix86_trampoline_init): Likewise.
13590 (ix86_expand_divmod_libfunc): Likewise.
13591 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
13592 (ia64_expand_compare): Likewise.
13593 (ia64_profile_hook): Likewise.
13594 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
13595 (nonlocal_goto): Likewise.
13596 (restore_stack_nonlocal): Likewise.
13597 * config/m32r/m32r.c (block_move_call): Likewise.
13598 (m32r_trampoline_init): Likewise.
13599 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
13600 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
13601 (m68k_call_m68k_read_tp): Likewise.
13602 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
13603 (microblaze_expand_divide): Likewise.
13604 * config/mips/mips.h (mips_args): Likewise.
13605 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
13606 (MIPS_ICACHE_SYNC): Likewise.
13607 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
13608 (nios2_trampoline_init): Likewise.
13609 * config/pa/pa.c (hppa_tls_call): Likewise.
13610 (pa_trampoline_init): Likewise.
13611 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
13612 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
13613 (expand_strn_compare): Likewise.
13614 (rs6000_generate_compare): Likewise.
13615 (rs6000_expand_float128_convert): Likewise.
13616 (output_profile_hook): Likewise.
13617 (rs6000_trampoline_init): Likewise.
13618 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
13619 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
13620 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
13621 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
13622 (rs6000_generate_compare): Likewise.
13623 (rs6000_expand_float128_convert): Likewise.
13624 (output_profile_hook): Likewise.
13625 (rs6000_trampoline_init): Likewise.
13626 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
13627 * config/sh/sh.c (sh_trampoline_init): Likewise.
13628 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
13629 (sparc_emit_float_lib_cmp): Likewise.
13630 (sparc32_initialize_trampoline): Likewise.
13631 (sparc64_initialize_trampoline): Likewise.
13632 (sparc_profile_hook): Likewise.
13633 * config/spu/spu.c (ea_load_store): Likewise.
13634 * config/spu/spu.md (floatunssidf2): Likewise.
13635 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
13636 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
13637 * config/visium/visium.c (expand_block_move_4): Likewise.
13638 (expand_block_move_2): Likewise.
13639 (expand_block_move_1): Likewise.
13640 (expand_block_set_4): Likewise.
13641 (expand_block_set_2): Likewise.
13642 (expand_block_set_1): Likewise.
13643 (visium_trampoline_init): Likewise.
13644 (visium_profile_hook): Likewise.
13645 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
13646 (xtensa_setup_frame_addresses): Likewise.
13647 (xtensa_trampoline_init): Likewise.
13648 * except.c (sjlj_emit_function_enter): Likewise.
13649 (sjlj_emit_function_exit): Likewise.
13650 * explow.c (allocate_dynamic_stack_space): Likewise.
13651 (probe_stack_range): Likewise.
13652 * expr.c (convert_mode_scalar): Likewise.
13653 * optabs.c (expand_binop): Likewise.
13654 (expand_twoval_binop_libfunc): Likewise.
13655 (expand_unop): Likewise.
13656 (prepare_cmp_insn): Likewise.
13657 (prepare_float_lib_cmp): Likewise.
13658 (expand_float): Likewise.
13659 (expand_fix): Likewise.
13660 (expand_fixed_convert): Likewise.
13661 (maybe_emit_sync_lock_test_and_set): Likewise.
13662 (expand_atomic_compare_and_swap): Likewise.
13663 (expand_mem_thread_fence): Likewise.
13664 (expand_atomic_fetch_op): Likewise.
13666 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
13668 * doc/generic.texi (OpenACC): Adjust URL.
13669 * doc/invoke.texi (C Dialect Options): Ditto.
13671 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
13673 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
13674 predicate for operand 1. Add (m,<S>) constraint.
13675 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
13676 Prevent memory operand 1 with register operand 2.
13678 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
13680 PR rtl-optimization/82024
13681 * combine.c (try_combine): If the combination result is a PARALLEL,
13682 and we only need to retain the SET in there that would be placed
13683 at I2, check that we can place that at I3 instead, before doing so.
13685 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13688 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
13690 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
13691 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
13694 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
13695 Jeff Law <law@redhat.com>
13697 * varasm.c (bss_initializer_p): Do not put constants into .bss
13698 (categorize_decl_for_section): Handle bss_initializer_p returning
13699 false when DECL_INITIAL is NULL.
13701 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13704 * config/s390/s390.c (s390_can_inline_p): New function.
13706 2017-09-01 Jeff Law <law@redhat.com>
13708 PR tree-optimization/82052
13709 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13710 Always initialize the returned slot after a hash table miss
13711 when INSERT is true.
13713 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
13715 * config/s390/s390.md (mem_signal_fence): Remove.
13716 * doc/md.texi (mem_signal_fence): Remove.
13717 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
13719 * target-insns.def (mem_signal_fence): Remove.
13721 2017-09-01 Jakub Jelinek <jakub@redhat.com>
13724 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
13727 * asan.c (create_odr_indicator): Strip name encoding from assembler
13728 name before appending it after __odr_asan_.
13730 2017-09-01 Martin Liska <mliska@suse.cz>
13732 PR tree-optimization/82059
13733 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
13734 frequency only when an edge is redirected.
13736 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13738 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
13739 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
13740 (arc_conditional_register_usage): Remove ARC600 lp_count
13742 (arc_file_start): Emit Tag_ARC_CPU_variation.
13743 (arc_can_use_doloop_p): New conditions to use ZOLs.
13744 (hwloop_fail): New function.
13745 (hwloop_optimize): Likewise.
13746 (hwloop_pattern_reg): Likewise.
13747 (arc_doloop_hooks): New struct, to be used with reorg_loops.
13748 (arc_reorg_loops): New function, calls reorg_loops.
13749 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
13750 (arc600_corereg_hazard): Remove ZOL checking, case handled by
13752 (arc_loop_hazard): Remove function, functionality moved into
13754 (arc_hazard): Remove arc_loop_hazard call.
13755 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
13756 into hwloop_optimize.
13757 (arc_label_align): Remove ZOL handling.
13758 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
13759 * config/arc/arc.md (doloop_begin): Remove pattern.
13760 (doloop_begin_i): Likewise.
13761 (doloop_end_i): Likewise.
13762 (doloop_fallback): Likewise.
13763 (doloop_fallback_m): Likewise.
13764 (doloop_end): Reimplement expand.
13765 (arc_lp): New pattern for LP instruction.
13766 (loop_end): New pattern.
13767 (loop_fail): Likewise.
13768 (decrement_and_branch_until_zero): Likewise.
13769 * config/arc/arc.opt (mlpc-width): New option.
13770 * doc/invoke.texi (mlpc-width): Document option.
13772 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13774 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
13775 (arc_ccfsm_advance): Fix checking for delay slots.
13776 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
13778 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13780 * config/arc/arc.md (movqi_insn): Add stores to save constant long
13782 (movhi_insn): Update store instruction constraint which are saving
13783 6-bit short immediates.
13784 (movsi_insn): Consider also short scaled load operations.
13785 (zero_extendhisi2_i): Use Usd constraint instead of T.
13786 (extendhisi2_i): Add q constraint.
13787 (arc_clzsi2): Add type and length attributes.
13788 (arc_ctzsi2): Likewise.
13789 * config/arc/constraints.md (Usc): Update constraint, the
13790 assembler can parse two relocations for a single instruction.
13792 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
13794 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
13795 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
13797 2017-08-31 Olivier Hainque <hainque@adacore.com>
13799 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
13800 match as powerpc-wrs-vxworks*.
13802 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
13804 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
13805 register constraint for by-element operand.
13806 (aarch64_mls_elt_merge<mode>): Likewise.
13808 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13810 * config/arc/arc.c (arc_can_follow_jump): Check for short
13813 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13815 * config.gcc: Use g.opt for arc.
13816 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
13817 functionality moved to ...
13818 (legitimate_scaled_address_p): New function, ...here.
13819 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
13820 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
13821 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
13823 (arc_override_options): Handle G option.
13824 (arc_output_pic_addr_const): Correct function definition.
13825 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
13826 (arc_decl_anon_ns_mem_p): Delete.
13827 (arc_in_small_data_p): Overhaul this function to take into
13828 consideration the value given via G option.
13829 (arc_rewrite_small_data_1): Renamed and corrected old
13830 arc_rewrite_small_data function.
13831 (arc_rewrite_small_data): New function.
13832 (small_data_pattern): Don't use pic_offset_table_rtx.
13833 * config/arc/arc.h (CC1_SPEC): Recognize G option.
13834 * config/arc/simdext.md (movmisalignv2hi): Use
13835 prepare_move_operands function.
13837 (movmisalign*): Likewise.
13838 * doc/invoke.texi (ARC options): Document -G option.
13840 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
13842 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
13844 * config/arc/arc.c (arc_print_operand): Output scalled address for
13845 sdata whenever is possible.
13846 (arc_in_small_data_p): Allow sdata for 64bit datum when double
13847 load/stores are available.
13848 (compact_sda_memory_operand): Check for the alignment required by
13849 code density instructions.
13850 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
13852 * config/arc/constraints.md (Usd): Update constraint.
13853 (Us0): New constraint.
13854 (Usc): Update constraint.
13856 2017-08-31 Richard Biener <rguenther@suse.de>
13858 PR middle-end/82054
13859 * dwarf2out.c (dwarf2out_early_global_decl): Process each
13860 function only once.
13862 2017-08-31 Tamar Christina <tamar.christina@arm.com>
13864 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
13865 Resize type_signature.
13867 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13868 Alan Hayward <alan.hayward@arm.com>
13869 David Sherwood <david.sherwood@arm.com>
13871 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
13872 subregs whose inner modes can be stored in GPRs.
13873 (aarch64_classify_index): Likewise.
13875 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13876 Alan Hayward <alan.hayward@arm.com>
13877 David Sherwood <david.sherwood@arm.com>
13879 * config/aarch64/iterators.md (V_cmp_result): Rename to...
13880 (V_INT_EQUIV): ...this.
13881 (v_cmp_result): Rename to...
13882 (v_int_equiv): ...this.
13883 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
13884 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
13885 (copysign<mode>3): Likewise.
13886 (aarch64_simd_bsl<mode>_internal): Likewise.
13887 (aarch64_simd_bsl<mode>): Likewise.
13888 (vec_cmp<mode><mode>): Likewise.
13889 (vcond<mode><mode>): Likewise.
13890 (vcond<v_cmp_mixed><mode>): Likewise.
13891 (vcondu<mode><v_cmp_mixed>): Likewise.
13892 (aarch64_cm<optab><mode>): Likewise.
13893 (aarch64_cmtst<mode>): Likewise.
13894 (aarch64_fac<optab><mode>): Likewise.
13895 (vec_perm_const<mode>): Likewise.
13896 (vcond_mask_<mode><v_cmp_result>): Rename to...
13897 (vcond_mask_<mode><v_int_equiv>): ...this.
13898 (vec_cmp<mode><v_cmp_result>): Rename to...
13899 (vec_cmp<mode><v_int_equiv>): ...this.
13901 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
13902 Alan Hayward <alan.hayward@arm.com>
13903 David Sherwood <david.sherwood@arm.com>
13905 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
13907 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
13908 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
13909 (UNSPEC_LD4_DREG): New unspecs.
13910 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
13911 (aarch64_ld2<mode>_dreg_be): Replace with...
13912 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
13914 (aarch64_ld3<mode>_dreg_le)
13915 (aarch64_ld3<mode>_dreg_be): Replace with...
13916 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
13918 (aarch64_ld4<mode>_dreg_le)
13919 (aarch64_ld4<mode>_dreg_be): Replace with...
13920 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
13923 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13925 PR tree-optimization/81987
13926 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
13927 insert an initializer in a location not dominated by the stride
13930 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13932 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
13933 on the entire header of the finally block in the fallthru case.
13935 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
13937 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
13939 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
13941 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
13942 rs6000_emit_move_from_cr and call renamed function.
13943 (rs6000_emit_prologue): Call renamed functions.
13944 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
13945 movesi_from_cr, remove volatile CRs.
13947 2017-08-30 Jon Beniston <jon@beniston.com>
13948 Richard Biener <rguenther@suse.de>
13950 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
13951 of VECTOR_MODE_P check.
13952 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
13953 element vector types.
13955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13957 * df.h (df_read_modify_subreg_p): Remove in favor of...
13958 * rtl.h (read_modify_subreg_p): ...this new function. Take a
13959 const_rtx instead of an rtx.
13960 * cprop.c (local_cprop_find_used_regs): Update accordingly.
13961 * df-problems.c (df_word_lr_mark_ref): Likewise.
13962 * ira-lives.c (mark_pseudo_reg_live): Likewise.
13963 (mark_pseudo_reg_dead): Likewise.
13964 (mark_ref_dead): Likewise.
13965 * reginfo.c (init_subregs_of_mode): Likewise.
13966 * sched-deps.c (sched_analyze_1): Likewise.
13967 * df-scan.c (df_def_record_1): Likewise.
13968 (df_uses_record): Likewise.
13969 (df_read_modify_subreg_p): Remove in favor of...
13970 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
13971 const_rtx instead of an rtx.
13973 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13974 Alan Hayward <alan.hayward@arm.com>
13975 David Sherwood <david.sherwood@arm.com>
13977 * rtl.h (partial_subreg_p): New function.
13978 * caller-save.c (save_call_clobbered_regs): Use it.
13979 * calls.c (expand_call): Likewise.
13980 * combine.c (combinable_i3pat): Likewise.
13981 (simplify_set): Likewise.
13982 (make_extraction): Likewise.
13983 (make_compound_operation_int): Likewise.
13984 (gen_lowpart_or_truncate): Likewise.
13985 (force_to_mode): Likewise.
13986 (make_field_assignment): Likewise.
13987 (reg_truncated_to_mode): Likewise.
13988 (record_truncated_value): Likewise.
13989 (move_deaths): Likewise.
13990 * cse.c (record_jump_cond): Likewise.
13991 (cse_insn): Likewise.
13992 * cselib.c (cselib_lookup_1): Likewise.
13993 * expmed.c (extract_bit_field_using_extv): Likewise.
13994 * function.c (assign_parm_setup_reg): Likewise.
13995 * ifcvt.c (noce_convert_multiple_sets): Likewise.
13996 * ira-build.c (create_insn_allocnos): Likewise.
13997 * lra-coalesce.c (merge_pseudos): Likewise.
13998 * lra-constraints.c (match_reload): Likewise.
13999 (simplify_operand_subreg): Likewise.
14000 (curr_insn_transform): Likewise.
14001 * lra-lives.c (process_bb_lives): Likewise.
14002 * lra.c (new_insn_reg): Likewise.
14003 (lra_substitute_pseudo): Likewise.
14004 * regcprop.c (mode_change_ok): Likewise.
14005 (maybe_mode_change): Likewise.
14006 (copyprop_hardreg_forward_1): Likewise.
14007 * reload.c (push_reload): Likewise.
14008 (find_reloads): Likewise.
14009 (find_reloads_subreg_address): Likewise.
14010 * reload1.c (alter_reg): Likewise.
14011 (eliminate_regs_1): Likewise.
14012 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14014 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
14016 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
14017 back to if statements, including unpack.
14019 2017-08-30 Martin Liska <mliska@suse.cz>
14021 PR inline-asm/82001
14022 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
14024 (func_checker::compare_asm_inputs_outputs): ... this function.
14025 (func_checker::compare_gimple_asm): Use the function to compare
14026 also ASM constrains.
14027 * ipa-icf-gimple.h: Rename the function.
14029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14030 Alan Hayward <alan.hayward@arm.com>
14031 David Sherwood <david.sherwood@arm.com>
14033 * coretypes.h (complex_mode): New type.
14034 * gdbhooks.py (build_pretty_printer): Handle it.
14035 * machmode.h (complex_mode): New class.
14036 (complex_mode::includes_p): New function.
14037 (is_complex_int_mode): Likewise.
14038 (is_complex_float_mode): Likewise.
14039 * genmodes.c (get_mode_class): Handle complex mode classes.
14040 * function.c (expand_function_end): Use is_complex_int_mode.
14042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14043 Alan Hayward <alan.hayward@arm.com>
14044 David Sherwood <david.sherwood@arm.com>
14046 * coretypes.h (scalar_mode_pod): New typedef.
14047 * gdbhooks.py (build_pretty_printer): Handle it.
14048 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
14049 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
14050 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
14051 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
14052 as_a <scalar_mode>.
14054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14055 Alan Hayward <alan.hayward@arm.com>
14056 David Sherwood <david.sherwood@arm.com>
14058 * machmode.h (mode_for_vector): Take a scalar_mode instead
14060 * stor-layout.c (mode_for_vector): Likewise.
14061 * explow.c (promote_mode): Use as_a <scalar_mode>.
14062 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
14064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14065 Alan Hayward <alan.hayward@arm.com>
14066 David Sherwood <david.sherwood@arm.com>
14068 * target.def (preferred_simd_mode): Take a scalar_mode
14069 instead of a machine_mode.
14070 * targhooks.h (default_preferred_simd_mode): Likewise.
14071 * targhooks.c (default_preferred_simd_mode): Likewise.
14072 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
14073 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
14074 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
14075 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
14076 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
14077 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
14078 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
14079 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
14081 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
14082 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
14083 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
14084 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
14085 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
14086 * doc/tm.texi: Regenerate.
14087 * optabs-query.c (can_vec_mask_load_store_p): Return false for
14090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14091 Alan Hayward <alan.hayward@arm.com>
14092 David Sherwood <david.sherwood@arm.com>
14094 * target.def (scalar_mode_supported_p): Take a scalar_mode
14095 instead of a machine_mode.
14096 * targhooks.h (default_scalar_mode_supported_p): Likewise.
14097 * targhooks.c (default_scalar_mode_supported_p): Likewise.
14098 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
14099 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
14100 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
14101 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
14102 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
14103 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
14104 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
14105 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
14106 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
14107 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
14108 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
14109 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
14111 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
14112 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
14113 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
14114 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
14115 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
14117 * doc/tm.texi: Regenerate.
14119 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14120 Alan Hayward <alan.hayward@arm.com>
14121 David Sherwood <david.sherwood@arm.com>
14123 * coretypes.h (opt_scalar_mode): New typedef.
14124 * gdbhooks.py (build_pretty_printers): Handle it.
14125 * machmode.h (mode_iterator::get_2xwider): Add overload for
14127 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
14129 * expr.c (convert_mode_scalar): Likewise.
14130 * omp-low.c (omp_clause_aligned_alignment): Likewise.
14131 * optabs.c (expand_float): Likewise.
14132 (expand_fix): Likewise.
14133 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14135 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14136 Alan Hayward <alan.hayward@arm.com>
14137 David Sherwood <david.sherwood@arm.com>
14139 * optabs.c (expand_float): Explicitly check for scalars before
14140 using a branching expansion.
14141 (expand_fix): Likewise.
14143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14144 Alan Hayward <alan.hayward@arm.com>
14145 David Sherwood <david.sherwood@arm.com>
14147 * expr.c (convert_mode): Split scalar handling out into...
14148 (convert_mode_scalar): ...this new function. Treat the modes
14151 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14152 Alan Hayward <alan.hayward@arm.com>
14153 David Sherwood <david.sherwood@arm.com>
14155 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
14157 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
14159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14160 Alan Hayward <alan.hayward@arm.com>
14161 David Sherwood <david.sherwood@arm.com>
14163 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
14164 rather than a machine_mode.
14165 (fixed_from_string): Likewise.
14166 (fixed_convert): Likewise.
14167 (fixed_convert_from_int): Likewise.
14168 (fixed_convert_from_real): Likewise.
14169 (real_convert_from_fixed): Likewise.
14170 * fixed-value.c (fixed_from_double_int): Likewise.
14171 (fixed_from_string): Likewise.
14172 (fixed_convert): Likewise.
14173 (fixed_convert_from_int): Likewise.
14174 (fixed_convert_from_real): Likewise.
14175 (real_convert_from_fixed): Likewise.
14176 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
14178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14179 Alan Hayward <alan.hayward@arm.com>
14180 David Sherwood <david.sherwood@arm.com>
14182 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
14183 of separate mode class checks. Do not allow vector modes here.
14184 (immed_wide_int_const): Use as_a <scalar_mode>.
14185 * explow.c (trunc_int_for_mode): Likewise.
14186 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
14187 (wi::shwi): Likewise.
14188 (wi::min_value): Likewise.
14189 (wi::max_value): Likewise.
14190 * dwarf2out.c (loc_descriptor): Likewise.
14191 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
14192 for CONST_WIDE_INT.
14194 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14195 Alan Hayward <alan.hayward@arm.com>
14196 David Sherwood <david.sherwood@arm.com>
14198 * tree.h (SCALAR_TYPE_MODE): New macro.
14199 * expr.c (expand_expr_addr_expr_1): Use it.
14200 (expand_expr_real_2): Likewise.
14201 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
14202 (fold_convert_const_fixed_from_int): Likewise.
14203 (fold_convert_const_fixed_from_real): Likewise.
14204 (native_encode_fixed): Likewise
14205 (native_encode_complex): Likewise
14206 (native_encode_vector): Likewise.
14207 (native_interpret_fixed): Likewise.
14208 (native_interpret_real): Likewise.
14209 (native_interpret_complex): Likewise.
14210 (native_interpret_vector): Likewise.
14211 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
14212 (simd_clone_adjust_argument_types): Likewise.
14213 (simd_clone_init_simd_arrays): Likewise.
14214 (simd_clone_adjust): Likewise.
14215 * stor-layout.c (layout_type): Likewise.
14216 * tree.c (build_minus_one_cst): Likewise.
14217 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14218 * tree-inline.c (estimate_move_cost): Likewise.
14219 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
14220 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
14221 (vectorizable_reduction): Likewise.
14222 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
14223 (vect_recog_mixed_size_cond_pattern): Likewise.
14224 (check_bool_pattern): Likewise.
14225 (adjust_bool_pattern): Likewise.
14226 (search_type_for_mask_1): Likewise.
14227 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
14228 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14229 (vectorizable_load): Likewise.
14230 (vectorizable_store): Likewise.
14231 * ubsan.c (ubsan_encode_value): Likewise.
14232 * varasm.c (output_constant): Likewise.
14234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14235 Alan Hayward <alan.hayward@arm.com>
14236 David Sherwood <david.sherwood@arm.com>
14238 * coretypes.h (scalar_mode): New class.
14239 * machmode.h (scalar_mode): Likewise.
14240 (scalar_mode::includes_p): New function.
14241 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
14242 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
14243 * genmodes.c (get_mode_class): Handle remaining scalar modes.
14244 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
14245 * expmed.c (store_bit_field_1): Likewise.
14246 (extract_bit_field_1): Likewise.
14247 * expr.c (write_complex_part): Likewise.
14248 (read_complex_part): Likewise.
14249 (emit_move_complex_push): Likewise.
14250 (expand_expr_real_2): Likewise.
14251 * function.c (assign_parm_setup_reg): Likewise.
14252 (assign_parms_unsplit_complex): Likewise.
14253 * optabs.c (expand_binop): Likewise.
14254 * rtlanal.c (subreg_get_info): Likewise.
14255 * simplify-rtx.c (simplify_immed_subreg): Likewise.
14256 * varasm.c (output_constant_pool_2): Likewise.
14258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14259 Alan Hayward <alan.hayward@arm.com>
14260 David Sherwood <david.sherwood@arm.com>
14262 * expmed.c (extract_high_half): Use scalar_int_mode and remove
14264 (expmed_mult_highpart_optab): Likewise.
14265 (expmed_mult_highpart): Likewise.
14267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14268 Alan Hayward <alan.hayward@arm.com>
14269 David Sherwood <david.sherwood@arm.com>
14271 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
14272 instead of a machine_mode.
14273 (builtin_memset_read_str): Likewise.
14274 * builtins.c (c_readstr): Likewise.
14275 (builtin_memcpy_read_str): Likewise.
14276 (builtin_strncpy_read_str): Likewise.
14277 (builtin_memset_read_str): Likewise.
14278 (builtin_memset_gen_str): Likewise.
14279 (expand_builtin_signbit): Use scalar_int_mode for local variables.
14280 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
14281 instead of a machine_mode.
14282 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
14284 (make_extraction): Likewise.
14285 (try_widen_shift_mode): Take and return scalar_int_modes instead
14287 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
14288 a scalar_int_mode instead of a machine_mode.
14289 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
14290 (avr_addr_space_pointer_mode): Likewise.
14291 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
14292 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
14293 (msp430_unwind_word_mode): Likewise.
14294 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
14295 (spu_addr_space_pointer_mode): Likewise.
14296 (spu_addr_space_address_mode): Likewise.
14297 (spu_libgcc_cmp_return_mode): Likewise.
14298 (spu_libgcc_shift_count_mode): Likewise.
14299 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
14300 (rl78_addr_space_pointer_mode): Likewise.
14301 (fl78_unwind_word_mode): Likewise.
14302 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
14304 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
14305 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
14306 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
14307 (mips_valid_pointer_mode): Likewise.
14308 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
14309 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
14310 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
14312 (ft32_addr_space_address_mode): Likewise.
14313 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
14314 scalar_int_mode instead of a machine_mode.
14315 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
14317 (m32c_addr_space_address_mode): Likewise.
14318 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
14319 (rs6000_eh_return_filter_mode): Likewise.
14320 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
14321 (rs6000_eh_return_filter_mode): Likewise.
14322 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
14323 (s390_libgcc_shift_count_mode): Likewise.
14324 (s390_unwind_word_mode): Likewise.
14325 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
14327 * target.def (mode_rep_extended): Likewise.
14328 (valid_pointer_mode): Likewise.
14329 (addr_space.valid_pointer_mode): Likewise.
14330 (eh_return_filter_mode): Return a scalar_int_mode rather than
14332 (libgcc_cmp_return_mode): Likewise.
14333 (libgcc_shift_count_mode): Likewise.
14334 (unwind_word_mode): Likewise.
14335 (addr_space.pointer_mode): Likewise.
14336 (addr_space.address_mode): Likewise.
14337 * doc/tm.texi: Regenerate.
14338 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
14340 (do_jump): Use scalar_int_mode for local variables.
14341 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
14342 rather than a machine_mode.
14343 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
14344 (scompare_loc_descriptor_wide): Likewise.
14345 (scompare_loc_descriptor_narrow): Likewise.
14346 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
14348 * except.c (sjlj_emit_dispatch_table): Likewise.
14349 (expand_builtin_eh_copy_values): Likewise.
14350 * explow.c (convert_memory_address_addr_space_1): Likewise.
14351 Take a scalar_int_mode rather than a machine_mode.
14352 (convert_memory_address_addr_space): Take a scalar_int_mode rather
14353 than a machine_mode.
14354 (memory_address_addr_space): Use scalar_int_mode for local variables.
14355 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
14356 rather than a machine_mode.
14357 * expmed.c (mask_rtx): Likewise.
14358 (init_expmed_one_conv): Likewise.
14359 (expand_mult_highpart_adjust): Likewise.
14360 (extract_high_half): Likewise.
14361 (expmed_mult_highpart_optab): Likewise.
14362 (expmed_mult_highpart): Likewise.
14363 (expand_smod_pow2): Likewise.
14364 (expand_sdiv_pow2): Likewise.
14365 (emit_store_flag_int): Likewise.
14366 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
14368 (extract_low_bits): Likewise.
14369 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
14371 * expr.c (pieces_addr::adjust): Likewise.
14372 (can_store_by_pieces): Likewise.
14373 (store_by_pieces): Likewise.
14374 (clear_by_pieces_1): Likewise.
14375 (expand_expr_addr_expr_1): Likewise.
14376 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
14377 (expand_expr_real_1): Likewise.
14378 (try_casesi): Likewise.
14379 * final.c (shorten_branches): Likewise.
14380 * fold-const.c (fold_convert_const_int_from_fixed): Change the
14381 type of "mode" to machine_mode.
14382 * internal-fn.c (expand_arith_overflow_result_store): Take a
14383 scalar_int_mode rather than a machine_mode.
14384 (expand_mul_overflow): Use scalar_int_mode for local variables.
14385 * loop-doloop.c (doloop_modify): Likewise.
14386 (doloop_optimize): Likewise.
14387 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
14388 than a machine_mode.
14389 (expand_doubleword_shift_condmove): Likewise.
14390 (expand_doubleword_shift): Likewise.
14391 (expand_doubleword_clz): Likewise.
14392 (expand_doubleword_popcount): Likewise.
14393 (expand_doubleword_parity): Likewise.
14394 (expand_absneg_bit): Use scalar_int_mode for local variables.
14395 (prepare_float_lib_cmp): Likewise.
14396 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
14397 rather than a machine_mode.
14398 (convert_memory_address_addr_space): Likewise.
14399 (get_mode_bounds): Likewise.
14400 (get_address_mode): Return a scalar_int_mode rather than a
14402 * rtlanal.c (get_address_mode): Likewise.
14403 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
14404 than a machine_mode.
14405 * targhooks.c (default_mode_rep_extended): Likewise.
14406 (default_valid_pointer_mode): Likewise.
14407 (default_addr_space_valid_pointer_mode): Likewise.
14408 (default_eh_return_filter_mode): Return a scalar_int_mode rather
14409 than a machine_mode.
14410 (default_libgcc_cmp_return_mode): Likewise.
14411 (default_libgcc_shift_count_mode): Likewise.
14412 (default_unwind_word_mode): Likewise.
14413 (default_addr_space_pointer_mode): Likewise.
14414 (default_addr_space_address_mode): Likewise.
14415 * targhooks.h (default_eh_return_filter_mode): Likewise.
14416 (default_libgcc_cmp_return_mode): Likewise.
14417 (default_libgcc_shift_count_mode): Likewise.
14418 (default_unwind_word_mode): Likewise.
14419 (default_addr_space_pointer_mode): Likewise.
14420 (default_addr_space_address_mode): Likewise.
14421 (default_mode_rep_extended): Take a scalar_int_mode rather than
14423 (default_valid_pointer_mode): Likewise.
14424 (default_addr_space_valid_pointer_mode): Likewise.
14425 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
14427 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
14428 rather than a machine_mode.
14429 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
14430 for local variables.
14431 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14432 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
14433 than a machine_mode.
14435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14436 Alan Hayward <alan.hayward@arm.com>
14437 David Sherwood <david.sherwood@arm.com>
14439 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
14440 the mode argument to scalar_int_mode.
14441 (do_jump_by_parts_zero_rtx): Likewise.
14442 (do_jump_by_parts_equality_rtx): Likewise.
14443 (do_jump_by_parts_greater): Take a mode argument.
14444 (do_jump_by_parts_equality): Likewise.
14445 (do_jump_1): Update calls accordingly.
14447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14448 Alan Hayward <alan.hayward@arm.com>
14449 David Sherwood <david.sherwood@arm.com>
14451 * is-a.h (safe_dyn_cast): New function.
14452 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
14453 (jump_table_for_label): Likewise.
14454 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
14455 instead of an rtx_insn *.
14456 (shorten_branches): Use dyn_cast instead of LABEL_P and
14457 JUMP_TABLE_DATA_P. Use jump_table_for_label and
14458 rtx_jump_table_data::get_data_mode.
14459 (final_scan_insn): Likewise.
14461 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14462 Alan Hayward <alan.hayward@arm.com>
14463 David Sherwood <david.sherwood@arm.com>
14465 * combine.c (try_combine): Use is_a <scalar_int_mode> when
14466 trying to combine a full-register integer set with a subreg
14469 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14470 Alan Hayward <alan.hayward@arm.com>
14471 David Sherwood <david.sherwood@arm.com>
14473 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
14474 that is always either address_mode or pointer_mode.
14476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14477 Alan Hayward <alan.hayward@arm.com>
14478 David Sherwood <david.sherwood@arm.com>
14480 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
14481 when the two are known to be equal.
14483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14484 Alan Hayward <alan.hayward@arm.com>
14485 David Sherwood <david.sherwood@arm.com>
14487 * simplify-rtx.c (simplify_const_unary_operation): Use
14488 is_a <scalar_int_mode> instead of checking for a nonzero
14489 precision. Forcibly convert op_mode to a scalar_int_mode
14490 in that case. More clearly differentiate the operand and
14491 result modes and use the former when deciding what the value
14492 of a count-bits operation should be. Use is_int_mode instead
14493 of checking for a MODE_INT. Remove redundant check for whether
14494 this mode has a zero precision.
14496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14497 Alan Hayward <alan.hayward@arm.com>
14498 David Sherwood <david.sherwood@arm.com>
14500 * optabs.c (widen_leading): Change the type of the mode argument
14501 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
14502 (widen_bswap): Likewise.
14503 (expand_parity): Likewise.
14504 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
14505 (expand_ffs): Likewise.
14506 (epand_unop): Check for scalar integer modes before calling the
14509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14510 Alan Hayward <alan.hayward@arm.com>
14511 David Sherwood <david.sherwood@arm.com>
14513 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
14515 (expand_expr_real_1): Update call accordingly.
14517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14518 Alan Hayward <alan.hayward@arm.com>
14519 David Sherwood <david.sherwood@arm.com>
14521 * expmed.c (store_bit_field_using_insv): Add op0_mode and
14522 value_mode arguments. Use scalar_int_mode internally.
14523 (store_bit_field_1): Rename the new integer mode from imode
14524 to op0_mode and use it instead of GET_MODE (op0). Update calls
14525 to store_split_bit_field, store_bit_field_using_insv and
14526 store_fixed_bit_field.
14527 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
14528 Use scalar_int_mode internally. Use a bit count rather than a mode
14529 when calculating the largest bit size for get_best_mode.
14530 Update calls to store_split_bit_field and store_fixed_bit_field_1.
14531 (store_fixed_bit_field_1): Add mode and value_mode arguments.
14532 Remove assertion that OP0 has a scalar integer mode.
14533 (store_split_bit_field): Add op0_mode and value_mode arguments.
14534 Update calls to extract_fixed_bit_field.
14535 (extract_bit_field_using_extv): Add an op0_mode argument.
14536 Use scalar_int_mode internally.
14537 (extract_bit_field_1): Rename the new integer mode from imode to
14538 op0_mode and use it instead of GET_MODE (op0). Update calls to
14539 extract_split_bit_field, extract_bit_field_using_extv and
14540 extract_fixed_bit_field.
14541 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
14542 to extract_split_bit_field and extract_fixed_bit_field_1.
14543 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
14544 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
14545 on the target mode.
14546 (extract_split_bit_field): Add an op0_mode argument. Update call
14547 to extract_fixed_bit_field.
14549 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14550 Alan Hayward <alan.hayward@arm.com>
14551 David Sherwood <david.sherwood@arm.com>
14553 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
14554 * explow.c (hard_function_value): Likewise.
14555 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
14556 convert_to_mode call outside the loop.
14557 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
14558 for the mode iterator. Require the mode specified by max_pieces
14560 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
14562 (copy_blkmode_to_reg): Likewise.
14563 (set_storage_via_setmem): Likewise.
14564 * optabs.c (prepare_cmp_insn): Likewise.
14565 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
14566 * stor-layout.c (finish_bitfield_representative): Likewise.
14568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14569 Alan Hayward <alan.hayward@arm.com>
14570 David Sherwood <david.sherwood@arm.com>
14572 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
14573 * expr.c (convert_move): Use them.
14574 (convert_modes): Likewise.
14575 (store_expr_with_bounds): Likewise.
14577 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14578 Alan Hayward <alan.hayward@arm.com>
14579 David Sherwood <david.sherwood@arm.com>
14581 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
14582 parameter for the mode of "x". Remove the "known_x", "known_mode"
14583 and "known_ret" arguments. Change the type of the mode argument
14584 to scalar_int_mode.
14585 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
14586 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
14587 (reg_num_sign_bit_copies_for_combine): Likewise.
14588 * rtlanal.c (nonzero_bits1): Likewise.
14589 (num_sign_bit_copies1): Likewise.
14590 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
14591 (reg_num_sign_bit_copies_general): Likewise.
14592 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
14593 (reg_nonzero_bits_general): Likewise.
14595 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14596 Alan Hayward <alan.hayward@arm.com>
14597 David Sherwood <david.sherwood@arm.com>
14599 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
14600 than in subroutines. Return 1 for non-integer modes.
14601 (cached_num_sign_bit_copies): Change the type of the mode parameter
14602 to scalar_int_mode.
14603 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
14604 classes. Handle CONST_INT_P first and then check whether X also
14605 has a scalar integer mode. Check the same thing for inner registers
14606 of a SUBREG and for values that are being extended or truncated.
14608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14609 Alan Hayward <alan.hayward@arm.com>
14610 David Sherwood <david.sherwood@arm.com>
14612 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
14613 in subroutines. Return the mode mask for non-integer modes.
14614 (cached_nonzero_bits): Change the type of the mode parameter
14615 to scalar_int_mode.
14616 (nonzero_bits1): Likewise. Remove early exit for other mode
14617 classes. Handle CONST_INT_P first and then check whether X
14618 also has a scalar integer mode.
14620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14621 Alan Hayward <alan.hayward@arm.com>
14622 David Sherwood <david.sherwood@arm.com>
14624 * expr.c (widest_int_mode_for_size): Make the comment match the code.
14625 Return a scalar_int_mode and assert that the size is greater than
14627 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
14628 (op_by_pieces_d::op_by_pieces_d): Likewise.
14629 (op_by_pieces_d::run): Likewise.
14630 (can_store_by_pieces): Likewise.
14632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14633 Alan Hayward <alan.hayward@arm.com>
14634 David Sherwood <david.sherwood@arm.com>
14636 * combine.c (extract_left_shift): Add a mode argument and update
14638 (make_compound_operation_int): Change the type of the mode parameter
14639 to scalar_int_mode and update the call to extract_left_shift.
14641 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14642 Alan Hayward <alan.hayward@arm.com>
14643 David Sherwood <david.sherwood@arm.com>
14645 * combine.c (simplify_and_const_int): Change the type of the mode
14646 parameter to scalar_int_mode.
14647 (simplify_and_const_int_1): Likewise. Update recursive call.
14649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14650 Alan Hayward <alan.hayward@arm.com>
14651 David Sherwood <david.sherwood@arm.com>
14653 * combine.c (simplify_compare_const): Check that the mode is a
14654 scalar_int_mode (rather than VOIDmode) before testing its
14656 (simplify_comparison): Move COMPARISON_P handling out of the
14657 loop and restrict the latter part of the loop to scalar_int_modes.
14658 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
14659 and when considering SUBREG_REGs. Use is_int_mode instead of
14660 checking GET_MODE_CLASS against MODE_INT.
14662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14663 Alan Hayward <alan.hayward@arm.com>
14664 David Sherwood <david.sherwood@arm.com>
14666 * combine.c (try_widen_shift_mode): Move check for equal modes to...
14667 (simplify_shift_const_1): ...here. Use scalar_int_mode for
14668 shift_unit_mode and for modes involved in scalar shifts.
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 * combine.c (force_int_to_mode): New function, split out from...
14675 (force_to_mode): ...here. Keep xmode up-to-date and use it
14676 instead of GET_MODE (x).
14678 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14679 Alan Hayward <alan.hayward@arm.com>
14680 David Sherwood <david.sherwood@arm.com>
14682 * optabs-query.h (extraction_insn::struct_mode): Change type to
14683 opt_scalar_int_mode and update comment.
14684 (extraction_insn::field_mode): Change type to scalar_int_mode.
14685 (extraction_insn::pos_mode): Likewise.
14686 * combine.c (make_extraction): Update accordingly.
14687 * optabs-query.c (get_traditional_extraction_insn): Likewise.
14688 (get_optab_extraction_insn): Likewise.
14689 * recog.c (simplify_while_replacing): Likewise.
14690 * expmed.c (narrow_bit_field_mem): Change the type of the mode
14691 parameter to opt_scalar_int_mode.
14693 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14694 Alan Hayward <alan.hayward@arm.com>
14695 David Sherwood <david.sherwood@arm.com>
14697 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
14698 to a scalar_int_mode instead of a machine_mode.
14699 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
14700 (get_best_mode): Return a boolean and use a pointer argument to store
14701 the selected mode. Replace the limit mode parameter with a bit limit.
14702 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
14703 for the values returned by bit_field_mode_iterator::next_mode.
14704 (store_bit_field): Update call to get_best_mode.
14705 (store_fixed_bit_field): Likewise.
14706 (extract_fixed_bit_field): Likewise.
14707 * expr.c (optimize_bitfield_assignment_op): Likewise.
14708 * fold-const.c (optimize_bit_field_compare): Likewise.
14709 (fold_truth_andor_1): Likewise.
14710 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
14711 Update for new type of m_mode.
14712 (get_best_mode): As above.
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 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
14719 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
14720 (store_bit_field): Check is_a <scalar_int_mode> before calling
14721 strict_volatile_bitfield_p.
14722 (extract_bit_field): Likewise.
14724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14725 Alan Hayward <alan.hayward@arm.com>
14726 David Sherwood <david.sherwood@arm.com>
14728 * target.def (cstore_mode): Return a scalar_int_mode.
14729 * doc/tm.texi: Regenerate.
14730 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
14731 * targhooks.h (default_cstore_mode): Likewise.
14732 * targhooks.c (default_cstore_mode): Likewise, using a forced
14734 * expmed.c (emit_cstore): Expect the target of the cstore to be
14737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14738 Alan Hayward <alan.hayward@arm.com>
14739 David Sherwood <david.sherwood@arm.com>
14741 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
14743 (niter_desc): Likewise mode.
14744 (iv_analyze): Add a mode parameter.
14746 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
14747 and change its type to scalar_int_mode.
14748 * loop-iv.c: Update commentary at head of file.
14749 (iv_constant): Pass the mode paraeter before the rtx it describes
14750 and change its type to scalar_int_mode. Remove VOIDmode handling.
14751 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
14752 (iv_extend): Likewise.
14753 (shorten_into_mode): Likewise.
14754 (iv_add): Use scalar_int_mode.
14755 (iv_mult): Likewise.
14756 (iv_shift): Likewise.
14757 (canonicalize_iv_subregs): Likewise.
14758 (get_biv_step_1): Pass the outer_mode parameter before the rtx
14759 it describes and change its mode to scalar_int_mode. Also change
14760 the type of the returned inner_mode to scalar_int_mode.
14761 (get_biv_step): Likewise, turning outer_mode from a pointer
14762 into a direct parameter. Update call to get_biv_step_1.
14763 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
14764 iv_constant and get_biv_step.
14765 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
14766 and change its type to scalar_int_mode. Don't initialise iv->mode
14767 to VOIDmode and remove later checks for its still being VOIDmode.
14768 Update calls to iv_analyze_op and iv_analyze_expr. Check
14769 is_a <scalar_int_mode> when changing the mode under consideration.
14770 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
14771 Update call to iv_analyze_expr.
14772 (iv_analyze_op): Add a mode parameter. Reject subregs whose
14773 inner register is not also a scalar_int_mode. Update call to
14775 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
14776 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
14777 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
14778 separate mode class checks. Update calls to iv_analyze. Remove
14779 fix-up of VOIDmodes after iv_analyze_biv.
14780 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
14781 don't have a scalar_int_mode. Update call to biv_p.
14783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14784 Alan Hayward <alan.hayward@arm.com>
14785 David Sherwood <david.sherwood@arm.com>
14787 * cfgexpand.c (convert_debug_memory_address): Use
14788 as_a <scalar_int_mode>.
14789 * combine.c (expand_compound_operation): Likewise.
14790 (make_extraction): Likewise.
14791 (change_zero_ext): Likewise.
14792 (simplify_comparison): Likewise.
14793 * cse.c (cse_insn): Likewise.
14794 * dwarf2out.c (minmax_loc_descriptor): Likewise.
14795 (mem_loc_descriptor): Likewise.
14796 (loc_descriptor): Likewise.
14797 * expmed.c (init_expmed_one_mode): Likewise.
14798 (synth_mult): Likewise.
14799 (emit_store_flag_1): Likewise.
14800 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
14801 of a comparison with size.
14802 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
14803 (reduce_to_bit_field_precision): Likewise.
14804 * function.c (expand_function_end): Likewise.
14805 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
14806 * loop-doloop.c (doloop_modify): Likewise.
14807 * optabs.c (expand_binop): Likewise.
14808 (expand_unop): Likewise.
14809 (expand_copysign_absneg): Likewise.
14810 (prepare_cmp_insn): Likewise.
14811 (maybe_legitimize_operand): Likewise.
14812 * recog.c (const_scalar_int_operand): Likewise.
14813 * rtlanal.c (get_address_mode): Likewise.
14814 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14815 (simplify_cond_clz_ctz): Likewise.
14816 * tree-nested.c (get_nl_goto_field): Likewise.
14817 * tree.c (build_vector_type_for_mode): Likewise.
14818 * var-tracking.c (use_narrower_mode): Likewise.
14820 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14821 Alan Hayward <alan.hayward@arm.com>
14822 David Sherwood <david.sherwood@arm.com>
14824 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
14825 * builtins.c (expand_builtin_signbit): Use it.
14826 * cfgexpand.c (expand_debug_expr): Likewise.
14827 * dojump.c (do_jump): Likewise.
14828 (do_compare_and_jump): Likewise.
14829 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
14830 * expmed.c (make_tree): Likewise.
14831 * expr.c (expand_expr_real_2): Likewise.
14832 (expand_expr_real_1): Likewise.
14833 (try_casesi): Likewise.
14834 * fold-const-call.c (fold_const_call_ss): Likewise.
14835 * fold-const.c (unextend): Likewise.
14836 (extract_muldiv_1): Likewise.
14837 (fold_single_bit_test): Likewise.
14838 (native_encode_int): Likewise.
14839 (native_encode_string): Likewise.
14840 (native_interpret_int): Likewise.
14841 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
14842 * internal-fn.c (expand_addsub_overflow): Likewise.
14843 (expand_neg_overflow): Likewise.
14844 (expand_mul_overflow): Likewise.
14845 (expand_arith_overflow): Likewise.
14846 * match.pd: Likewise.
14847 * stor-layout.c (layout_type): Likewise.
14848 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14849 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
14850 * tree-ssanames.c (get_range_info): Likewise.
14851 * tree-switch-conversion.c (array_value_type) Likewise.
14852 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
14853 (vect_recog_divmod_pattern): Likewise.
14854 (vect_recog_mixed_size_cond_pattern): Likewise.
14855 * tree-vrp.c (extract_range_basic): Likewise.
14856 (simplify_float_conversion_using_ranges): Likewise.
14857 * tree.c (int_fits_type_p): Likewise.
14858 * ubsan.c (instrument_bool_enum_load): Likewise.
14859 * varasm.c (mergeable_string_section): Likewise.
14860 (narrowing_initializer_constant_valid_p): Likewise.
14861 (output_constant): Likewise.
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 * machmode.h (NARROWEST_INT_MODE): New macro.
14868 * expr.c (alignment_for_piecewise_move): Use it instead of
14869 GET_CLASS_NARROWEST_MODE (MODE_INT).
14870 (push_block): Likewise.
14871 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
14873 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
14875 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14876 Alan Hayward <alan.hayward@arm.com>
14877 David Sherwood <david.sherwood@arm.com>
14879 * postreload.c (move2add_valid_value_p): Change the type of the
14880 mode parameter to scalar_int_mode.
14881 (move2add_use_add2_insn): Add a mode parameter and use it instead
14883 (move2add_use_add3_insn): Likewise.
14884 (reload_cse_move2add): Update accordingly.
14886 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14887 Alan Hayward <alan.hayward@arm.com>
14888 David Sherwood <david.sherwood@arm.com>
14890 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
14892 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
14893 * optabs.c (expand_unop): Likewise.
14895 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14896 Alan Hayward <alan.hayward@arm.com>
14897 David Sherwood <david.sherwood@arm.com>
14899 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
14900 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
14901 (popcount_loc_descriptor): Likewise.
14902 (bswap_loc_descriptor): Likewise.
14903 (rotate_loc_descriptor): Likewise.
14904 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
14905 calling the functions above.
14907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14908 Alan Hayward <alan.hayward@arm.com>
14909 David Sherwood <david.sherwood@arm.com>
14911 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
14913 (try_combine): Likewise.
14914 (simplify_if_then_else): Likewise.
14915 * cse.c (cse_insn): Likewise.
14916 * dwarf2out.c (mem_loc_descriptor): Likewise.
14917 * emit-rtl.c (gen_lowpart_common): Likewise.
14918 * simplify-rtx.c (simplify_truncation): Likewise.
14919 (simplify_binary_operation_1): Likewise.
14920 (simplify_const_relational_operation): Likewise.
14921 (simplify_ternary_operation): Likewise.
14922 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
14924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14925 Alan Hayward <alan.hayward@arm.com>
14926 David Sherwood <david.sherwood@arm.com>
14928 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
14929 * reload.c (push_reload): Likewise.
14930 (find_reloads): Likewise.
14932 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14933 Alan Hayward <alan.hayward@arm.com>
14934 David Sherwood <david.sherwood@arm.com>
14936 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
14937 (make_compound_operation_int): Likewise.
14938 (change_zero_ext): Likewise.
14939 * expr.c (convert_move): Likewise.
14940 (convert_modes): Likewise.
14941 * fwprop.c (forward_propagate_subreg): Likewise.
14942 * loop-iv.c (get_biv_step_1): Likewise.
14943 * optabs.c (widen_operand): Likewise.
14944 * postreload.c (move2add_valid_value_p): Likewise.
14945 * recog.c (simplify_while_replacing): Likewise.
14946 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14947 (simplify_binary_operation_1): Likewise. Remove redundant
14948 mode equality check.
14950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14951 Alan Hayward <alan.hayward@arm.com>
14952 David Sherwood <david.sherwood@arm.com>
14954 * combine.c (combine_simplify_rtx): Add checks for
14955 is_a <scalar_int_mode>.
14956 (simplify_if_then_else): Likewise.
14957 (make_field_assignment): Likewise.
14958 (simplify_comparison): Likewise.
14959 * ifcvt.c (noce_try_bitop): Likewise.
14960 * loop-invariant.c (canonicalize_address_mult): Likewise.
14961 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14963 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14964 Alan Hayward <alan.hayward@arm.com>
14965 David Sherwood <david.sherwood@arm.com>
14967 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
14968 is_a <scalar_int_mode> instead of != BLKmode.
14970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14971 Alan Hayward <alan.hayward@arm.com>
14972 David Sherwood <david.sherwood@arm.com>
14974 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
14975 instead of != VOIDmode.
14976 * combine.c (if_then_else_cond): Likewise.
14977 (change_zero_ext): Likewise.
14978 * dwarf2out.c (mem_loc_descriptor): Likewise.
14979 (loc_descriptor): Likewise.
14980 * rtlanal.c (canonicalize_condition): Likewise.
14981 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
14983 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14984 Alan Hayward <alan.hayward@arm.com>
14985 David Sherwood <david.sherwood@arm.com>
14987 * simplify-rtx.c (simplify_binary_operation_1): Use
14988 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
14990 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
14991 Alan Hayward <alan.hayward@arm.com>
14992 David Sherwood <david.sherwood@arm.com>
14994 * wide-int.h (int_traits<unsigned char>) New class.
14995 (int_traits<unsigned short>) Likewise.
14996 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
14997 Use GET_MODE_UNIT_PRECISION and remove redundant test for
14999 * combine.c (set_nonzero_bits_and_sign_copies): Use
15000 is_a <scalar_int_mode>.
15001 (find_split_point): Likewise.
15002 (combine_simplify_rtx): Likewise.
15003 (simplify_logical): Likewise.
15004 (expand_compound_operation): Likewise.
15005 (expand_field_assignment): Likewise.
15006 (make_compound_operation): Likewise.
15007 (extended_count): Likewise.
15008 (change_zero_ext): Likewise.
15009 (simplify_comparison): Likewise.
15010 * dwarf2out.c (scompare_loc_descriptor): Likewise.
15011 (ucompare_loc_descriptor): Likewise.
15012 (minmax_loc_descriptor): Likewise.
15013 (mem_loc_descriptor): Likewise.
15014 (loc_descriptor): Likewise.
15015 * expmed.c (init_expmed_one_mode): Likewise.
15016 * lra-constraints.c (lra_constraint_offset): Likewise.
15017 * optabs.c (prepare_libcall_arg): Likewise.
15018 * postreload.c (move2add_note_store): Likewise.
15019 * reload.c (operands_match_p): Likewise.
15020 * rtl.h (load_extend_op): Likewise.
15021 * rtlhooks.c (gen_lowpart_general): Likewise.
15022 * simplify-rtx.c (simplify_truncation): Likewise.
15023 (simplify_unary_operation_1): Likewise.
15024 (simplify_binary_operation_1): Likewise.
15025 (simplify_const_binary_operation): Likewise.
15026 (simplify_const_relational_operation): Likewise.
15027 (simplify_subreg): Likewise.
15028 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15029 * var-tracking.c (adjust_mems): Likewise.
15030 (prepare_call_arguments): Likewise.
15032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15033 Alan Hayward <alan.hayward@arm.com>
15034 David Sherwood <david.sherwood@arm.com>
15036 * machmode.h (is_int_mode): New fuction.
15037 * combine.c (find_split_point): Use it.
15038 (combine_simplify_rtx): Likewise.
15039 (simplify_if_then_else): Likewise.
15040 (simplify_set): Likewise.
15041 (simplify_shift_const_1): Likewise.
15042 (simplify_comparison): Likewise.
15043 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
15044 * cse.c (notreg_cost): Likewise.
15045 (cse_insn): Likewise.
15046 * cselib.c (cselib_lookup_1): Likewise.
15047 * dojump.c (do_jump_1): Likewise.
15048 (do_compare_rtx_and_jump): Likewise.
15049 * dse.c (get_call_args): Likewise.
15050 * dwarf2out.c (rtl_for_decl_init): Likewise.
15051 (native_encode_initializer): Likewise.
15052 * expmed.c (emit_store_flag_1): Likewise.
15053 (emit_store_flag): Likewise.
15054 * expr.c (convert_modes): Likewise.
15055 (store_field): Likewise.
15056 (expand_expr_real_1): Likewise.
15057 * fold-const.c (fold_read_from_constant_string): Likewise.
15058 * gimple-ssa-sprintf.c (get_format_string): Likewise.
15059 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
15060 * optabs.c (expand_binop): Likewise.
15061 (expand_unop): Likewise.
15062 (expand_abs_nojump): Likewise.
15063 (expand_one_cmpl_abs_nojump): Likewise.
15064 * simplify-rtx.c (mode_signbit_p): Likewise.
15065 (val_signbit_p): Likewise.
15066 (val_signbit_known_set_p): Likewise.
15067 (val_signbit_known_clear_p): Likewise.
15068 (simplify_relational_operation_1): Likewise.
15069 * tree.c (vector_type_mode): Likewise.
15071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15072 Alan Hayward <alan.hayward@arm.com>
15073 David Sherwood <david.sherwood@arm.com>
15075 * machmode.h (smallest_mode_for_size): Fix formatting.
15076 (smallest_int_mode_for_size): New function.
15077 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
15078 instead of smallest_mode_for_size.
15079 * combine.c (make_extraction): Likewise.
15080 * config/arc/arc.c (arc_expand_movmem): Likewise.
15081 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
15082 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15083 * config/s390/s390.c (s390_expand_insv): Likewise.
15084 * config/sparc/sparc.c (assign_int_registers): Likewise.
15085 * config/spu/spu.c (spu_function_value): Likewise.
15086 (spu_function_arg): Likewise.
15087 * coverage.c (get_gcov_type): Likewise.
15088 (get_gcov_unsigned_t): Likewise.
15089 * dse.c (find_shift_sequence): Likewise.
15090 * expmed.c (store_bit_field_1): Likewise.
15091 * expr.c (convert_move): Likewise.
15092 (store_field): Likewise.
15093 * internal-fn.c (expand_arith_overflow): Likewise.
15094 * optabs-query.c (get_best_extraction_insn): Likewise.
15095 * optabs.c (expand_twoval_binop_libfunc): Likewise.
15096 * stor-layout.c (layout_type): Likewise.
15097 (initialize_sizetypes): Likewise.
15098 * targhooks.c (default_get_mask_mode): Likewise.
15099 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
15101 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15102 Alan Hayward <alan.hayward@arm.com>
15103 David Sherwood <david.sherwood@arm.com>
15105 * machmode.h (opt_mode::else_blk): New function.
15106 (int_mode_for_mode): Declare.
15107 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
15108 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
15110 * cfgexpand.c (expand_debug_expr): Likewise.
15111 * combine.c (gen_lowpart_or_truncate): Likewise.
15112 (gen_lowpart_for_combine): Likewise.
15113 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
15114 * config/avr/avr.c (avr_to_int_mode): Likewise.
15115 (avr_out_plus_1): Likewise.
15116 (avr_out_plus): Likewise.
15117 (avr_out_round): Likewise.
15118 * config/i386/i386.c (ix86_split_to_parts): Likewise.
15119 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15120 (s390_expand_vcond): Likewise.
15121 * config/spu/spu.c (spu_split_immediate): Likewise.
15122 (spu_expand_mov): Likewise.
15123 * dse.c (get_stored_val): Likewise.
15124 * expmed.c (store_bit_field_1): Likewise.
15125 (convert_extracted_bit_field): Use int_mode_for_mode instead of
15127 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
15128 (extract_low_bits): Likewise.
15129 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
15130 handling rather than repeating the check.
15131 (emit_group_store): Likewise.
15132 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
15133 * optabs.c (expand_absneg_bit): Likewise.
15134 (expand_copysign_absneg): Likewise.
15135 (expand_copysign_bit): Likewise.
15136 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
15137 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
15138 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
15139 * var-tracking.c (prepare_call_arguments): Likewise.
15140 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
15141 int_mode_for_mode instead of mode_for_size.
15142 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15145 Alan Hayward <alan.hayward@arm.com>
15146 David Sherwood <david.sherwood@arm.com>
15148 * machmode.h (int_mode_for_size): New function.
15149 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
15150 instead of mode_for_size.
15151 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
15153 * combine.c (expand_field_assignment): Use int_mode_for_size
15154 instead of mode_for_size.
15155 (make_extraction): Likewise.
15156 (simplify_shift_const_1): Likewise.
15157 (simplify_comparison): Likewise.
15158 * dojump.c (do_jump): Likewise.
15159 * dwarf2out.c (mem_loc_descriptor): Likewise.
15160 * emit-rtl.c (init_derived_machine_modes): Likewise.
15161 * expmed.c (flip_storage_order): Likewise.
15162 (convert_extracted_bit_field): Likewise.
15163 * expr.c (copy_blkmode_from_reg): Likewise.
15164 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
15165 * internal-fn.c (expand_mul_overflow): Likewise.
15166 * lower-subreg.c (simple_move): Likewise.
15167 * optabs-libfuncs.c (init_optabs): Likewise.
15168 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15169 * tree.c (vector_type_mode): Likewise.
15170 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
15171 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
15172 * tree-vect-generic.c (expand_vector_parallel): Likewise.
15173 * tree-vect-stmts.c (vectorizable_load): Likewise.
15174 (vectorizable_store): Likewise.
15176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15177 Alan Hayward <alan.hayward@arm.com>
15178 David Sherwood <david.sherwood@arm.com>
15180 * coretypes.h (pod_mode): New type.
15181 (scalar_int_mode_pod): New typedef.
15182 * machmode.h (pod_mode): New class.
15183 (int_n_data_t::m): Change type to scalar_int_mode_pod.
15184 * genmodes.c (emit_mode_int_n): Update accordingly.
15185 * lower-subreg.h (target_lower_subreg): Change type to
15186 scalar_int_mode_pod.
15187 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
15188 scalar_int_mode_pod.
15190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15191 Alan Hayward <alan.hayward@arm.com>
15192 David Sherwood <david.sherwood@arm.com>
15194 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
15195 machine_mode to scalar_int_mode.
15196 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
15197 (rs6000_option_override_internal): Remove cast to int.
15198 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
15199 machine_mode to scalar_int_mode.
15200 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
15201 (rs6000_option_override_internal): Remove cast to int.
15202 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
15203 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
15205 * config/s390/s390.c (s390_expand_builtin): Likewise.
15206 * coretypes.h (scalar_int_mode): New type.
15207 (opt_scalar_int_mode): New typedef.
15208 * machmode.h (scalar_int_mode): New class.
15209 (scalar_int_mode::includes_p): New function.
15210 (byte_mode): Change type to scalar_int_mode.
15211 (word_mode): Likewise.
15212 (ptr_mode): Likewise.
15213 * emit-rtl.c (byte_mode): Likewise.
15214 (word_mode): Likewise.
15215 (ptr_mode): Likewise.
15216 (init_derived_machine_modes): Update accordingly.
15217 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
15218 and MODE_PARTIAL_INT.
15219 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
15220 opt_scalar_int_mode.
15222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15223 Alan Hayward <alan.hayward@arm.com>
15224 David Sherwood <david.sherwood@arm.com>
15226 * target.def (libgcc_floating_mode_supported_p): Take a
15228 * doc/tm.texi: Regenerate.
15229 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
15231 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15232 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
15235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15236 Alan Hayward <alan.hayward@arm.com>
15237 David Sherwood <david.sherwood@arm.com>
15239 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
15240 * doc/tm.texi: Regenerate.
15241 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
15242 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
15243 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
15244 * targhooks.h (default_floatn_mode): Likewise.
15245 * targhooks.c (default_floatn_mode): Likewise.
15246 * tree.c (build_common_tree_nodes): Update accordingly.
15248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15249 Alan Hayward <alan.hayward@arm.com>
15250 David Sherwood <david.sherwood@arm.com>
15252 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
15253 (mode_iterator::iterate_p): Likewise.
15254 (mode_iterator::get_wider): Likewise.
15255 * expr.c (init_expr_target): Use opt_scalar_float_mode.
15257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15258 Alan Hayward <alan.hayward@arm.com>
15259 David Sherwood <david.sherwood@arm.com>
15261 * coretypes.h (opt_scalar_float_mode): New typedef.
15262 * machmode.h (float_mode_for_size): New function.
15263 * emit-rtl.c (double_mode): Delete.
15264 (init_emit_once): Use float_mode_for_size.
15265 * stor-layout.c (layout_type): Likewise.
15266 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
15268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15269 Alan Hayward <alan.hayward@arm.com>
15270 David Sherwood <david.sherwood@arm.com>
15272 * output.h (assemble_real): Take a scalar_float_mode.
15273 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
15274 * config/arm/arm.md (consttable_4): Likewise.
15275 (consttable_8): Likewise.
15276 (consttable_16): Likewise.
15277 * config/mips/mips.md (consttable_float): Likewise.
15278 * config/s390/s390.c (s390_output_pool_entry): Likewise.
15279 * varasm.c (assemble_real): Take a scalar_float_mode.
15280 (output_constant_pool_2): Update accordingly.
15281 (output_constant): Likewise.
15283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15284 Alan Hayward <alan.hayward@arm.com>
15285 David Sherwood <david.sherwood@arm.com>
15287 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
15288 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
15289 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
15290 (native_encode_real): Likewise.
15291 (native_interpret_real): Likewise.
15292 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
15293 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15295 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15296 Alan Hayward <alan.hayward@arm.com>
15297 David Sherwood <david.sherwood@arm.com>
15299 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
15300 <scalar_float_mode>. Simplify.
15301 (gen_extend_conv_libfunc): Likewise.
15303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15304 Alan Hayward <alan.hayward@arm.com>
15305 David Sherwood <david.sherwood@arm.com>
15307 * coretypes.h (scalar_float_mode): New type.
15308 * machmode.h (mode_traits::from_int): Use machine_mode if
15309 USE_ENUM_MODES is defined.
15310 (is_a): New function.
15312 (dyn_cast): Likewise.
15313 (scalar_float_mode): New class.
15314 (scalar_float_mode::includes_p): New function.
15315 (is_float_mode): Likewise.
15316 * gdbhooks.py (MachineModePrinter): New class.
15317 (build_pretty_printer): Use it for scalar_float_mode.
15318 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
15319 (format_helper::format_helper): Turn into a template.
15320 * genmodes.c (get_mode_class): New function.
15321 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
15322 or machine_mode if none.
15323 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
15324 as_a <scalar_float_mode>.
15325 * dwarf2out.c (mem_loc_descriptor): Likewise.
15326 (insert_float): Likewise.
15327 (add_const_value_attribute): Likewise.
15328 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15329 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
15330 (expand_unop): Update accordingly.
15331 (expand_abs_nojump): Likewise.
15332 (expand_copysign_absneg): Take a scalar_float_mode.
15333 (expand_copysign_bit): Likewise.
15334 (expand_copysign): Update accordingly.
15336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15337 Alan Hayward <alan.hayward@arm.com>
15338 David Sherwood <david.sherwood@arm.com>
15340 * coretypes.h (opt_mode): New class.
15341 * machmode.h (opt_mode): Likewise.
15342 (opt_mode::else_void): New function.
15343 (opt_mode::require): Likewise.
15344 (opt_mode::exists): Likewise.
15345 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
15346 (GET_MODE_2XWIDER_MODE): Likewise.
15347 (mode_iterator::get_wider): Update accordingly.
15348 (mode_iterator::get_2xwider): Likewise.
15349 (mode_iterator::get_known_wider): Likewise, turning into a template.
15350 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
15351 forcing a wider mode to exist.
15352 * config/cr16/cr16.h (LONG_REG_P): Likewise.
15353 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15354 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
15355 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
15356 * lower-subreg.c (init_lower_subreg): Likewise.
15357 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
15358 on the final iteration.
15359 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
15360 a wider mode exists before asking for a move pattern.
15361 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
15362 forcing a wider mode to exist.
15363 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
15364 returning false if no such mode exists.
15365 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
15366 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
15367 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
15368 Avoid checking for a MODE_INT if we already know the mode is not a
15370 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
15371 forcing a wider mode to exist.
15372 (expmed_mult_highpart_optab): Likewise.
15373 (expmed_mult_highpart): Likewise.
15374 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
15376 * lto-streamer-in.c (lto_input_mode_table): Likewise.
15377 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
15378 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
15379 * internal-fn.c (expand_mul_overflow): Update use of
15380 GET_MODE_2XWIDER_MODE.
15381 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15382 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
15383 GET_MODE_WIDER_MODE.
15384 (convert_plusminus_to_widen): Likewise.
15385 * tree-switch-conversion.c (array_value_type): Likewise.
15386 * var-tracking.c (emit_note_insn_var_location): Likewise.
15387 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
15388 Return false inside rather than outside the loop if no wider mode
15390 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
15391 and GET_MODE_2XWIDER_MODE
15392 (can_compare_p): Use else_void.
15393 * gdbhooks.py (OptMachineModePrinter): New class.
15394 (build_pretty_printer): Use it for opt_mode.
15396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15397 Alan Hayward <alan.hayward@arm.com>
15398 David Sherwood <david.sherwood@arm.com>
15400 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
15401 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
15403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15404 Alan Hayward <alan.hayward@arm.com>
15405 David Sherwood <david.sherwood@arm.com>
15407 * machmode.h (mode_traits): New structure.
15408 (get_narrowest_mode): New function.
15409 (mode_iterator::start): Likewise.
15410 (mode_iterator::iterate_p): Likewise.
15411 (mode_iterator::get_wider): Likewise.
15412 (mode_iterator::get_known_wider): Likewise.
15413 (mode_iterator::get_2xwider): Likewise.
15414 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
15415 (FOR_EACH_MODE): Likewise.
15416 (FOR_EACH_MODE_FROM): Likewise.
15417 (FOR_EACH_MODE_UNTIL): Likewise.
15418 (FOR_EACH_WIDER_MODE): Likewise.
15419 (FOR_EACH_2XWIDER_MODE): Likewise.
15420 * builtins.c (expand_builtin_strlen): Use new mode iterators.
15421 * combine.c (simplify_comparison): Likewise
15422 * config/i386/i386.c (type_natural_mode): Likewise.
15423 * cse.c (cse_insn): Likewise.
15424 * dse.c (find_shift_sequence): Likewise.
15425 * emit-rtl.c (init_derived_machine_modes): Likewise.
15426 (init_emit_once): Likewise.
15427 * explow.c (hard_function_value): Likewise.
15428 * expmed.c (extract_fixed_bit_field_1): Likewise.
15429 (extract_bit_field_1): Likewise.
15430 (expand_divmod): Likewise.
15431 (emit_store_flag_1): Likewise.
15432 * expr.c (init_expr_target): Likewise.
15433 (convert_move): Likewise.
15434 (alignment_for_piecewise_move): Likewise.
15435 (widest_int_mode_for_size): Likewise.
15436 (emit_block_move_via_movmem): Likewise.
15437 (copy_blkmode_to_reg): Likewise.
15438 (set_storage_via_setmem): Likewise.
15439 (compress_float_constant): Likewise.
15440 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15441 * optabs-query.c (get_best_extraction_insn): Likewise.
15442 * optabs.c (expand_binop): Likewise.
15443 (expand_twoval_unop): Likewise.
15444 (expand_twoval_binop): Likewise.
15445 (widen_leading): Likewise.
15446 (widen_bswap): Likewise.
15447 (expand_parity): Likewise.
15448 (expand_unop): Likewise.
15449 (prepare_cmp_insn): Likewise.
15450 (prepare_float_lib_cmp): Likewise.
15451 (expand_float): Likewise.
15452 (expand_fix): Likewise.
15453 (expand_sfix_optab): Likewise.
15454 * postreload.c (move2add_use_add2_insn): Likewise.
15455 * reg-stack.c (reg_to_stack): Likewise.
15456 * reginfo.c (choose_hard_reg_mode): Likewise.
15457 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
15458 * stor-layout.c (mode_for_size): Likewise.
15459 (smallest_mode_for_size): Likewise.
15460 (mode_for_vector): Likewise.
15461 (finish_bitfield_representative): Likewise.
15462 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
15463 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
15464 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15465 * var-tracking.c (prepare_call_arguments): Likewise.
15467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15468 Alan Hayward <alan.hayward@arm.com>
15469 David Sherwood <david.sherwood@arm.com>
15471 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
15472 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
15473 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
15474 * machmode.h (mode_size): Move earlier in file.
15475 (mode_precision): Likewise.
15476 (mode_inner): Likewise.
15477 (mode_nunits): Likewise.
15478 (mode_unit_size): Likewise.
15479 (unit_unit_precision): Likewise.
15480 (mode_wider): Likewise.
15481 (mode_2xwider): Likewise.
15482 (machine_mode): New class.
15483 (mode_to_bytes): New function.
15484 (mode_to_bits): Likewise.
15485 (mode_to_precision): Likewise.
15486 (mode_to_inner): Likewise.
15487 (mode_to_unit_size): Likewise.
15488 (mode_to_unit_precision): Likewise.
15489 (mode_to_nunits): Likewise.
15490 (GET_MODE_SIZE): Use mode_to_bytes.
15491 (GET_MODE_BITSIZE): Use mode_to_bits.
15492 (GET_MODE_PRECISION): Use mode_to_precision.
15493 (GET_MODE_INNER): Use mode_to_inner.
15494 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
15495 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
15496 (GET_MODE_NUNITS): Use mode_to_nunits.
15497 * system.h (ALWAYS_INLINE): New macro.
15498 * config/powerpcspe/powerpcspe-c.c
15499 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
15500 int for arg1_mode and arg2_mode.
15502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15503 Alan Hayward <alan.hayward@arm.com>
15504 David Sherwood <david.sherwood@arm.com>
15506 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15507 Prefix mode names with E_ in case statements.
15508 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15509 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
15510 (aarch64_split_simd_move): Likewise.
15511 (aarch64_gen_storewb_pair): Likewise.
15512 (aarch64_gen_loadwb_pair): Likewise.
15513 (aarch64_gen_store_pair): Likewise.
15514 (aarch64_gen_load_pair): Likewise.
15515 (aarch64_get_condition_code_1): Likewise.
15516 (aarch64_constant_pool_reload_icode): Likewise.
15517 (get_rsqrte_type): Likewise.
15518 (get_rsqrts_type): Likewise.
15519 (get_recpe_type): Likewise.
15520 (get_recps_type): Likewise.
15521 (aarch64_gimplify_va_arg_expr): Likewise.
15522 (aarch64_simd_container_mode): Likewise.
15523 (aarch64_emit_load_exclusive): Likewise.
15524 (aarch64_emit_store_exclusive): Likewise.
15525 (aarch64_expand_compare_and_swap): Likewise.
15526 (aarch64_gen_atomic_cas): Likewise.
15527 (aarch64_emit_bic): Likewise.
15528 (aarch64_emit_atomic_swap): Likewise.
15529 (aarch64_emit_atomic_load_op): Likewise.
15530 (aarch64_evpc_trn): Likewise.
15531 (aarch64_evpc_uzp): Likewise.
15532 (aarch64_evpc_zip): Likewise.
15533 (aarch64_evpc_ext): Likewise.
15534 (aarch64_evpc_rev): Likewise.
15535 (aarch64_evpc_dup): Likewise.
15536 (aarch64_gen_ccmp_first): Likewise.
15537 (aarch64_gen_ccmp_next): Likewise.
15538 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15539 (alpha_emit_xfloating_libcall): Likewise.
15540 (emit_insxl): Likewise.
15541 (alpha_arg_type): Likewise.
15542 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
15543 (arc_preferred_simd_mode): Likewise.
15544 (arc_secondary_reload): Likewise.
15545 (get_arc_condition_code): Likewise.
15546 (arc_print_operand): Likewise.
15547 (arc_legitimate_constant_p): Likewise.
15548 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15549 * config/arc/arc.md (casesi_load): Likewise.
15550 (casesi_compact_jump): Likewise.
15551 * config/arc/predicates.md (proper_comparison_operator): Likewise.
15552 (cc_use_register): Likewise.
15553 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15554 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
15555 (arm_init_iwmmxt_builtins): Likewise.
15556 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
15557 (neon_expand_vector_init): Likewise.
15558 (arm_attr_length_move_neon): Likewise.
15559 (maybe_get_arm_condition_code): Likewise.
15560 (arm_emit_vector_const): Likewise.
15561 (arm_preferred_simd_mode): Likewise.
15562 (arm_output_iwmmxt_tinsr): Likewise.
15563 (thumb1_output_casesi): Likewise.
15564 (thumb2_output_casesi): Likewise.
15565 (arm_emit_load_exclusive): Likewise.
15566 (arm_emit_store_exclusive): Likewise.
15567 (arm_expand_compare_and_swap): Likewise.
15568 (arm_evpc_neon_vuzp): Likewise.
15569 (arm_evpc_neon_vzip): Likewise.
15570 (arm_evpc_neon_vrev): Likewise.
15571 (arm_evpc_neon_vtrn): Likewise.
15572 (arm_evpc_neon_vext): Likewise.
15573 (arm_validize_comparison): Likewise.
15574 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
15575 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
15576 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
15577 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
15578 (c6x_preferred_simd_mode): Likewise.
15579 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
15580 (epiphany_rtx_costs): Likewise.
15581 * config/epiphany/predicates.md (proper_comparison_operator):
15583 * config/frv/frv.c (condexec_memory_operand): Likewise.
15584 (frv_emit_move): Likewise.
15585 (output_move_single): Likewise.
15586 (output_condmove_single): Likewise.
15587 (frv_hard_regno_mode_ok): Likewise.
15588 (frv_matching_accg_mode): Likewise.
15589 * config/h8300/h8300.c (split_adds_subs): Likewise.
15590 (h8300_rtx_costs): Likewise.
15591 (h8300_print_operand): Likewise.
15592 (compute_mov_length): Likewise.
15593 (output_logical_op): Likewise.
15594 (compute_logical_op_length): Likewise.
15595 (compute_logical_op_cc): Likewise.
15596 (h8300_shift_needs_scratch_p): Likewise.
15597 (output_a_shift): Likewise.
15598 (compute_a_shift_length): Likewise.
15599 (compute_a_shift_cc): Likewise.
15600 (expand_a_rotate): Likewise.
15601 (output_a_rotate): Likewise.
15602 * config/i386/i386.c (classify_argument): Likewise.
15603 (function_arg_advance_32): Likewise.
15604 (function_arg_32): Likewise.
15605 (function_arg_64): Likewise.
15606 (function_value_64): Likewise.
15607 (ix86_gimplify_va_arg): Likewise.
15608 (ix86_legitimate_constant_p): Likewise.
15609 (put_condition_code): Likewise.
15610 (split_double_mode): Likewise.
15611 (ix86_avx256_split_vector_move_misalign): Likewise.
15612 (ix86_expand_vector_logical_operator): Likewise.
15613 (ix86_split_idivmod): Likewise.
15614 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
15615 (ix86_build_const_vector): Likewise.
15616 (ix86_build_signbit_mask): Likewise.
15617 (ix86_match_ccmode): Likewise.
15618 (ix86_cc_modes_compatible): Likewise.
15619 (ix86_expand_branch): Likewise.
15620 (ix86_expand_sse_cmp): Likewise.
15621 (ix86_expand_sse_movcc): Likewise.
15622 (ix86_expand_int_sse_cmp): Likewise.
15623 (ix86_expand_vec_perm_vpermi2): Likewise.
15624 (ix86_expand_vec_perm): Likewise.
15625 (ix86_expand_sse_unpack): Likewise.
15626 (ix86_expand_int_addcc): Likewise.
15627 (ix86_split_to_parts): Likewise.
15628 (ix86_vectorize_builtin_gather): Likewise.
15629 (ix86_vectorize_builtin_scatter): Likewise.
15630 (avx_vpermilp_parallel): Likewise.
15631 (inline_memory_move_cost): Likewise.
15632 (ix86_tieable_integer_mode_p): Likewise.
15633 (x86_maybe_negate_const_int): Likewise.
15634 (ix86_expand_vector_init_duplicate): Likewise.
15635 (ix86_expand_vector_init_one_nonzero): Likewise.
15636 (ix86_expand_vector_init_one_var): Likewise.
15637 (ix86_expand_vector_init_concat): Likewise.
15638 (ix86_expand_vector_init_interleave): Likewise.
15639 (ix86_expand_vector_init_general): Likewise.
15640 (ix86_expand_vector_set): Likewise.
15641 (ix86_expand_vector_extract): Likewise.
15642 (emit_reduc_half): Likewise.
15643 (ix86_emit_i387_round): Likewise.
15644 (ix86_mangle_type): Likewise.
15645 (ix86_expand_round_sse4): Likewise.
15646 (expand_vec_perm_blend): Likewise.
15647 (canonicalize_vector_int_perm): Likewise.
15648 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
15649 (expand_vec_perm_1): Likewise.
15650 (expand_vec_perm_interleave3): Likewise.
15651 (expand_vec_perm_even_odd_pack): Likewise.
15652 (expand_vec_perm_even_odd_1): Likewise.
15653 (expand_vec_perm_broadcast_1): Likewise.
15654 (ix86_vectorize_vec_perm_const_ok): Likewise.
15655 (ix86_expand_vecop_qihi): Likewise.
15656 (ix86_expand_mul_widen_hilo): Likewise.
15657 (ix86_expand_sse2_abs): Likewise.
15658 (ix86_expand_pextr): Likewise.
15659 (ix86_expand_pinsr): Likewise.
15660 (ix86_preferred_simd_mode): Likewise.
15661 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
15662 * config/i386/sse.md (*andnot<mode>3): Likewise.
15663 (<mask_codefor><code><mode>3<mask_name>): Likewise.
15664 (*<code><mode>3): Likewise.
15665 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
15666 (ia64_expand_atomic_op): Likewise.
15667 (ia64_arg_type): Likewise.
15668 (ia64_mode_to_int): Likewise.
15669 (ia64_scalar_mode_supported_p): Likewise.
15670 (ia64_vector_mode_supported_p): Likewise.
15671 (expand_vec_perm_broadcast): Likewise.
15672 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
15673 (iq2000_function_arg_advance): Likewise.
15674 (iq2000_function_arg): Likewise.
15675 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
15676 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
15677 (m68k_libcall_value): Likewise.
15678 (m68k_function_value): Likewise.
15679 (sched_attr_op_type): Likewise.
15680 * config/mcore/mcore.c (mcore_output_move): Likewise.
15681 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
15683 (microblaze_function_arg): Likewise.
15684 * config/mips/mips.c (mips16_build_call_stub): Likewise.
15685 (mips_print_operand): Likewise.
15686 (mips_mode_ok_for_mov_fmt_p): Likewise.
15687 (mips_vector_mode_supported_p): Likewise.
15688 (mips_preferred_simd_mode): Likewise.
15689 (mips_expand_vpc_loongson_even_odd): Likewise.
15690 (mips_expand_vec_unpack): Likewise.
15691 (mips_expand_vi_broadcast): Likewise.
15692 (mips_expand_vector_init): Likewise.
15693 (mips_expand_vec_reduc): Likewise.
15694 (mips_expand_msa_cmp): Likewise.
15695 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
15696 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
15697 (cc_flags_for_mode): Likewise.
15698 * config/msp430/msp430.c (msp430_print_operand): Likewise.
15699 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
15700 (nds32_output_casesi_pc_relative): Likewise.
15701 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15702 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
15703 (nvptx_gen_unpack): Likewise.
15704 (nvptx_gen_pack): Likewise.
15705 (nvptx_gen_shuffle): Likewise.
15706 (nvptx_gen_wcast): Likewise.
15707 (nvptx_preferred_simd_mode): Likewise.
15708 * config/pa/pa.c (pa_secondary_reload): Likewise.
15709 * config/pa/predicates.md (base14_operand): Likewise.
15710 * config/powerpcspe/powerpcspe-c.c
15711 (altivec_resolve_overloaded_builtin): Likewise.
15712 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
15714 (rs6000_preferred_simd_mode): Likewise.
15715 (output_vec_const_move): Likewise.
15716 (rs6000_expand_vector_extract): Likewise.
15717 (rs6000_split_vec_extract_var): Likewise.
15718 (reg_offset_addressing_ok_p): Likewise.
15719 (rs6000_legitimate_offset_address_p): Likewise.
15720 (rs6000_legitimize_address): Likewise.
15721 (rs6000_emit_set_const): Likewise.
15722 (rs6000_const_vec): Likewise.
15723 (rs6000_emit_move): Likewise.
15724 (spe_build_register_parallel): Likewise.
15725 (rs6000_darwin64_record_arg_recurse): Likewise.
15726 (swap_selector_for_mode): Likewise.
15727 (spe_init_builtins): Likewise.
15728 (paired_init_builtins): Likewise.
15729 (altivec_init_builtins): Likewise.
15730 (do_load_for_compare): Likewise.
15731 (rs6000_generate_compare): Likewise.
15732 (rs6000_expand_float128_convert): Likewise.
15733 (emit_load_locked): Likewise.
15734 (emit_store_conditional): Likewise.
15735 (rs6000_output_function_epilogue): Likewise.
15736 (rs6000_handle_altivec_attribute): Likewise.
15737 (rs6000_function_value): Likewise.
15738 (emit_fusion_gpr_load): Likewise.
15739 (emit_fusion_p9_load): Likewise.
15740 (emit_fusion_p9_store): Likewise.
15741 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
15742 (fusion_gpr_mem_load): Likewise.
15743 (fusion_addis_mem_combo_load): Likewise.
15744 (fusion_addis_mem_combo_store): Likewise.
15745 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
15746 (fusion_gpr_mem_load): Likewise.
15747 (fusion_addis_mem_combo_load): Likewise.
15748 (fusion_addis_mem_combo_store): Likewise.
15749 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15751 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
15752 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15753 (rs6000_preferred_simd_mode): Likewise.
15754 (output_vec_const_move): Likewise.
15755 (rs6000_expand_vector_extract): Likewise.
15756 (rs6000_split_vec_extract_var): Likewise.
15757 (reg_offset_addressing_ok_p): Likewise.
15758 (rs6000_legitimate_offset_address_p): Likewise.
15759 (rs6000_legitimize_address): Likewise.
15760 (rs6000_emit_set_const): Likewise.
15761 (rs6000_const_vec): Likewise.
15762 (rs6000_emit_move): Likewise.
15763 (rs6000_darwin64_record_arg_recurse): Likewise.
15764 (swap_selector_for_mode): Likewise.
15765 (paired_init_builtins): Likewise.
15766 (altivec_init_builtins): Likewise.
15767 (rs6000_expand_float128_convert): Likewise.
15768 (emit_load_locked): Likewise.
15769 (emit_store_conditional): Likewise.
15770 (rs6000_output_function_epilogue): Likewise.
15771 (rs6000_handle_altivec_attribute): Likewise.
15772 (rs6000_function_value): Likewise.
15773 (emit_fusion_gpr_load): Likewise.
15774 (emit_fusion_p9_load): Likewise.
15775 (emit_fusion_p9_store): Likewise.
15776 * config/rx/rx.c (rx_gen_move_template): Likewise.
15777 (flags_from_mode): Likewise.
15778 * config/s390/predicates.md (s390_alc_comparison): Likewise.
15779 (s390_slb_comparison): Likewise.
15780 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
15781 (s390_vector_mode_supported_p): Likewise.
15782 (s390_cc_modes_compatible): Likewise.
15783 (s390_match_ccmode_set): Likewise.
15784 (s390_canonicalize_comparison): Likewise.
15785 (s390_emit_compare_and_swap): Likewise.
15786 (s390_branch_condition_mask): Likewise.
15787 (s390_rtx_costs): Likewise.
15788 (s390_secondary_reload): Likewise.
15789 (__SECONDARY_RELOAD_CASE): Likewise.
15790 (s390_expand_cs): Likewise.
15791 (s390_preferred_simd_mode): Likewise.
15792 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
15793 * config/sh/sh.c (sh_print_operand): Likewise.
15794 (dump_table): Likewise.
15795 (sh_secondary_reload): Likewise.
15796 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15797 * config/sh/sh.md (casesi_worker_1): Likewise.
15798 (casesi_worker_2): Likewise.
15799 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
15800 (fcc_comparison_operator): Likewise.
15801 * config/sparc/sparc.c (sparc_expand_move): Likewise.
15802 (emit_soft_tfmode_cvt): Likewise.
15803 (sparc_preferred_simd_mode): Likewise.
15804 (output_cbranch): Likewise.
15805 (sparc_print_operand): Likewise.
15806 (sparc_expand_vec_perm_bmask): Likewise.
15807 (vector_init_bshuffle): Likewise.
15808 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15809 (spu_vector_mode_supported_p): Likewise.
15810 (spu_expand_insv): Likewise.
15811 (spu_emit_branch_or_set): Likewise.
15812 (spu_handle_vector_attribute): Likewise.
15813 (spu_builtin_splats): Likewise.
15814 (spu_builtin_extract): Likewise.
15815 (spu_builtin_promote): Likewise.
15816 (spu_expand_sign_extend): Likewise.
15817 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15818 (tilegx_simd_int): Likewise.
15819 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
15820 (tilepro_simd_int): Likewise.
15821 * config/v850/v850.c (const_double_split): Likewise.
15822 (v850_print_operand): Likewise.
15823 (ep_memory_offset): Likewise.
15824 * config/vax/vax.c (vax_rtx_costs): Likewise.
15825 (vax_output_int_move): Likewise.
15826 (vax_output_int_add): Likewise.
15827 (vax_output_int_subtract): Likewise.
15828 * config/visium/predicates.md (visium_branch_operator): Likewise.
15829 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
15830 (visium_print_operand_address): Likewise.
15831 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
15832 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
15833 (xtensa_expand_conditional_branch): Likewise.
15834 (xtensa_copy_incoming_a7): Likewise.
15835 (xtensa_output_literal): Likewise.
15836 * dfp.c (decimal_real_maxval): Likewise.
15837 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
15839 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15840 Alan Hayward <alan.hayward@arm.com>
15841 David Sherwood <david.sherwood@arm.com>
15843 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
15844 (mode_nunits_inline): Likewise.
15845 (mode_inner_inline): Likewise.
15846 (mode_unit_size_inline): Likewise.
15847 (mode_unit_precision_inline): Likewise.
15848 (emit_insn_modes_h): Likewise. Also emit a #define of the
15850 (emit_mode_wider): Add an E_ prefix to mode names.
15851 (emit_mode_complex): Likewise.
15852 (emit_mode_inner): Likewise.
15853 (emit_mode_adjustments): Likewise.
15854 (emit_mode_int_n): Likewise.
15855 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
15856 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
15857 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
15858 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
15859 (CRC32_BUILTIN, ENTRY): Likewise.
15860 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
15861 (aarch64_pop_regs): Likewise.
15862 (aarch64_process_components): Likewise.
15863 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
15864 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
15865 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
15866 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
15867 * config/arm/arm.c (arm_init_libfuncs): Likewise.
15868 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
15870 * config/i386/i386-builtin.def (pcmpestr): Likewise.
15871 (pcmpistr): Likewise.
15872 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
15873 * config/mmix/mmix.c (mmix_output_condition): Likewise.
15874 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15876 * config/rl78/rl78.c (mduc_regs): Likewise.
15877 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15878 (htm_expand_builtin): Likewise.
15879 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
15880 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
15881 * config/xtensa/xtensa.c (print_operand): Likewise.
15882 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
15883 (NUM_MODE_VECTOR_INT): Likewise.
15884 * genoutput.c (null_operand): Likewise.
15885 (output_operand_data): Likewise.
15886 * genrecog.c (print_parameter_value): Likewise.
15887 * lra.c (debug_operand_data): Likewise.
15889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15891 * dwarf2out.c (scompare_loc_descriptor_wide)
15892 (scompare_loc_descriptor_narrow): New functions, split out from...
15893 (scompare_loc_descriptor): ...here.
15894 * expmed.c (emit_store_flag_int): New function, split out from...
15895 (emit_store_flag): ...here.
15897 2017-08-30 Richard Biener <rguenther@suse.de>
15899 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
15900 (dwarf2out_early_finish): Move setting of AT_pubnames from
15901 early debug output to early finish.
15903 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15905 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
15906 and -mdata-region to the assembler.
15908 2017-08-30 Richard Biener <rguenther@suse.de>
15910 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
15912 (gen_subprogram_die): Add DW_AT_object_pointer only early.
15913 (dwarf2out_early_global_decl): Only generate a DIE for the
15914 abstract origin if it doesn't already exist or is a declaration DIE.
15915 (resolve_addr): Do not add the linkage name twice when
15916 generating a stub DIE for the DW_TAG_GNU_call_site target.
15918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15920 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15921 Use machine_mode rather than int for arg1_mode.
15923 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
15926 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
15927 that the second argument of the built-in functions to unpack
15928 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
15929 switch statement instead a lot of if statements.
15930 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
15931 Allow 64-bit values to be in Altivec registers as well as
15932 traditional floating point registers.
15933 (pack<mode>, FMOVE128_VSX iterator): Likewise.
15935 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
15937 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
15938 MAX_REGS_PER_ADDRESS == 1.
15940 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
15942 * config/i386/i386.opt (flag_fentry): Do not init to -1.
15943 * config/i386/i386.c (ix86_option_override_internal): Simplify
15944 setting of opts->x_flag_entry.
15946 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15947 Jakub Jelinek <jakub@redhat.com>
15948 Richard Biener <rguenther@suse.de>
15950 PR tree-optimization/81503
15951 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
15952 folded constant fits in the target type; reorder tests for clarity.
15954 2017-08-29 Martin Liska <mliska@suse.cz>
15956 * passes.def: Include pass_lower_switch.
15957 * stmt.c (dump_case_nodes): Remove and move to
15958 tree-switch-conversion.
15959 (case_values_threshold): Likewise.
15960 (expand_switch_as_decision_tree_p): Likewise.
15961 (emit_case_decision_tree): Likewise.
15962 (expand_case): Likewise.
15963 (balance_case_nodes): Likewise.
15964 (node_has_low_bound): Likewise.
15965 (node_has_high_bound): Likewise.
15966 (node_is_bounded): Likewise.
15967 (emit_case_nodes): Likewise.
15968 (struct simple_case_node): New struct.
15969 (add_case_node): Remove.
15970 (emit_case_dispatch_table): Use vector instead of case_list.
15971 (reset_out_edges_aux): Remove.
15972 (compute_cases_per_edge): Likewise.
15973 (expand_case): Build list of simple_case_node.
15974 (expand_sjlj_dispatch_table): Use it.
15975 * tree-switch-conversion.c (struct case_node): Moved from
15976 stmt.c and adjusted.
15977 (emit_case_nodes): Likewise.
15978 (node_has_low_bound): Likewise.
15979 (node_has_high_bound): Likewise.
15980 (node_is_bounded): Likewise.
15981 (case_values_threshold): Likewise.
15982 (reset_out_edges_aux): Likewise.
15983 (compute_cases_per_edge): Likewise.
15984 (add_case_node): Likewise.
15985 (dump_case_nodes): Likewise.
15986 (balance_case_nodes): Likewise.
15987 (expand_switch_as_decision_tree_p): Likewise.
15988 (emit_jump): Likewise.
15989 (emit_case_decision_tree): Likewise.
15990 (try_switch_expansion): Likewise.
15991 (do_jump_if_equal): Likewise.
15992 (emit_cmp_and_jump_insns): Likewise.
15993 (fix_phi_operands_for_edge): New function.
15994 (record_phi_operand_mapping): Likewise.
15995 (class pass_lower_switch): New pass.
15996 (pass_lower_switch::execute): New function.
15997 (make_pass_lower_switch): Likewise.
15998 (conditional_probability):
15999 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
16000 * tree-pass.h: Add make_pass_lower_switch.
16002 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16005 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
16008 2017-08-29 Richard Biener <rguenther@suse.de>
16010 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
16011 we do not add a DW_AT_inline attribute twice.
16012 (gen_subprogram_die): Remove code setting DW_AT_inline on
16013 DECL_ABSTRACT_P nodes.
16015 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
16017 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
16018 calls to internal functions.
16019 (gimplify_modify_expr): Likewise.
16020 * tree-call-cdce.c (use_internal_fn): Likewise.
16021 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
16022 (convert_to_divmod): Set the nothrow flag.
16023 * tree-if-conv.c (predicate_mem_writes): Likewise.
16024 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16025 (vectorizable_call): Likewise.
16026 (vectorizable_store): Likewise.
16027 (vectorizable_load): Likewise.
16028 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
16029 (vect_recog_mask_conversion_pattern): Likewise.
16031 2017-08-29 Martin Liska <mliska@suse.cz>
16034 * gcc.c (driver_handle_option): Add new argument.
16035 * opts-common.c (handle_option): Pass
16036 target_option_override_hook.
16037 * opts-global.c (lang_handle_option): Add new option.
16038 (set_default_handlers): Add new argument.
16039 (decode_options): Likewise.
16040 * opts.c (target_handle_option): Likewise.
16041 (common_handle_option): Call target_option_override_hook.
16042 * opts.h (struct cl_option_handler_func): Add hook for
16043 target option override.
16044 (struct cl_option_handlers): Likewise.
16045 (set_default_handlers): Add new argument.
16046 (decode_options): Likewise.
16047 (common_handle_option): Likewise.
16048 (target_handle_option): Likewise.
16049 * toplev.c (toplev::main): Pass targetm.target_option.override
16052 2017-08-29 Richard Biener <rguenther@suse.de>
16053 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
16055 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
16056 life to the active subtree.
16058 2017-08-28 Jeff Law <law@redhat.com>
16060 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
16061 derive_equivalences.
16062 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
16063 Code moved into....
16064 (edge_info::derive_equivalences): New private member function
16066 * tree-ssa-dom.c (class edge_info): Changed from a struct
16067 to a class. Add ctor/dtor, methods and data members.
16068 (edge_info::edge_info): Renamed from allocate_edge_info.
16069 Initialize additional members.
16070 (edge_info::~edge_info): New.
16071 (free_dom_edge_info): Delete the edge info.
16072 (record_edge_info): Use new class & associated member functions.
16073 Tighten forms for testing for edge equivalences.
16074 (record_temporary_equivalences): Iterate over the simple
16075 equivalences rather than assuming there's only one per edge.
16076 (cprop_into_successor_phis): Iterate over the simple
16077 equivalences rather than assuming there's only one per edge.
16078 (optimize_stmt): Use operand_equal_p rather than pointer
16079 equality for mini-DSE code.
16081 2017-08-28 Nathan Sidwell <nathan@acm.org>
16083 * gcc.c (execute): Fold SIGPIPE handling into switch
16084 statement. Adjust internal error message.
16086 2017-08-28 Richard Biener <rguenther@suse.de>
16089 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
16090 Do nothing for removed DIEs.
16092 2017-08-28 Richard Biener <rguenther@suse.de>
16094 PR tree-optimization/81977
16095 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
16098 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
16101 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
16102 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
16103 using targetm.gen_mem_thread_fence.
16105 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
16108 * config/i386/i386.md (*<btsc><mode>): Change operand 2
16109 predicate to register_operand. Reorder operands.
16110 (*btr<mode>): Ditto.
16111 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
16112 (*btr<mode>_mask): Ditto.
16114 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
16116 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
16117 * config/rs6000/xmmintrin.h: New file.
16118 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
16120 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16123 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
16124 parameter and_insn and return it.
16125 (recombine_lvx_pattern): Insert a copy to ensure availability of
16126 the base register of the copied masking operation at the point of
16127 the instruction replacement.
16128 (recombine_stvx_pattern): Likewise.
16130 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
16132 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
16133 undocumented switches.
16134 (-mpower9-dform-vector): Likewise.
16135 (-mpower9-dform): Likewise.
16136 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
16137 comments to delete references to -mpower9-dform* switches.
16138 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
16139 Delete reference to -mpower9-dform* switches, test for
16140 -mpower9-vector instead.
16141 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
16142 (OTHER_P9_VECTOR_MASKS): Likewise.
16143 (POWERPC_MASKS): Likewise.
16144 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
16145 tests against -mpower9-dform* to -mpower9-vector. Delete code
16146 that checked for -mpower9-dform* consistancy with other options.
16147 Add test for -mpower9-misc to enable other power9 switches.
16148 (rs6000_init_hard_regno_mode_ok): Likewise.
16149 (rs6000_option_override_internal): Likewise.
16150 (rs6000_emit_prologue): Likewise.
16151 (rs6000_emit_epilogue): Likewise.
16152 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
16153 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
16154 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
16156 (emit_fusion_p9_store): Likewise.
16157 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
16158 resetting these macros if the assembler does not support ISA 3.0
16160 (TARGET_P9_DFORM_VECTOR): Likewise.
16161 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
16162 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
16164 2017-08-25 Alan Modra <amodra@gmail.com>
16166 PR rtl-optimization/81747
16167 * cse.c (cse_extended_basic_block): Don't attempt to record
16168 equivalences for degenerate conditional branches that branch
16169 to their fall-through.
16171 2017-08-24 Martin Sebor <msebor@redhat.com>
16173 PR middle-end/81908
16174 * gimple-fold.c (size_must_be_zero_p): New function.
16175 (gimple_fold_builtin_memory_op): Call it.
16177 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
16179 * config/rs6000/mm_malloc.h: New file.
16181 2017-08-24 Bin Cheng <bin.cheng@arm.com>
16183 PR tree-optimization/81913
16184 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
16185 analysis when either IVs in condition can wrap.
16187 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
16189 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
16190 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
16192 2017-08-24 Richard Biener <rguenther@suse.de>
16195 * targhooks.c (default_target_can_inline_p): Properly
16196 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
16197 is present and always compare.
16198 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
16199 infer -mfpmath=sse from TARGET_SSE_P.
16200 (ix86_can_inline_p): Properly use target_option_default_node when
16201 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
16203 2017-08-24 Richard Biener <rguenther@suse.de>
16206 * dwarf2out.c (output_die): Handle flag_generate_offload like
16208 (output_comp_unit): Likewise.
16209 (gen_array_type_die): Likewise.
16210 (dwarf2out_early_finish): Likewise.
16211 (note_variable_value_in_expr): Likewise.
16212 (dwarf2out_finish): Likewise. Adjust assert.
16213 * cgraphunit.c (symbol_table::compile): Move setting of
16214 flag_generate_offload earlier ...
16215 (symbol_table::finalize_compilation_unit): ... here, before
16216 early debug finalization.
16218 2017-08-24 Richard Biener <rguenther@suse.de>
16220 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
16221 and ipa-fnsummary.h.
16222 (ix86_can_inline_p): When ix86_fpmath flags do not match
16223 check whether the callee uses FP math at all.
16225 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
16227 PR middle-end/81931
16228 * tree-ssanames.c (get_nonzero_bits): Use element_precision
16229 instead of TYPE_PRECISION.
16231 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
16232 Alan Hayward <alan.hayward@arm.com>
16233 David Sherwood <david.sherwood@arm.com>
16235 * combine.c (make_extraction): Use subreg_offset_from_lsb.
16237 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
16239 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
16241 (ix86_frame::stack_realign_allocate): New field.
16242 (struct machine_frame_state): Modify comments.
16243 (machine_frame_state::sp_realigned_fp_end): New field.
16244 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
16245 layout calculation.
16246 (sp_valid_at): Add assertion to assure no attempt to access invalid
16247 offset of a realigned stack.
16248 (fp_valid_at): Likewise.
16249 (choose_baseaddr): Modify comments.
16250 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
16251 ix86_expand_prologue.
16252 (ix86_expand_prologue): Modify stack realignment and allocation.
16253 (ix86_expand_epilogue): Modify comments.
16254 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
16255 avx2_runtime, avx512f, and avx512f_runtime.
16257 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
16259 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
16260 (mstackrealign): Do not init to -1.
16261 * config/i386/i386.c (ix86_option_override_internal):
16262 Check opts_set, not opts when setting default value of
16263 opts->x_ix86_force_align_arg_pointer.
16265 2017-08-23 Richard Biener <rguenther@suse.de>
16267 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
16268 lang_hooks.decl_printable_name.
16269 * print-rtl-function.c (print_rtx_function): Likewise.
16270 * tree-pretty-print.c (dump_function_header): Likewise.
16272 2017-08-23 Richard Biener <rguenther@suse.de>
16275 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
16276 -g0 at compile-time.
16278 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16280 PR middle-end/19706
16281 * doc/sourcebuild.texi (Other hardware attributes):
16284 2017-08-23 Tamar Christina <tamar.christina@arm.com>
16286 PR middle-end/19706
16287 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
16288 Fix single-use check.
16290 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
16292 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
16294 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16296 * doc/install.texi: Modify to add more details on running selected
16299 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
16301 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
16302 is combined with -mabi=ms.
16303 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
16306 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16308 PR tree-optimization/81488
16309 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
16310 and cached_basis fields.
16311 (MAX_SPREAD): New constant.
16312 (alloc_cand_and_find_basis): Initialize new fields.
16313 (clear_visited): New function.
16314 (create_phi_basis_1): Rename from create_phi_basis, set visited
16315 and cached_basis fields.
16316 (create_phi_basis): New wrapper function.
16317 (phi_add_costs_1): Rename from phi_add_costs, add spread
16318 parameter, set visited field, short-circuit when limits reached.
16319 (phi_add_costs): New wrapper function.
16320 (record_phi_increments_1): Rename from record_phi_increments, set
16322 (record_phi_increments): New wrapper function.
16323 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
16324 (phi_incr_cost): New wrapper function.
16325 (all_phi_incrs_profitable_1): Rename from
16326 all_phi_incrs_profitable, set visited field.
16327 (all_phi_incrs_profitable): New wrapper function.
16329 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16330 Alan Hayward <alan.hayward@arm.com>
16331 David Sherwood <david.sherwood@arm.com>
16333 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
16334 that takes the outer and inner modes.
16335 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
16336 comparison as the canonical test for a paradoxical subreg.
16337 * combine.c (simplify_set): Use paradoxical_subreg_p.
16338 (make_extraction): Likewise.
16339 (force_to_mode): Likewise.
16340 (rtx_equal_for_field_assignment_p): Likewise.
16341 (gen_lowpart_for_combine): Likewise.
16342 (simplify_comparison): Likewise.
16343 * cse.c (equiv_constant): Likewise.
16344 * expmed.c (store_bit_field_1): Likewise.
16345 * final.c (alter_subreg): Likewise.
16346 * fwprop.c (propagate_rtx): Likewise.
16347 (forward_propagate_subreg): Likewise.
16348 * ira-conflicts.c (ira_build_conflicts): Likewise.
16349 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
16350 * lra-constraints.c (curr_insn_transform): Likewise.
16351 (split_reg): Likewise.
16352 * lra-eliminations.c (move_plus_up): Likewise.
16353 (lra_eliminate_regs_1): Likewise.
16354 * recog.c (general_operand): Likewise.
16355 * ree.c (combine_reaching_defs): Likewise.
16356 * reload.c (push_reload): Likewise.
16357 (find_reloads): Likewise.
16358 * reload1.c (elimination_effects): Likewise.
16359 (compute_reload_subreg_offset): Likewise.
16360 (choose_reload_regs): Likewise.
16361 * rtlanal.c (subreg_lsb_1): Likewise.
16362 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16363 (simplify_subreg): Likewise.
16364 * var-tracking.c (track_loc_p): Likewise.
16365 * emit-rtl.c (byte_lowpart_offset): Likewise.
16366 (paradoxical_subreg_p): Delete out-of-line definition.
16368 2017-08-22 Jeff Law <law@redhat.com>
16370 PR tree-optimization/81741
16371 PR tree-optimization/71947
16372 * tree-ssa-dom.c: Include tree-inline.h.
16373 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
16374 equivalences if one is more expensive to compute than the other.
16375 * tree-ssa-scopedtables.h (class const_or_copies): Make
16376 record_const_or_copy_raw method private.
16377 (class avail_exprs_stack): New method simplify_binary_operation.
16378 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
16379 avail_exprs_stack::simplify_binary_operation as needed.
16380 (avail_exprs_stack::simplify_binary_operation): New function.
16382 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
16384 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
16385 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
16386 (DOT_SYMBOLS): Likewise.
16387 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
16388 (RELOCATABLE_NEEDS_FIXUP): Likewise.
16389 (RS6000_ABI_NAME): Likewise.
16390 (TARGET_CMODEL): Likewise.
16391 (TOC_SECTION_ASM_OP): Likewise.
16392 (SET_CMODEL): New macro.
16393 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
16395 2017-08-22 Richard Biener <rguenther@suse.de>
16397 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
16398 to allow for free-lang-data replacements similar to verify_type_variant.
16400 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
16401 Alan Hayward <alan.hayward@arm.com>
16402 David Sherwood <david.sherwood@arm.com>
16404 * config/aarch64/aarch64.md (casesi): Use DImode rather than
16405 VOIDmode for the LABEL_REF.
16407 2017-08-22 Richard Biener <rguenther@suse.de>
16409 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
16411 2017-08-22 Richard Biener <rguenther@suse.de>
16413 * common.opt (feliminate-dwarf2-dups): Ignore.
16414 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
16415 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
16416 same_die_p_wrap, compute_section_prefix,
16417 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
16418 (comdat_symbol_id, comdat_symbol_number): Likewise.
16419 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
16421 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
16422 (output_die): Mark unreachable path unreachable.
16423 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
16424 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
16425 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
16426 (dwarf2out_early_finish): Likewise.
16428 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
16430 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
16432 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
16435 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
16436 not VAR_P. Filter attribute warnings with OPT_Wattributes.
16437 (avr_attribute_table) <io, io_low, address>: Initialize
16438 .decl_required with true.
16440 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
16442 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
16443 undocumented debugging options.
16444 (-mvsx-scalar-double): Likewise.
16445 (-mallow-df-permute): Likewise.
16446 (-mvectorize-builtins): Likewise.
16447 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16448 (rs6000_builtin_vectorized_function): Likewise.
16449 (rs6000_builtin_md_vectorized_function): Likewise.
16450 (rs6000_opt_vars): Likewise.
16452 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
16455 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
16456 (*btrq_imm): Rename from *btrq.
16457 (*btcq_imm): Rename from *btcq.
16458 (btsc): New code attribute.
16459 (*<btsc><mode>): New insn pattern.
16460 (*btr<mode>): Ditto.
16461 (*<btsc><mode>_mask): New insn_and_split pattern.
16462 (*btr<mode>_mask): Ditto.
16464 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16465 Alan Hayward <alan.hayward@arm.com>
16466 David Sherwood <david.sherwood@arm.com>
16468 * function.c (pad_below): Simplify padding calculation.
16470 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16471 Alan Hayward <alan.hayward@arm.com>
16472 David Sherwood <david.sherwood@arm.com>
16474 * target.def (function_prologue): Remove frame size argument.
16475 (function_epilogue): Likewise.
16476 * doc/tm.texi: Regenerate.
16477 * final.c (final_start_function): Update call to function_prologue.
16478 (final_end_function): Update call to function_epilogue.
16479 (default_function_pro_epilogue): Remove frame size argument.
16480 * output.h (default_function_pro_epilogue): Likewise.
16481 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
16482 (arm_output_function_prologue): Likewise.
16483 * config/frv/frv.c (frv_function_prologue): Likewise.
16484 (frv_function_epilogue): Likewise.
16485 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
16486 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
16487 (ia64_output_function_epilogue): Likewise.
16488 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
16489 (m32r_output_function_epilogue): Likewise.
16490 * config/microblaze/microblaze.c (microblaze_function_prologue)
16491 (microblaze_function_epilogue): Likewise.
16492 * config/mips/mips.c (mips_output_function_prologue): Likewise.
16493 (mips_output_function_epilogue): Likewise.
16494 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
16495 (mmix_target_asm_function_epilogue): Likewise.
16496 * config/msp430/msp430.c (msp430_start_function): Likewise.
16497 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
16498 (nds32_asm_function_epilogue): Likewise.
16499 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
16500 * config/pa/pa.c (pa_output_function_prologue): Likewise.
16501 (pa_output_function_epilogue): Likewise.
16502 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
16503 (rs6000_output_function_epilogue): Likewise.
16504 * config/rl78/rl78.c (rl78_start_function): Likewise.
16505 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
16506 (rs6000_output_function_epilogue): Likewise.
16507 * config/rx/rx.c (rx_output_function_prologue): Likewise.
16508 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
16509 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
16510 (sparc_asm_function_epilogue): Likewise.
16512 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16514 * tree.h (type_has_mode_precision_p): New function.
16515 * convert.c (convert_to_integer_1): Use it.
16516 * expr.c (expand_expr_real_2): Likewise.
16517 (expand_expr_real_1): Likewise.
16518 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
16519 * match.pd: Likewise.
16520 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
16521 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
16522 * tree-tailcall.c (process_assignment): Likewise.
16523 * tree-vect-loop.c (vectorizable_reduction): Likewise.
16524 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
16525 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
16526 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16527 (vectorizable_assignment): Likewise.
16528 (vectorizable_shift): Likewise.
16529 (vectorizable_operation): Likewise.
16530 * tree-vrp.c (register_edge_assert_for_2): Likewise.
16532 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
16534 * match.pd: Add pow (C, x) simplification.
16536 2017-08-21 Richard Biener <rguenther@suse.de>
16538 PR tree-optimization/81900
16539 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
16540 for blocks with abnormal predecessors.
16541 (compute_antic): Do not set visited flag prematurely.
16543 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
16546 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
16548 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16550 * stor-layout.h (vector_type_mode): Move to...
16551 * tree.h (vector_type_mode): ...here.
16552 * stor-layout.c (vector_type_mode): Move to...
16553 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
16555 2017-08-21 Richard Biener <rguenther@suse.de>
16557 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
16558 register_external_die hooks.
16559 (debug_false_tree_charstarstar_uhwistar): Declare.
16560 (debug_nothing_tree_charstar_uhwi): Likewise.
16561 * debug.c (do_nothing_debug_hooks): Adjust.
16562 (debug_false_tree_charstarstar_uhwistar): New do nothing.
16563 (debug_nothing_tree_charstar_uhwi): Likewise.
16564 * dbxout.c (dbx_debug_hooks): Adjust.
16565 (xcoff_debug_hooks): Likewise.
16566 * sdbout.c (sdb_debug_hooks): Likewise.
16567 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16568 * dwarf2out.c (macinfo_label_base): New global.
16569 (dwarf2out_register_external_die): New function for the
16570 register_external_die hook.
16571 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
16572 (dwarf2_debug_hooks): Use them.
16573 (dwarf2_lineno_debug_hooks): Adjust.
16574 (struct die_struct): Add with_offset flag.
16575 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
16576 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
16577 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
16578 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
16579 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
16580 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
16581 defining section names for the early LTO debug variants.
16582 (reset_indirect_string): New helper.
16583 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
16584 (print_dw_val): Add support for offsetted symbol references.
16585 (get_ultimate_context): Split out from is_cxx.
16586 (is_cxx): Use get_ultimate_context.
16587 (is_fortran): Add decl overload.
16588 (compute_comp_unit_symbol): Split out worker from
16589 compute_section_prefix.
16590 (compute_section_prefix): Call compute_comp_unit_symbol and
16591 set comdat_type_p here.
16592 (output_die): Skip DIE symbol output for the LTO added one.
16593 Handle DIE symbol references with offset.
16594 (output_comp_unit): Guard section name mangling properly.
16595 For LTO debug sections emit a symbol at the section beginning
16596 which we use to refer to its DIEs.
16597 (add_abstract_origin_attribute): For DIEs registered via
16598 dwarf2out_register_external_die directly refer to the early
16599 DIE rather than indirectly through the shadow one we created.
16600 Remove obsolete call to dwarf2out_abstract_function for
16601 non-function/block origins.
16602 (gen_array_type_die): When generating early LTO debug do
16603 not emit DW_AT_string_length.
16604 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
16605 late when in LTO. As suggested place a gcc_unreachable for
16606 the DECL_ABSTRACT_P case.
16607 (gen_subprogram_die): Avoid another specification DIE
16608 for early built declarations/definitions for the late LTO case.
16609 (gen_variable_die): Add type references for late duplicated VLA dies
16611 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
16612 we have the abstract instance already.
16613 (process_scope_var): Adjust decl DIE contexts in LTO which
16614 first puts them in limbo.
16615 (gen_decl_die): Do not generate type DIEs late apart from
16616 types for VLAs or for decls we do not yet have a DIE. Do not
16617 call dwarf2out_abstract_function late.
16618 (dwarf2out_early_global_decl): Make sure to create DIEs
16619 for abstract instances of a decl first.
16620 (dwarf2out_late_global_decl): Adjust comment.
16621 (output_macinfo_op): With multiple macro sections use
16622 macinfo_label_base to distinguish labels.
16623 (output_macinfo): Likewise. Update macinfo_label_base.
16624 Pass in the line info label.
16625 (note_variable_value_in_expr): When generating LTO resolve
16626 all variable values here by generating DIEs as needed.
16627 (init_sections_and_labels): Add early LTO debug flag parameter
16628 and generate different sections and names if set. Add generation
16629 counter for the labels so we can have multiple of them.
16630 (reset_dies): Helper to allow DIEs to be output multiple times.
16631 (dwarf2out_finish): When outputting DIEs to the fat part of an
16632 LTO object first reset DIEs.
16633 (dwarf2out_early_finish): Output early DIEs when generating LTO.
16634 (modified_type_die): Check for decl_ultimate_origin being self
16636 (gen_type_die_with_usage): Likewise.
16637 (gen_typedef_die): Allow decl_ultimate_origin being self.
16638 (set_decl_abstract_flags): Remove.
16639 (set_block_abstract_flags): Likewise.
16640 (dwarf2out_abstract_function): Treat the early generated DIEs
16641 as the abstract copy and only add DW_AT_inline and
16642 DW_AT_artificial here and call set_decl_origin_self.
16643 If the DIE has an abstract origin don't do anything.
16644 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
16645 if we have none yet (Go fails to build one, PR78628).
16646 (variably_modified_type_p): Prevent endless recursion for Ada
16647 cyclic pointer types.
16648 * lto-streamer-in.c: Include debug.h.
16649 (dref_queue): New global.
16650 (lto_read_tree_1): Stream in DIE references.
16651 (lto_input_tree): Register DIE references.
16652 (input_function): Stream DECL_DEBUG_ARGS.
16653 * lto-streamer-out.c: Include debug.h.
16654 (lto_write_tree_1): Output DIE references.
16655 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
16656 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
16657 (output_function): Stream DECL_DEBUG_ARGS.
16658 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
16659 Stream DECL_ABSTRACT_ORIGIN.
16660 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
16661 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
16662 DECL_CONTEXT for file-scope decls.
16663 * lto-streamer.h (struct dref_entry): Declare.
16664 (dref_queue): Likewise.
16665 * cfgexpand.c (pass_expand::execute): Do not call the
16666 outlining_inline_function hook here.
16667 * lto-wrapper.c (debug_obj): New global.
16668 (tool_cleanup): Unlink it if required.
16669 (debug_objcopy): New function.
16670 (run_gcc): Handle early debug sections in the IL files by
16671 extracting them to separate files, partially linkin them and
16672 feeding the result back as result to the linker.
16673 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
16674 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
16675 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
16676 sections into a separate segment.
16677 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
16679 (darwin_asm_dwarf_section): Likewise.
16680 (darwin_asm_output_dwarf_offset): Likewise.
16681 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
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 * read-md.h (md_reader::record_potential_iterator_use): Replace
16688 pointer argument with an rtx and an index.
16689 * read-rtl.c (iterator_group::apply_iterator): Likewise.
16690 (apply_mode_iterator): Likewise.
16691 (apply_code_iterator): Likewise.
16692 (apply_int_iterator): Likewise.
16693 (apply_subst_iterator): Likewise.
16694 (record_iterator_use): Likewise.
16695 (record_attribute_use): Likewise.
16696 (md_reader::record_potential_iterator_use): Likewise. Update calls
16697 to record_iterator_use and apply_iterator.
16698 (iterator_use): Replace ptr with x and index.
16699 (attribute_use): Likewise.
16700 (apply_attribute_uses): Update calls to apply_iterator.
16701 (apply_iterators): Likewise. Update initialization of iterator_use.
16702 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
16703 and record_potential_iterator_use.
16704 (rtx_reader::read_rtx_operand): Likewise.
16706 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
16707 Alan Hayward <alan.hayward@arm.com>
16708 David Sherwood <david.sherwood@arm.com>
16710 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
16713 2017-08-21 Richard Biener <rguenther@suse.de>
16715 PR middle-end/81884
16716 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
16717 at struct end conservatively when comparing common bases.
16719 2017-08-21 Richard Biener <rguenther@suse.de>
16721 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
16722 (mem_ref_in_stmt): Remove.
16723 (determine_max_movement): Use ref index to get at the reference.
16724 (invariantness_dom_walker::before_dom_children): Deal with
16725 lim data already initialized.
16726 (gather_mem_refs_stmt): Initialize lim data and record ref index.
16728 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16730 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
16731 (TARGET_ISA_ROUND): Ditto.
16732 (TARGET_ROUND): Ditto.
16733 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
16734 * config/i386/i386.md: Ditto.
16735 * config/i386/sse.md: Ditto.
16736 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
16737 with OPTION_MASK_ISA_SSE4_1.
16739 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
16742 * doc/extend.texi (x86 Built-in Functions): Correct the name of
16743 __builtin_ia32_lzcnt_u16.
16745 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
16748 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
16749 (rs6000_set_current_function): Rewrite function to use it.
16751 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
16754 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
16755 and TYPE_WARN_IF_NOT_ALIGN.
16756 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
16757 (handle_warn_if_not_align): New.
16758 (place_union_field): Call handle_warn_if_not_align.
16759 (place_field): Call handle_warn_if_not_align.
16760 Copy TYPE_WARN_IF_NOT_ALIGN.
16761 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
16762 (layout_type): Likewise.
16763 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
16765 (tree_decl_common): Add warn_if_not_align.
16766 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
16767 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
16768 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
16769 (DECL_WARN_IF_NOT_ALIGN): Likewise.
16770 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
16771 * doc/extend.texi: Document warn_if_not_aligned attribute.
16772 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
16774 2017-08-17 Martin Liska <mliska@suse.cz>
16777 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
16778 (get_data_dependence): Use it as pointer type.
16779 (distribute_loop): Likewise.
16781 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16783 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
16784 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
16785 (p8_vmrgow_<mode>_direct): New define_insn.
16786 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
16787 handle endianness for vmrgew and vmrgow permute patterns.
16789 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16791 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
16792 * config/rs6000/rs6000-cpus.def: Remove comment.
16793 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
16794 (POWERPC_MASKS): Likewise.
16795 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
16796 use of TARGET_VSX_TIMODE.
16797 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
16798 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
16799 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
16800 (rs6000_option_override_internal): Remove dead code.
16801 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
16802 (rs6000_legitimize_reload_address): Likewise.
16803 (rs6000_legitimate_address_p): Likewise.
16804 (rs6000_opt_masks): Delete "vsx-timode".
16805 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
16806 from function comment.
16807 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
16808 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
16809 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
16811 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
16812 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
16813 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
16814 (VSX_TI): Likewise.
16816 (define_peephole2): Likewise.
16818 2017-08-17 Martin Sebor <msebor@redhat.com>
16821 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
16822 past the end of an array.
16823 (test_pp_format): Add test cases.
16825 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
16827 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
16828 missing ECF_NOTHROW flags.
16830 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
16833 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
16834 operands residing in integer registers.
16835 (*vsx_le_perm_load_<mode>): Likewise.
16836 (*vsx_le_perm_store_<mode>): Likewise.
16837 (define_peephole2): Add peepholes to optimize the above.
16839 2017-08-17 Marek Polacek <polacek@redhat.com>
16841 PR middle-end/81814
16842 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
16843 to mimic what shorten_compare did. Change the return type to bool.
16844 (fold_cond_expr_with_comparison): Update call to
16845 operand_equal_for_comparison_p.
16846 (fold_ternary_loc): Likewise.
16848 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
16850 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
16852 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
16854 2017-08-17 Richard Biener <rguenther@suse.de>
16856 * tree-ssa-structalias.c (solve_graph): When propagating
16857 to successors update the graphs succ edges and avoid duplicate work.
16859 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
16862 * config/i386/i386.c (ix86_option_override_internal): Save target
16863 specific options after ix86_stack_protector_guard_reg was changed.
16865 2017-08-17 Richard Biener <rguenther@suse.de>
16867 PR tree-optimization/81827
16868 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
16869 (new_var_info): Initialize it conservatively.
16870 (get_call_vi): Mark register vars.
16871 (new_scalar_tmp_constraint_exp): Likewise.
16872 (handle_rhs_call): Likewise.
16873 (handle_const_call): Likewise.
16874 (create_function_info_for): Likewise.
16875 (solve_constraints): Sort varinfos to separate register from
16876 non-register vars to pack points-to solution bitmaps during
16879 2017-08-17 Marek Polacek <polacek@redhat.com>
16881 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
16883 2017-08-17 Richard Biener <rguenther@suse.de>
16885 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
16886 to true when overflow is undefined and we saturated the result.
16888 2017-08-17 Alan Modra <amodra@gmail.com>
16891 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
16892 Don't use store multiple if only one reg needs saving.
16893 (interesting_frame_related_regno): New function.
16894 (rs6000_frame_related): Don't emit frame info for regs that
16896 (rs6000_emit_epilogue): Likewise.
16898 2017-08-16 Nathan Sidwell <nathan@acm.org>
16900 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
16901 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
16902 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
16903 * tree.c (free_lang_data_in_type): Use else-if chain. Always
16904 clear TYPE_LANG_1. Remove obsolete member-function stripping.
16905 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
16906 (verify_type): Adjust for TYPE_BINFO move.
16907 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
16908 process TYPE_BINFO directly.
16909 (hash_tree): Likewise.
16910 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16912 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16915 2017-08-16 David Malcolm <dmalcolm@redhat.com>
16917 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
16919 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
16922 * config/i386/i386.md (*anddi_1_btr): Change predicates of
16923 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
16924 Add ix86_binary_operator_ok to insn constraint.
16925 (*iordi_1_bts): Ditto.
16926 (*xordi_1_btc): Ditto.
16927 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
16928 Update corresponding peephole2 pattern.
16932 2017-08-16 Bin Cheng <bin.cheng@arm.com>
16934 PR tree-optimization/81832
16935 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
16936 copy loop header which has IFN_LOOP_DIST_ALIAS call.
16938 2017-08-16 Marek Polacek <polacek@redhat.com>
16941 * fold-const.c (fold_indirect_ref_1): Restore original behavior
16942 regarding size_zero_node.
16944 2017-08-16 Martin Liska <mliska@suse.cz>
16947 * config.gcc: Respect previously set extra_objs in case
16950 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
16952 PR tree-optimization/81835
16953 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
16954 the phi SSA_NAME. Check that the condition in a COND_EXPR does
16955 not depend on the phi.
16957 2017-08-16 Alan Modra <amodra@gmail.com>
16959 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
16962 2017-08-16 Alan Modra <amodra@gmail.com>
16964 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
16965 (save_reg_p): ..into this. Update all callers.
16966 (first_reg_to_save): Simplify.
16968 2017-08-16 Alan Modra <amodra@gmail.com>
16970 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
16973 2017-08-15 Joseph Myers <joseph@codesourcery.com>
16977 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
16978 constant count if that count is less than 32.
16980 2017-08-15 Nathan Sidwell <nathan@acm.org>
16982 * gcc.c (execute): Emit friendlier message if inferior is killed
16983 by an external cause.
16985 2017-08-15 Richard Biener <rguenther@suse.de>
16987 PR tree-optimization/81790
16988 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
16989 CONSTRUCTORs from simplifying and VN.
16991 2017-08-14 Martin Sebor <msebor@redhat.com>
16993 * builtin-attrs.def: Add comments.
16995 2017-08-14 Martin Sebor <msebor@redhat.com>
16998 * doc/extend.texi (attribute nonstring): Document new attribute.
17000 2017-08-14 Martin Sebor <msebor@redhat.com>
17003 * tree-diagnostic.c (default_tree_printer): Handle %G.
17004 * gimple-pretty-print.h (percent_G_format): Declare new function.
17005 * gimple-pretty-print.c (percent_G_format): Define.
17006 * tree-pretty-print.c (percent_K_format): Add argument.
17008 2017-08-14 Martin Sebor <msebor@redhat.com>
17010 PR translation/79998
17011 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
17012 Remove a stray space.
17014 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
17017 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
17018 (*iordi_1_bts): Ditto.
17019 (*xordi_1_btc): Ditto.
17021 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17024 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
17025 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17027 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
17028 quoted strings, and make more translator-friendly.
17029 (darwin_rs6000_override_options): Likewise.
17030 (rs6000_option_override_internal): Likewise.
17031 (rs6000_return_in_memory): Fix overlong line.
17032 (init_cmulative_args): Use quoted strings, and make more
17033 translator-friendly.
17034 (rs6000_pass_by_reference): Fix overlong line.
17035 (def_builtin): Use quoted strings.
17036 (altivec_expand_predicate_builtin): Use quoted strings, and make
17037 more translator-friendly.
17038 (htm_expand_builtin): Use quoted strings.
17039 (cpu_expand_builtin): Use quoted strings, and make more
17040 translator-friendly.
17041 (altivec_expand_builtin): Likewise.
17042 (paired_expand_predicate_builtin): Likewise.
17043 (rs6000_invalid_builtin): Likewise.
17044 (builtin_function_type): Use quoted strings.
17045 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
17046 more translator-friendly.
17047 (rs6000_trampoline_init): Likewise.
17048 (rs6000_handle_altivec_attribute): Likewise.
17049 (rs6000_inner_target_options): Use quoted strings.
17050 (rs6000_disable_incompatible_switches): Likewise.
17051 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
17052 strings, and make more translator-friendly.
17053 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
17055 2017-08-14 Bin Cheng <bin.cheng@arm.com>
17057 PR tree-optimization/81799
17058 * tree-loop-distribution.c (version_loop_by_alias_check): Force
17059 cond_expr to simple gimple operand.
17061 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
17063 PR middle-end/46932
17064 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
17066 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
17070 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
17071 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
17072 TARGET_GASISR_PROLOGUES.
17073 * config/avr/avr.c (avr_option_override): Same.
17074 (avr_pass_pre_proep::execute): Same.
17076 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
17079 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
17080 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
17082 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
17084 * config/i386/i386.md (*load_tp_<mode>): Redefine as
17085 define_insn_and_split. Split to a memory load from 0 in
17086 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
17087 using PTR mode iterator.
17088 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
17089 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
17090 (*add_tp_<mode>): Redefine as define_insn_and_split.
17091 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
17092 address space. Merge with *add_tp_x32 using PTR mode iterator.
17093 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
17094 Split to an add with a memory load from 0 in
17095 DEFAULT_TLS_SEG_REG address space.
17097 2017-08-12 Andrew Pinski <apinski@cavium.com>
17099 * config/aarch64/aarch64-option-extensions.def (rdma):
17100 Fix feature string to what Linux prints out in /proc/cpuinfo.
17102 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
17105 * dwarf2out.c (modified_type_die): For C typedef types that have
17106 an ultimate origin, process the ultimate origin instead of the
17108 (gen_typedef_die): Assert that input DECLs have no ultimate
17110 (gen_type_die_with_usage): For typedef variants that have an
17111 ultimate origin, just call gen_decl_die on the original DECL.
17112 (process_scope_var): Avoid creating DIEs for local typedefs and
17113 concrete static variables.
17115 2017-08-12 Alan Modra <amodra@gmail.com>
17119 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
17120 match gnu-user.h startfile.
17121 (ENDFILE_LINUX_SPEC): Similarly.
17123 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
17126 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
17128 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
17130 2017-08-11 Tamar Christina <tamar.christina@arm.com>
17131 * config/aarch64/aarch64.md (mov<mode>): Change.
17132 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
17133 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
17134 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
17136 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
17138 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
17139 for storage order barriers.
17141 2017-08-11 Martin Liska <mliska@suse.cz>
17144 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
17145 variables of void type.
17147 2017-08-11 Martin Liska <mliska@suse.cz>
17149 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
17150 TARGET_SUPPORTS_ALIASES.
17151 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
17152 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
17153 (optimize_weakref): Likewise.
17154 * symtab.c (symtab_node::noninterposable_alias): Likewise.
17155 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
17156 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
17158 2017-08-11 Martin Liska <mliska@suse.cz>
17161 * config/i386/i386.c (make_resolver_func): Do complete
17162 refactoring of the function.
17164 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17167 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
17168 * config/i386/i386.c (ix86_stack_protect_guard): Use
17169 ix86_stack_protect_guard_symbol_str to generate varible declaration.
17170 * doc/invoke.texi (x86 Options): Document
17171 -mstack-protector-guard-symbol= option.
17173 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
17175 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
17176 * config/i386/i386.c (ix86_split_stack_guard): New function.
17177 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
17178 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
17179 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
17180 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
17181 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
17182 (split_stack_space_check): Call ix86_split_stack_guard.
17184 2017-08-10 Martin Sebor <msebor@redhat.com>
17186 * print-tree.c (print_node): Print location using the established
17188 Replace spaces with colons.
17189 (debug_raw, debug): Ditto.
17191 2017-08-10 Martin Sebor <msebor@redhat.com>
17194 * pretty-print.c (pp_format): Correct the handling of %s precision.
17196 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
17199 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17201 (ix86_finalize_stack_frame_flags): This. Also clear
17202 frame_pointer_needed if -fno-omit-frame-pointer is used without
17204 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17205 with ix86_finalize_stack_frame_flags.
17206 (ix86_expand_epilogue): Likewise.
17207 (ix86_expand_split_stack_prologue): Likewise.
17208 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
17210 2017-08-10 Martin Liska <mliska@suse.cz>
17213 * c-attribs.c (handle_target_attribute):
17214 Report warning for an empty string argument of target attribute.
17216 2017-08-09 Jakub Jelinek <jakub@redhat.com>
17219 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
17221 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
17222 or DECL_NONLOCAL labels.
17223 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
17224 or DECL_NONLOCAL labels here.
17226 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
17228 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
17229 to indicate when early gimple folding has been disabled.
17230 (rs6000_gimple_fold_builtin): Add debug content.
17231 (rs6000_invalid_builtin): Fix whitespace.
17232 (rs6000_expand_builtin): Fix whitespace.
17233 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
17235 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
17238 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
17239 SAVE_MULTIPLE if not all the registers that saves, should be saved.
17241 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
17243 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
17244 (qdf24xx): Likewise.
17245 * config/aarch64/aarch64.md: Include falkor.md.
17246 * config/aarch64/falkor.md: New.
17248 2017-08-09 Marek Polacek <polacek@redhat.com>
17251 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
17252 * diagnostic.c (emit_diagnostic): Add a comment.
17253 (emit_diagnostic_valist): New function.
17255 2017-08-09 Marek Polacek <polacek@redhat.com>
17258 * input.c (make_location): New overload.
17259 * input.h (make_location): Declare.
17261 2017-08-08 Alan Modra <amodra@gmail.com>
17262 H.J. Lu <hongjiu.lu@intel.com>
17265 * gcc.c (NO_PIE_SPEC): Delete.
17266 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
17268 (LINK_PIE_SPEC): ..to here.
17269 (LINK_COMMAND_SPEC): Support -no-pie.
17270 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
17271 chain of crtbegin*.o selection, update for PIE_SPEC changes and
17273 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
17274 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
17275 (ENDFILE_CRTEND_SPEC): Similarly.
17277 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
17280 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
17281 (mstack-protector-guard-offset=): Ditto.
17282 * config/i386/i386.c (ix86_option_override): Handle
17283 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
17285 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
17286 ix86_stack_protect_guard_offset variables.
17287 (TARGET_STACK_PROTECT_GUARD): Always define.
17288 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
17289 and -mstack-protector-guard-offset= options.
17291 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17293 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
17294 boundary case for the last candidate.
17296 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17298 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
17299 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
17301 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17303 PR middle-end/19706
17304 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
17305 * config/aarch64/aarch64-builtins.c
17306 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
17307 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
17308 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
17310 2017-08-08 Tamar Christina <tamar.christina@arm.com>
17311 Andrew Pinski <pinskia@gmail.com>
17313 PR middle-end/19706
17314 * internal-fn.def (XORSIGN): New.
17315 * optabs.def (xorsign_optab): New.
17316 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
17317 (convert_expand_mult_copysign): New.
17318 (pass_optimize_widening_mul::execute): Call
17319 convert_expand_mult_copysign.
17321 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17323 PR tree-optimization/81354
17324 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
17325 Insert on edges rather than explicitly creating landing pads.
17326 (analyze_candidates_and_replace): Commit edge inserts.
17328 2017-08-08 Richard Biener <rguenther@suse.de>
17330 PR middle-end/81719
17331 * tree-ssa-loop-niter.c: Include tree-dfa.h.
17332 (expand_simple_operations): Also look through ADDR_EXPRs with
17333 MEM_REF bases treating them as POINTER_PLUS_EXPR.
17335 2017-08-08 Richard Biener <rguenther@suse.de>
17337 PR tree-optimization/81723
17338 * tree-vect-slp.c (struct bst_traits): New hash traits.
17339 (bst_fail): New global.
17340 (vect_build_slp_tree_2): New worker, split out from ...
17341 (vect_build_slp_tree): ... this now wrapping it with using
17342 bst_fail set to cache SLP tree build fails. Properly handle
17344 (vect_analyze_slp_instance): Allocate and free bst_fail.
17346 2017-08-08 Martin Liska <mliska@suse.cz>
17349 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
17350 LABEL_DECLs that can be from a different function.
17352 2017-08-08 Bin Cheng <bin.cheng@arm.com>
17354 PR tree-optimization/81744
17355 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
17356 loop's number of iterations.
17358 2017-08-08 Martin Liska <mliska@suse.cz>
17360 * asan.c: Include header files.
17361 * attribs.c (build_decl_attribute_variant): New function moved
17363 (build_type_attribute_qual_variant): Likewise.
17364 (cmp_attrib_identifiers): Likewise.
17365 (simple_cst_list_equal): Likewise.
17366 (omp_declare_simd_clauses_equal): Likewise.
17367 (attribute_value_equal): Likewise.
17368 (comp_type_attributes): Likewise.
17369 (build_type_attribute_variant): Likewise.
17370 (lookup_ident_attribute): Likewise.
17371 (remove_attribute): Likewise.
17372 (merge_attributes): Likewise.
17373 (merge_type_attributes): Likewise.
17374 (merge_decl_attributes): Likewise.
17375 (merge_dllimport_decl_attributes): Likewise.
17376 (handle_dll_attribute): Likewise.
17377 (attribute_list_equal): Likewise.
17378 (attribute_list_contained): Likewise.
17379 * attribs.h (lookup_attribute): New function moved from tree.[ch].
17380 (lookup_attribute_by_prefix): Likewise.
17381 * bb-reorder.c: Include header files.
17382 * builtins.c: Likewise.
17383 * calls.c: Likewise.
17384 * cfgexpand.c: Likewise.
17385 * cgraph.c: Likewise.
17386 * cgraphunit.c: Likewise.
17387 * convert.c: Likewise.
17388 * dwarf2out.c: Likewise.
17389 * final.c: Likewise.
17390 * fold-const.c: Likewise.
17391 * function.c: Likewise.
17392 * gimple-expr.c: Likewise.
17393 * gimple-fold.c: Likewise.
17394 * gimple-pretty-print.c: Likewise.
17395 * gimple.c: Likewise.
17396 * gimplify.c: Likewise.
17397 * hsa-common.c: Likewise.
17398 * hsa-gen.c: Likewise.
17399 * internal-fn.c: Likewise.
17400 * ipa-chkp.c: Likewise.
17401 * ipa-cp.c: Likewise.
17402 * ipa-devirt.c: Likewise.
17403 * ipa-fnsummary.c: Likewise.
17404 * ipa-inline.c: Likewise.
17405 * ipa-visibility.c: Likewise.
17407 * lto-cgraph.c: Likewise.
17408 * omp-expand.c: Likewise.
17409 * omp-general.c: Likewise.
17410 * omp-low.c: Likewise.
17411 * omp-offload.c: Likewise.
17412 * omp-simd-clone.c: Likewise.
17413 * opts-global.c: Likewise.
17414 * passes.c: Likewise.
17415 * predict.c: Likewise.
17416 * sancov.c: Likewise.
17417 * sanopt.c: Likewise.
17418 * symtab.c: Likewise.
17419 * toplev.c: Likewise.
17420 * trans-mem.c: Likewise.
17421 * tree-chkp.c: Likewise.
17422 * tree-eh.c: Likewise.
17423 * tree-into-ssa.c: Likewise.
17424 * tree-object-size.c: Likewise.
17425 * tree-parloops.c: Likewise.
17426 * tree-profile.c: Likewise.
17427 * tree-ssa-ccp.c: Likewise.
17428 * tree-ssa-live.c: Likewise.
17429 * tree-ssa-loop.c: Likewise.
17430 * tree-ssa-sccvn.c: Likewise.
17431 * tree-ssa-structalias.c: Likewise.
17432 * tree-ssa.c: Likewise.
17433 * tree-streamer-in.c: Likewise.
17434 * tree-vectorizer.c: Likewise.
17435 * tree-vrp.c: Likewise.
17436 * tsan.c: Likewise.
17437 * ubsan.c: Likewise.
17438 * varasm.c: Likewise.
17439 * varpool.c: Likewise.
17440 * tree.c: Remove functions moved to attribs.[ch].
17441 * tree.h: Likewise.
17442 * config/aarch64/aarch64.c: Add attrs.h header file.
17443 * config/alpha/alpha.c: Likewise.
17444 * config/arc/arc.c: Likewise.
17445 * config/arm/arm.c: Likewise.
17446 * config/avr/avr.c: Likewise.
17447 * config/bfin/bfin.c: Likewise.
17448 * config/c6x/c6x.c: Likewise.
17449 * config/cr16/cr16.c: Likewise.
17450 * config/cris/cris.c: Likewise.
17451 * config/darwin.c: Likewise.
17452 * config/epiphany/epiphany.c: Likewise.
17453 * config/fr30/fr30.c: Likewise.
17454 * config/frv/frv.c: Likewise.
17455 * config/ft32/ft32.c: Likewise.
17456 * config/h8300/h8300.c: Likewise.
17457 * config/i386/winnt.c: Likewise.
17458 * config/ia64/ia64.c: Likewise.
17459 * config/iq2000/iq2000.c: Likewise.
17460 * config/lm32/lm32.c: Likewise.
17461 * config/m32c/m32c.c: Likewise.
17462 * config/m32r/m32r.c: Likewise.
17463 * config/m68k/m68k.c: Likewise.
17464 * config/mcore/mcore.c: Likewise.
17465 * config/microblaze/microblaze.c: Likewise.
17466 * config/mips/mips.c: Likewise.
17467 * config/mmix/mmix.c: Likewise.
17468 * config/mn10300/mn10300.c: Likewise.
17469 * config/moxie/moxie.c: Likewise.
17470 * config/msp430/msp430.c: Likewise.
17471 * config/nds32/nds32-isr.c: Likewise.
17472 * config/nds32/nds32.c: Likewise.
17473 * config/nios2/nios2.c: Likewise.
17474 * config/nvptx/nvptx.c: Likewise.
17475 * config/pa/pa.c: Likewise.
17476 * config/pdp11/pdp11.c: Likewise.
17477 * config/powerpcspe/powerpcspe.c: Likewise.
17478 * config/riscv/riscv.c: Likewise.
17479 * config/rl78/rl78.c: Likewise.
17480 * config/rx/rx.c: Likewise.
17481 * config/s390/s390.c: Likewise.
17482 * config/sh/sh.c: Likewise.
17483 * config/sol2.c: Likewise.
17484 * config/sparc/sparc.c: Likewise.
17485 * config/spu/spu.c: Likewise.
17486 * config/stormy16/stormy16.c: Likewise.
17487 * config/tilegx/tilegx.c: Likewise.
17488 * config/tilepro/tilepro.c: Likewise.
17489 * config/v850/v850.c: Likewise.
17490 * config/vax/vax.c: Likewise.
17491 * config/visium/visium.c: Likewise.
17492 * config/xtensa/xtensa.c: Likewise.
17494 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17497 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
17498 constraints since the -mupper-regs-* switches have been
17500 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
17501 into a vector from a double word element that was extracted from
17502 another vector, and eliminate extra XXPERMDI instructions.
17503 (vsx_concat_<mode>_2): Likewise.
17504 (vsx_concat_<mode>_3): Likewise.
17505 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
17506 concat to allow optimizing inserts from previous extracts.
17508 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
17510 * config/i386/i386.c (ix86_stack_protect_guard): Generate
17511 memory reference to a SSP offset in TLS address space.
17512 (ix86_print_operand) <case '@'>: Remove.
17513 (ix86_print_operand_punct_valid_p): Remove '@' code.
17514 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
17515 UNSPEC_SP_TLS_TEST.
17516 (stack_tls_protect_set_<mode>): Remove.
17517 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
17518 (stack_tls_protect_test_<mode>): Remove.
17519 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
17521 2017-08-07 Olivier Hainque <hainque@adacore.com>
17524 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
17526 2017-08-07 Douglas Rupp <rupp@adacore.com>
17528 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
17529 variable was referenced as multidir in command.
17531 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17534 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
17537 2017-08-07 Martin Liska <mliska@suse.cz>
17539 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
17540 * config/rl78/rl78.c: Add include of attribs.h.
17541 * config/sh/sh.c: Likewise.
17542 * config/v850/v850.c: Likewise.
17544 2017-08-07 Tom de Vries <tom@codesourcery.com>
17546 PR middle-end/78266
17547 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
17549 2017-08-07 Martin Liska <mliska@suse.cz>
17551 * config/mips/mips.c: Include attribs.h.
17553 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
17556 * doc/invoke.texi: Document change in behvaior for -Ofast for
17559 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
17561 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
17563 (aarch64_pop_regs): Likewise.
17564 (aarch64_gen_load_pair): Likewise.
17565 (aarch64_save_callee_saves): Likewise.
17566 (aarch64_restore_callee_saves): Likewise.
17568 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17570 * config/i386/i386.c: Revert the last change.
17572 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17575 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
17577 (ix86_finalize_stack_frame_flags): This. Also clear
17578 frame_pointer_needed if -fno-omit-frame-pointer is used without
17580 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
17581 with ix86_finalize_stack_frame_flags.
17582 (ix86_expand_epilogue): Likewise.
17583 (ix86_expand_split_stack_prologue): Likewise.
17585 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
17588 * config/i386/i386.c (get_builtin_code_for_version): Set priority
17589 to P_AES for Westmere.
17591 2017-08-07 Jonathan Yong <10walls@gmail.com>
17593 * config/i386/mingw.opt (fset-stack-executable): Removed.
17594 * config/i386/cygming.opt (fset-stack-executable): Moved
17596 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
17598 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
17600 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
17602 2017-08-07 Marek Polacek <polacek@redhat.com>
17604 PR middle-end/81737
17605 * fold-const.c (fold_indirect_ref_1): Check type_domain.
17607 2017-08-07 Martin Liska <mliska@suse.cz>
17609 * attribs.h (canonicalize_attr_name): New function.
17610 (cmp_attribs): Move from c-format.c and adjusted.
17611 (is_attribute_p): Moved from tree.h.
17612 * tree-inline.c: Add new includes.
17613 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
17614 (private_is_attribute_p): Remove.
17615 (private_lookup_attribute): Likewise.
17616 (private_lookup_attribute_by_prefix): Simplify.
17617 (remove_attribute): Use is_attribute_p.
17618 * tree.h: Remove removed declarations.
17620 2017-08-07 Jakub Jelinek <jakub@redhat.com>
17622 PR middle-end/81698
17623 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
17624 instead of computing it in the function. Formatting fix.
17625 (expand_case): Don't rely on default_edge being the first edge,
17626 clear it if removing it, pass default_edge to
17627 emit_case_dispatch_table.
17628 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
17631 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
17633 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
17634 insn in the function, emit NOP after the insn.
17636 2017-08-06 Tom de Vries <tom@codesourcery.com>
17638 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
17641 2017-08-06 Tom de Vries <tom@codesourcery.com>
17643 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
17646 2017-08-04 Yury Gribov <tetra2005@gmail.com>
17648 PR tree-optimization/57371
17649 * match.pd: New pattern.
17651 2017-08-04 Marek Polacek <polacek@redhat.com>
17653 PR middle-end/81695
17654 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
17655 perform the computation in offset_int.
17657 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17659 PR tree-optimization/81136
17660 * tree-vectorizer.h: Include tree-hash-traits.h.
17661 (vec_base_alignments): New typedef.
17662 (vec_info): Add a base_alignments field.
17663 (vect_record_base_alignments): Declare.
17664 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
17666 (DR_IS_CONDITIONAL_IN_STMT): New macro.
17667 (create_data_ref): Add an is_conditional_in_stmt argument.
17668 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
17669 the is_conditional_in_stmt field.
17670 (data_ref_loc): Add an is_conditional_in_stmt field.
17671 (get_references_in_stmt): Set the is_conditional_in_stmt field.
17672 (find_data_references_in_stmt): Update call to create_data_ref.
17673 (graphite_find_data_references_in_stmt): Likewise.
17674 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
17675 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
17676 (vect_record_base_alignment): New function.
17677 (vect_record_base_alignments): Likewise.
17678 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
17679 for nested statements even if we fail to compute a misalignment.
17680 Use pooled base alignments for unconditional references.
17681 (vect_find_same_alignment_drs): Compare base addresses instead
17683 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
17684 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
17686 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17688 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
17689 Add an explicit name for the enum. Use auto_vec for slp_instances
17690 and grouped_stores.
17691 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
17693 (_bb_vec_info): Add a constructor and destructor.
17694 (vinfo_for_stmt): Return NULL for uids of -1 as well.
17695 (destroy_loop_vec_info): Delete.
17696 (vect_destroy_datarefs): Likewise.
17697 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
17698 (vec_info::vec_info): New function.
17699 (vec_info::~vec_info): Likewise.
17700 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
17701 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
17702 destroy_loop_vec_info.
17703 * tree-vect-loop.c (new_loop_vec_info): Replace with...
17704 (_loop_vec_info::_loop_vec_info): ...this.
17705 (destroy_loop_vec_info): Replace with...
17706 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
17707 the stmt_vec_infos. Leave handling of vec_info information to its
17708 destructor. Remove explicit vector releases.
17709 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
17710 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
17711 * tree-vect-slp.c (new_bb_vec_info): Replace with...
17712 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
17713 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
17714 (destroy_bb_vec_info): Replace with...
17715 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
17716 information to its destructor.
17717 (vect_slp_analyze_bb_1): Use new and delete instead of
17718 new_bb_vec_info and destroy_bb_vec_info.
17719 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
17722 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
17724 * tree-data-ref.h (subscript): Add access_fn field.
17725 (data_dependence_relation): Add could_be_independent_p.
17726 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
17727 (same_access_functions): Move to tree-data-ref.c.
17728 * tree-data-ref.c (ref_contains_union_access_p): New function.
17729 (access_fn_component_p): Likewise.
17730 (access_fn_components_comparable_p): Likewise.
17731 (dr_analyze_indices): Add a reference to access_fn_component_p.
17732 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
17734 (constant_access_functions): Likewise.
17735 (add_other_self_distances): Likewise.
17736 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
17737 (initialize_data_dependence_relation): Use XCNEW and remove
17738 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
17739 of access functions that have the same type. Allow the
17740 subsequence to end with different bases in some circumstances.
17741 Record the chosen access functions in SUB_ACCESS_FN.
17742 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
17743 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
17744 (subscript_dependence_tester_1): Likewise dra and drb.
17745 (build_classic_dist_vector): Update calls accordingly.
17746 (subscript_dependence_tester): Likewise.
17747 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
17748 DDR_COULD_BE_INDEPENDENT_P.
17749 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
17750 comp_alias_ddrs instead of may_alias_ddrs.
17751 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
17753 (vect_analyze_data_ref_dependence): Use it if
17754 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
17755 distance vectors if that fails.
17756 (dependence_distance_ge_vf): New function.
17757 (vect_prune_runtime_alias_test_list): Use it. Don't clear
17758 LOOP_VINFO_MAY_ALIAS_DDRS.
17760 2017-08-04 Richard Biener <rguenther@suse.de>
17762 PR middle-end/81705
17763 * fold-const.c (fold_binary_loc): Properly restrict
17764 minus_var0 && minus_var1 case when associating undefined overflow
17767 2017-08-04 Tom de Vries <tom@codesourcery.com>
17769 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
17771 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17773 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17774 Don't start diagnostic messages with a capital letter.
17775 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17777 (rs6000_invalid_builtin): Likewise.
17778 (rs6000_trampoline_init): Likewise.
17780 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17783 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
17784 after setting changeable df flags.
17786 2017-08-03 Richard Biener <rguenther@suse.de>
17788 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
17789 up if the use is in USE - X.
17791 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
17793 * toplev.c (dumpfile.h): New include.
17794 (internal_error_reentered): New static function. Use it...
17795 (internal_error_function): ...here to handle reentered internal_error.
17797 2017-08-03 Richard Biener <rguenther@suse.de>
17799 PR middle-end/81148
17800 * fold-const.c (split_tree): Add minus_var and minus_con
17801 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
17802 here but always use minus_*.
17803 (associate_trees): Assert we never associate with MINUS_EXPR
17804 and NULL first operand. Do not recurse for PLUS_EXPR operands
17805 when associating as MINUS_EXPR either.
17806 (fold_binary_loc): Track minus_var and minus_con.
17808 2017-08-03 Tom de Vries <tom@codesourcery.com>
17811 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
17812 ACCEL_COMPILER, apply finish_options on
17813 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
17815 2017-08-03 Tom de Vries <tom@codesourcery.com>
17818 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
17819 function_entry_patch_area_size > 0.
17821 2017-08-03 Jakub Jelinek <jakub@redhat.com>
17824 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
17825 instead of 10??LU, perform unit multiplication in wide_int,
17826 don't change alloc_object_size_limit if the limit is larger
17829 PR tree-optimization/81655
17830 PR tree-optimization/81588
17831 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
17832 the case when ranges[i].low and high are 1 for unsigned type with
17835 PR middle-end/81052
17836 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
17837 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
17839 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17841 * tree-vrp.h: Add include guard.
17843 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
17846 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
17847 (ud2): New insn pattern.
17848 * config/i386/i386.c (ix86_expand_epilogue):
17849 For naked functions, generate ud2 instead of trap insn.
17851 2017-08-02 Marek Polacek <polacek@redhat.com>
17854 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
17856 2017-08-02 Tom de Vries <tom@codesourcery.com>
17857 Cesar Philippidis <cesar@codesourcery.com>
17859 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
17860 Add missing edge probabilities.
17862 2017-08-02 Tamar Christina <tamar.christina@arm.com>
17864 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
17865 Correct endianness.
17867 2017-08-02 Jakub Jelinek <jakub@redhat.com>
17869 PR middle-end/79499
17870 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
17871 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
17872 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
17873 prologue_seq sequences - if any.
17875 2017-08-02 Richard Biener <rguenther@suse.de>
17877 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
17878 via vectors if supported, integer extracts via punning if supported
17879 or otherwise vector extracts.
17881 2017-08-02 Richard Biener <rguenther@suse.de>
17883 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
17885 (bitmap_insert_into_set): ... this.
17887 2017-08-02 Richard Biener <rguenther@suse.de>
17889 PR tree-optimization/81633
17891 2015-08-17 Alan Hayward <alan.hayward@arm.com>
17893 PR tree-optimization/71752
17894 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
17896 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
17898 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
17899 (machine_function::call_ms2sysv_pad_out): Remove field.
17900 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
17901 (ix86_compute_frame_layout): Likewise.
17903 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
17906 * config/i386/i386.c (ix86_set_func_type): Disallow naked
17907 attribute with interrupt attribute.
17909 2017-08-01 Andrew Pinski <apinski@cavium.com>
17911 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
17912 BIT_INSERT_EXPR's operand 1
17913 to see if the types precision matches.
17915 2017-08-01 Martin Liska <mliska@suse.cz>
17917 PR middle-end/70140
17918 * builtins.c (expand_builtin_memcpy_args): Remove.
17919 (expand_builtin_memcpy): Call newly added function
17920 expand_builtin_memory_copy_args.
17921 (expand_builtin_memcpy_with_bounds): Likewise.
17922 (expand_builtin_mempcpy): Remove last argument.
17923 (expand_builtin_mempcpy_with_bounds): Likewise.
17924 (expand_builtin_memory_copy_args): New function created from
17925 expand_builtin_mempcpy_args with small modifications.
17926 (expand_builtin_mempcpy_args): Remove.
17927 (expand_builtin_stpcpy): Remove unused argument.
17928 (expand_builtin): Likewise.
17929 (expand_builtin_with_bounds): Likewise.
17931 2017-08-01 Martin Liska <mliska@suse.cz>
17934 Make mempcpy more optimal (PR middle-end/70140).
17936 2017-08-01 Jakub Jelinek <jakub@redhat.com>
17939 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
17940 __builtin_vec_cmpne verify both arguments are compatible vectors
17941 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
17942 verify arg1_type is a pointer or array type. For __builtin_vec_st,
17943 move computation of aligned to after checking the argument types.
17947 * config/rs6000/vsx.md (vextract_fp_from_shorth,
17948 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
17951 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
17952 Doug Rupp <rupp@adacore.com>
17953 Olivier Hainque <hainque@adacore.com>
17955 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
17956 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
17958 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
17959 for TARGET_OS_CPP_BUILTIN.
17960 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
17961 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
17963 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
17964 passing required abi options to the assembler for EABI configurations.
17965 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
17966 of .text.hot and .text.unlikely sections for kernel modules when
17967 using ARM style exceptions.
17968 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
17969 options. Add EXTRA_CC1_SPEC.
17970 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
17972 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
17974 (ARM_TARGET2_DWARF_FORMAT): Define.
17975 * config/arm/t-vxworks: Adjust multilib control to removal of the
17976 Diab command line options.
17978 2017-08-01 Martin Liska <mliska@suse.cz>
17980 PR gcov-profile/81561
17981 * gcov.c (unblock): Make unblocking safe as we need to preserve
17982 index correspondence of blocks and block_lists.
17984 2017-08-01 Richard Biener <rguenther@suse.de>
17986 PR tree-optimization/81181
17987 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
17988 (compute_antic): ... end of iteration here.
17990 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
17992 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
17993 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
17994 (ftree-slp-vectorize): Likewise.
17995 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
17996 can no longer be set independent of flag_tree_loop_vectorize.
17997 * omp-general.c (emp_max_vf): Likewise.
17998 * opts.c (enable_fdo_optimizations): Remove references to
17999 flag_tree_vectorize, these are now implicit.
18000 (common_handle_option): Remove handling for OPT_ftree_vectorize,
18001 and leave it for the options machinery.
18003 2017-08-01 Martin Liska <mliska@suse.cz>
18005 PR middle-end/70140
18006 * builtins.c (expand_builtin_memcpy_args): Remove.
18007 (expand_builtin_memcpy): Call newly added function
18008 expand_builtin_memory_copy_args.
18009 (expand_builtin_memcpy_with_bounds): Likewise.
18010 (expand_builtin_mempcpy): Remove last argument.
18011 (expand_builtin_mempcpy_with_bounds): Likewise.
18012 (expand_builtin_memory_copy_args): New function created from
18013 expand_builtin_mempcpy_args with small modifications.
18014 (expand_builtin_mempcpy_args): Remove.
18015 (expand_builtin_stpcpy): Remove unused argument.
18016 (expand_builtin): Likewise.
18017 (expand_builtin_with_bounds): Likewise.
18019 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18022 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
18023 print "ds:" only for immediates in generic address space.
18025 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
18028 * config/i386/i386.c (ix86_funciton_naked): New prototype.
18029 (ix86_function_ok_for_sibcall): Return false for naked functions.
18031 2017-08-01 Richard Biener <rguenther@suse.de>
18033 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
18034 (compute_antic): Seed worklist with exit block predecessors.
18035 * cfganal.c (dfs_find_deadend): For a cycle return the source
18036 of the edge closing it.
18038 2017-08-01 Tamar Christina <tamar.christina@arm.com>
18040 * config/aarch64/aarch64.c
18041 (aarch64_can_const_movi_rtx_p): Move 0 check.
18043 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18045 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
18046 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
18048 * match.pd: Ditto in address comparison pattern.
18050 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18052 PR tree-optimization/81627
18053 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
18054 closed ssa form for store-store chain.
18056 2017-08-01 Bin Cheng <bin.cheng@arm.com>
18058 PR tree-optimization/81620
18059 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
18060 for store-store chain.
18062 2017-08-01 Jakub Jelinek <jakub@redhat.com>
18064 PR tree-optimization/81588
18065 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
18066 ranges[i].in_p, invert comparison code ccode. For >/>=,
18067 swap rhs1 and rhs2 and comparison code unconditionally,
18068 for </<= don't do that. Don't swap rhs1/rhs2 again if
18069 ranges[i].in_p, instead invert comparison code ccode if
18070 opcode or oe->rank is BIT_IOR_EXPR.
18073 * optabs.def (vec_extract_optab, vec_init_optab): Change from
18074 a direct optab to conversion optab.
18075 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
18076 with GET_MODE_INNER as last argument instead of optab_handler.
18077 * expmed.c (extract_bit_field_1): Likewise. Use vector from
18078 vector extraction if possible and optab is available.
18079 * expr.c (store_constructor): Use convert_optab_handler instead
18080 of optab_handler. Use vector initialization from smaller
18081 vectors if possible and optab is available.
18082 * tree-vect-stmts.c (vectorizable_load): Likewise.
18083 * doc/md.texi (vec_extract, vec_init): Document that the optabs
18084 now have two modes.
18085 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
18086 of vec_init from half-sized vectors with the same element mode.
18087 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
18088 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
18089 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
18090 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
18091 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
18092 after mode in gen_vec_extract* calls.
18093 (vec_extract<mode>): Renamed to ...
18094 (vec_extract<mode><ssescalarmodelower>): ... this.
18095 (vec_extract<mode><ssehalfvecmodelower>): New expander.
18096 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
18097 element mode after mode in gen_vec_init* calls.
18098 (VEC_INIT_HALF_MODE): New mode iterator.
18099 (vec_init<mode>): Renamed to ...
18100 (vec_init<mode><ssescalarmodelower>): ... this.
18101 (vec_init<mode><ssehalfvecmodelower>): New expander.
18102 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
18103 (vec_extractv2sfsf): ... this.
18104 (vec_initv2sf): Renamed to ...
18105 (vec_initv2sfsf): ... this.
18106 (vec_extractv2si): Renamed to ...
18107 (vec_extractv2sisi): ... this.
18108 (vec_initv2si): Renamed to ...
18109 (vec_initv2sisi): ... this.
18110 (vec_extractv4hi): Renamed to ...
18111 (vec_extractv4hihi): ... this.
18112 (vec_initv4hi): Renamed to ...
18113 (vec_initv4hihi): ... this.
18114 (vec_extractv8qi): Renamed to ...
18115 (vec_extractv8qiqi): ... this.
18116 (vec_initv8qi): Renamed to ...
18117 (vec_initv8qiqi): ... this.
18118 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
18119 (vec_init<mode>): Renamed to ...
18120 (vec_init<mode><VEC_base_l>): ... this.
18121 (vec_extract<mode>): Renamed to ...
18122 (vec_extract<mode><VEC_base_l>): ... this.
18123 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
18124 (vec_initv2sfsf): ... this.
18125 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
18126 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18127 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
18128 element mode after mode in gen_vec_init* calls.
18129 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
18130 (vec_init<mode><Vel>): ... this.
18131 (vec_extract<mode>): Renamed to ...
18132 (vec_extract<mode><Vel>): ... this.
18133 * config/aarch64/iterators.md (Vel): New mode attribute.
18134 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
18135 Add element mode after mode in gen_vec_extract* calls.
18136 * config/s390/vector.md (non_vec_l): New mode attribute.
18137 (vec_extract<mode>): Renamed to ...
18138 (vec_extract<mode><non_vec_l>): ... this.
18139 (vec_init<mode>): Renamed to ...
18140 (vec_init<mode><non_vec_l>): ... this.
18141 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
18142 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
18144 * config/arm/iterators.md (V_elem_l): New mode attribute.
18145 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
18146 (vec_extract<mode><V_elem_l>): ... this.
18147 (vec_extractv2di): Renamed to ...
18148 (vec_extractv2didi): ... this.
18149 (vec_init<mode>): Renamed to ...
18150 (vec_init<mode><V_elem_l>): ... this.
18151 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
18152 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
18153 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
18154 Add element mode after gen_vec_extract* calls.
18155 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
18156 (vec_init<mode><unitmode>): ... this.
18157 (vec_extract<mode>): Renamed to ...
18158 (vec_extract<mode><unitmode>): ... this.
18159 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
18160 (vec_init<mode><unitmode>): ... this.
18161 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
18162 (vec_initv2sfsf): ... this.
18163 (vec_extractv2sf): Renamed to ...
18164 (vec_extractv2sfsf): ... this.
18165 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
18166 Add element mode after gen_vec_extract* calls.
18167 * config/mips/mips.md (unitmode): New mode iterator.
18168 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
18169 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
18170 * config/spu/spu.md (inner_l): New mode attribute.
18171 (vec_init<mode>): Renamed to ...
18172 (vec_init<mode><inner_l>): ... this.
18173 (vec_extract<mode>): Renamed to ...
18174 (vec_extract<mode><inner_l>): ... this.
18175 * config/sparc/sparc.md (veltmode): New mode iterator.
18176 (vec_init<VMALL:mode>): Renamed to ...
18177 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
18178 * config/ia64/vect.md (vec_initv2si): Renamed to ...
18179 (vec_initv2sisi): ... this.
18180 (vec_initv2sf): Renamed to ...
18181 (vec_initv2sfsf): ... this.
18182 (vec_extractv2sf): Renamed to ...
18183 (vec_extractv2sfsf): ... this.
18184 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
18185 (vec_init<mode>): Renamed to ...
18186 (vec_init<mode><VEC_base_l>): ... this.
18187 (vec_extract<mode>): Renamed to ...
18188 (vec_extract<mode><VEC_base_l>): ... this.
18189 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
18190 (vec_initv2sfsf): ... this.
18191 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
18192 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
18193 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
18194 gen_vec_init* calls.
18196 2017-08-01 Richard Biener <rguenther@suse.de>
18198 PR tree-optimization/81297
18199 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
18200 TREE_OVERFLOW from INTEGER_CSTs.
18202 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
18204 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
18206 2017-07-31 Carl Love <cel@us.ibm.com>
18208 * config/rs6000/rs6000-c: Add support for built-in functions
18209 vector signed char vec_xl_be (signed long long, signed char *);
18210 vector unsigned char vec_xl_be (signed long long, unsigned char *);
18211 vector signed int vec_xl_be (signed long long, signed int *);
18212 vector unsigned int vec_xl_be (signed long long, unsigned int *);
18213 vector signed long long vec_xl_be (signed long long, signed long long *);
18214 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
18215 vector signed short vec_xl_be (signed long long, signed short *);
18216 vector unsigned short vec_xl_be (signed long long, unsigned short *);
18217 vector double vec_xl_be (signed long long, double *);
18218 vector float vec_xl_be (signed long long, float *);
18219 * config/rs6000/altivec.h (vec_xl_be): Add #define.
18220 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
18221 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
18223 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
18224 (altivec_expand_builtin): Add switch statement to call
18225 altivec_expand_xl_be for each builtin.
18226 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
18227 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
18228 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
18229 __builtin_vsx_le_be_v16qi.
18230 * doc/extend.texi: Update the built-in documentation file for the
18231 new built-in functions.
18233 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18236 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
18238 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
18240 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18242 * config.gcc: Add z14.
18243 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
18244 CPU model numbers for z13s and z14.
18245 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
18247 * config/s390/s390-opts.h (enum processor_type): Rename
18248 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18249 * config/s390/s390.c (processor_table): Add field for CPU name to
18250 be passed to Binutils.
18251 (s390_asm_output_machine_for_arch): Use the new field in
18252 processor_table for Binutils.
18253 (s390_expand_builtin): Replace arch12 with z14.
18254 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
18255 (s390_get_sched_attrmask): Likewise.
18256 (s390_get_unit_mask): Likewise.
18257 * config/s390/s390.opt: Add z14 to processor_type enum.
18259 2017-07-31 Martin Jambor <mjambor@suse.cz>
18262 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
18263 regardless of optimization level.
18265 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
18266 Martin Liska <mliska@suse.cz>
18268 * predict.def: Remove old comment and adjust probability.
18269 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
18270 PREDICT statements.
18272 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
18275 * config/i386/i386.c (ix86_function_naked): New function.
18276 (ix86_can_use_return_insn_p): Return false for naked functions.
18277 (ix86_expand_prologue): Skip prologue for naked functions.
18278 (ix86_expand_epilogue): Skip epilogue for naked functions
18279 and emit trap instruction.
18280 (ix86_warn_func_return): New function.
18281 (ix86_attribute_table): Add "naked" attribute specification.
18282 (TARGET_WARN_FUNC_RETURN): Define.
18283 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
18285 2017-07-31 Martin Liska <mliska@suse.cz>
18287 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
18288 (dump_gimple_bb_header): Always dump BB info.
18289 (pp_cfg_jump): Do not append info about BB when dumping a jump.
18291 2017-07-31 Martin Liska <mliska@suse.cz>
18294 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
18295 also with current_function_decl non-null equality.
18297 2017-07-31 Jakub Jelinek <jakub@redhat.com>
18300 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
18301 change type to the element type, instead add eltype variable and
18302 use it where we are interested in the element type.
18304 PR tree-optimization/81603
18305 * ipa-polymorphic-call.c
18306 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
18307 offset arithmetic in offset_int, bail out if the resulting bit offset
18308 doesn't fit into shwi.
18310 2017-07-31 Martin Liska <mliska@suse.cz>
18312 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
18313 (gimplify_save_expr): Fix comment.
18315 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
18318 * config/i386/i386.c (ix86_function_arg): Update arguments for
18320 (ix86_compute_frame_layout): Set the initial stack offset to
18321 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
18322 INCOMING_FRAME_SP_OFFSET.
18323 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
18324 stack before exception handler returns.
18325 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
18326 the 'ERROR_CODE' for exception handler.
18328 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
18330 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
18331 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
18332 (ASM_OUTPUT_REG_POP): Ditto.
18333 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
18334 instead of asm_fprintf to output pure string.
18336 2017-07-29 Jakub Jelinek <jakub@redhat.com>
18338 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
18339 to imported_module_or_decl hook.
18340 (debug_nothing_tree_tree_tree_bool): Remove.
18341 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
18342 * debug.c (do_nothing_debug_hooks): Use
18343 debug_nothing_tree_tree_tree_bool_bool instead of
18344 debug_nothing_tree_tree_tree_bool.
18345 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18346 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
18347 * sdbout.c (sdb_debug_hooks): Likewise.
18348 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
18349 (gen_namespace_die): Add DW_AT_export_symbols attribute if
18351 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
18352 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
18353 attribute, don't add anything.
18355 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18357 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
18358 (fold_build2_stat_loc): Likewise.
18359 (fold_build3_stat_loc): Likewise.
18360 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
18361 (fold_build1_loc): Remove macro.
18362 (fold_build2_loc): Likewise.
18363 (fold_build3_loc): Likewise.
18365 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18367 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
18368 (gimple_build_debug_bind_source_stat): Likewise.
18369 * gimple.h (gimple_build_debug_bind): Remove macro.
18370 (gimple_build_debug_bind_source): Likewise.
18372 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18374 * bitmap.c (bitmap_alloc): Adjust.
18375 (bitmap_gc_alloc): Likewise.
18376 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
18378 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18380 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
18381 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
18382 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
18383 (bitmap_gc_alloc_stat): Likewise.
18384 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
18386 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18388 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
18389 * rtl.h (shallow_copy_rtx): Remove macro.
18391 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18393 * emit-rtl.c (gen_raw_REG): Adjust.
18394 * gengenrtl.c (gendef): Likewise.
18395 * rtl.c (rtx_alloc_stat): Remove _stat from name.
18396 * rtl.h (rtx_alloc): Remove macro.
18398 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18400 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
18401 (build_tree_list_stat): Likewise.
18402 * tree.h (build_tree_list): Remove macro.
18403 (build_tree_list_vec): Likewise.
18405 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18407 * tree.c (make_vector_stat): Remove _stat from name.
18408 (build_vector_stat): Likewise.
18409 * tree.h (make_vector_stat): Remove macro.
18410 (build_vector_stat): Likewise.
18412 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18414 * tree.h (build_var_debug_value): Remove prototype.
18416 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18418 * tree.c (tree_cons_stat): Remove _stat from name.
18419 * tree.h (tree_cons): Remove macro.
18421 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18423 * tree.c (build_vl_exp_stat): Remove _stat from name.
18424 * tree.h (build_vl_exp): Remove macro.
18426 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18428 * tree.c (build_decl_stat): Remove _stat from name.
18429 * tree.h (build_decl): Remove macro.
18431 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18433 * gimple.c (gimple_build_with_ops_stat): Adjust.
18434 (gimple_alloc_stat): Remove _stat from name.
18435 * gimple.h (gimple_alloc): Remove macro.
18437 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18439 * tree.c (make_tree_vec_stat): Remove _stat from name.
18440 (grow_tree_vec_stat): Likewise.
18441 * tree.h (make_tree_vec_stat): Adjust prototype.
18442 (grow_tree_vec_stat): Likewise.
18443 (make_tree_vec): Remove macro.
18444 (grow_tree_vec): Likewise.
18446 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18448 * fold-const.c (fold_build1_stat_loc): Adjust.
18449 (fold_build2_stat_loc): Likewise.
18450 (fold_build3_stat_loc): Likewise.
18451 * tree.c (build0_stat): Remove _stat from name.
18452 (build1_stat): Likewise.
18453 (build2_stat): Likewise.
18454 (build3_stat): Likewise.
18455 (build4_stat): Likewise.
18456 (build5_stat): Likewise.
18457 * tree.h (build1_loc): Remove macro, and rename _stat function
18459 (build2_loc): Likewise.
18460 (build3_loc): Likewise.
18461 (build4_loc): Likewise.
18462 (build5_loc): Likewise.
18464 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18466 * tree.c (make_int_cst_stat): Remove _stat from name.
18467 * tree.h (make_int_cst_stat): Adjust prototype.
18468 (make_int_cst): Remove macro.
18470 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18472 * tree.c (make_tre_binfo_stat): Remove _stat from name.
18473 * tree.h (make_tree_binfo_stat): Adjust prototype.
18474 (make_tree_binfo): Remove.
18476 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18478 * tree.c (copy_node_stat): Rename to copy_node.
18479 (build_distinct_type_copy): Adjust.
18480 * tree.h (copy_node_stat): Adjust prototype.
18481 (copy_node): Remove macro.
18483 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18485 * tree.c (make_node_stat): rename to make_node.
18486 (build_tree_list_stat): Adjust.
18487 (build0_stat): Likewise.
18488 (build2_stat): Likewise.
18489 (build3_stat): Likewise.
18490 (build4_stat): Likewise.
18491 (build5_stat): Likewise.
18492 (build_decl_stat): Likewise.
18493 * tree.h (make_node_stat): Adjust prototype.
18494 (make_node): remove macro.
18496 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
18498 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
18499 (PPC_FEATURE2_SCV): Likewise.
18500 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
18502 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18504 * config/aarch64/aarch64.c
18505 (aarch64_internal_mov_immediate): Add new special pattern.
18506 * config/aarch64/aarch64.md (*movdi_aarch64):
18507 Add reg/32bit const mov case.
18509 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18510 Richard Sandiford <richard.sandiford@linaro.org>
18512 * config/aarch64/aarch64.md (mov<mode>): Generalize.
18513 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18514 Add integer and movi cases.
18515 (movi-split-hf-df-sf split, fp16): New.
18516 (enabled): Added TARGET_FP_F16INST.
18517 * config/aarch64/iterators.md (GPF_HF): New.
18518 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
18520 2017-07-28 Tamar Christina <tamar.christina@arm.com>
18522 * config/aarch64/aarch64.c
18523 (aarch64_simd_container_mode): Add prototype.
18524 (aarch64_expand_mov_immediate): Add HI support.
18525 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
18526 (aarch64_can_const_movi_rtx_p): New.
18527 (aarch64_preferred_reload_class):
18528 Remove restrictions of using FP registers for certain SIMD operations.
18529 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
18530 (aarch64_valid_floating_const): Add integer move validation.
18531 (aarch64_simd_imm_scalar_p): Remove.
18532 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
18533 (aarch64_legitimate_constant_p): Expand list of supported cases.
18534 * config/aarch64/aarch64-protos.h
18535 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
18536 (aarch64_reinterpret_float_as_int): New.
18537 (aarch64_simd_imm_scalar_p): Remove.
18538 * config/aarch64/constraints.md (Uvi): New.
18539 (Dd): Split into Ds and new Dd.
18540 * config/aarch64/aarch64.md (*movsi_aarch64):
18542 (*movdi_aarch64): Add SIMD mov case.
18544 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18546 * tree-predcom.c: (struct chain): Handle store-store chain in which
18547 stores for elimination only store loop invariant values.
18548 (execute_pred_commoning_chain): Ditto.
18549 (prepare_initializers_chain_store_elim): Ditto.
18550 (prepare_finalizers): Ditto.
18551 (is_inv_store_elimination_chain): New function.
18552 (initialize_root_vars_store_elim_1): New function.
18554 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18556 * tree-predcom.c: Revise general description of the pass.
18557 (enum chain_type): New enum type for store elimination.
18558 (struct chain): New field supporting store elimination.
18559 (struct component): Ditto.
18560 (dump_chain): Dump store-stores chain.
18561 (release_chain): Release resources.
18562 (split_data_refs_to_components): Compute and create component
18563 contains only stores for elimination.
18564 (get_chain_last_ref_at): New function.
18565 (make_invariant_chain): Initialization.
18566 (make_rooted_chain): Specify chain type in parameter and record it.
18567 (add_looparound_copies): Skip for store-stores chain.
18568 (determine_roots_comp): Compute type of chain and pass it to
18570 (initialize_root_vars_store_elim_2): New function.
18571 (finalize_eliminated_stores): New function.
18572 (remove_stmt): Handle store for elimination.
18573 (execute_pred_commoning_chain): Execute predictive commoning on
18574 store-store chains.
18575 (determine_unroll_factor): Skip unroll for store-stores chain.
18576 (prepare_initializers_chain_store_elim): New function.
18577 (prepare_initializers_chain): Hanlde store-store chain.
18578 (prepare_finalizers_chain, prepare_finalizers): New function.
18579 (tree_predictive_commoning_loop): Return integer value indicating
18580 if loop is unrolled or lcssa form is corrupted.
18581 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
18583 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18585 * tree-predcom.c (initialize_root): Delete.
18586 (execute_pred_commoning_chain): Initialize root vars and replace
18587 reference of non-combined chain directly, rather than call above
18590 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18592 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
18593 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
18595 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18597 * tree-predcom.c (struct chain): New field init_seq.
18598 (release_chain): Release init_seq.
18599 (prepare_initializers_chain): Record intialization stmts in above
18601 (insert_init_seqs): New function.
18602 (tree_predictive_commoning_loop): Call insert_init_seqs.
18604 2017-07-28 Bin Cheng <bin.cheng@arm.com>
18606 * tree-predcom.c (determine_roots_comp): Skip trivial components.
18608 2017-07-28 Richard Biener <rguenther@suse.de>
18610 * match.pd: Remove superfluous :c.
18611 * genmatch.c (simplify::id): Add member.
18612 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
18614 (current_id): New global.
18615 (dt_node::parent): Move from ...
18616 (dt_operand::parent): ... here. Add for_id member.
18617 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
18618 (decision_tree::find_node): Relax order requirement when
18619 merging DT_TRUE nodes to ones inbetween the current simplify
18620 and the one we try to merge with. Add diagnostic whenever
18621 we need to enforce pattern order by not merging.
18622 (decision_tree::insert): Set current_id.
18623 (decision_tree::print_node): Dump parent node and for_id.
18624 (parser::last_id): Add member.
18625 (parser::push_simplify): Assign unique id.
18626 (parser::parser): Initialize last_id.
18628 2017-07-28 Martin Liska <mliska@suse.cz>
18631 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
18632 gimple_build_debug_bind.
18634 2017-07-28 Richard Biener <rguenther@suse.de>
18636 PR tree-optimization/81502
18637 * match.pd: Add pattern combining BIT_INSERT_EXPR with
18639 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
18641 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
18642 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
18643 for BIT_FIELD_REF args.
18644 * fold-const.c (make_bit_field_ref): Likewise.
18645 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
18647 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18650 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
18651 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
18652 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
18653 Or it into SANITIZER_UNDEFINED.
18654 * ubsan.c: Include gimple-fold.h and varasm.h.
18655 (ubsan_expand_ptr_ifn): New function.
18656 (instrument_pointer_overflow): New function.
18657 (maybe_instrument_pointer_overflow): New function.
18658 (instrument_object_size): Formatting fix.
18659 (pass_ubsan::execute): Call instrument_pointer_overflow
18660 and maybe_instrument_pointer_overflow.
18661 * internal-fn.c (expand_UBSAN_PTR): New function.
18662 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
18663 * sanitizer.def (__ubsan_handle_pointer_overflow,
18664 __ubsan_handle_pointer_overflow_abort): New builtins.
18665 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
18666 * internal-fn.def (UBSAN_PTR): New internal function.
18667 * opts.c (sanitizer_opts): Add pointer-overflow.
18668 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
18669 * fold-const.c (build_range_check): Compute pointer range check in
18670 integral type if pointer arithmetics would be needed. Formatting
18673 2017-07-28 Martin Liska <mliska@suse.cz>
18676 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
18677 parameters that are of a variable-length.
18679 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
18681 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
18683 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
18684 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
18685 (CRT_CALL_STATIC_FUNCTION): Likewise.
18686 (ASM_DEFAULT_SPEC): New define.
18687 (ASM_SPEC32): Likewise.
18688 (ASM_SPEC64): Likewise.
18689 (ASM_SPEC_COMMON): Likewise.
18690 (ASM_SPEC): Likewise.
18691 (INVALID_64BIT): Likewise.
18692 (LINK_OS_DEFAULT_SPEC): Likewise.
18693 (LINK_OS_SPEC32): Likewise.
18694 (LINK_OS_SPEC64): Likewise.
18695 (POWERPC_LINUX): Likewise.
18696 (PTRDIFF_TYPE): Likewise.
18697 (RESTORE_FP_PREFIX): Likewise.
18698 (RESTORE_FP_SUFFIX): Likewise.
18699 (SAVE_FP_PREFIX): Likewise.
18700 (SAVE_FP_SUFFIX): Likewise.
18701 (SIZE_TYPE): Likewise.
18702 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
18703 (TARGET_64BIT): Likewise.
18704 (TARGET_64BIT): Likewise.
18705 (TARGET_AIX): Likewise.
18706 (WCHAR_TYPE_SIZE): Likewise.
18707 (WCHAR_TYPE): Undefine.
18708 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
18709 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
18710 (CPP_OS_RTEMS_SPEC): Delete.
18711 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
18712 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
18714 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
18716 2017-07-28 Jakub Jelinek <jakub@redhat.com>
18718 PR tree-optimization/81578
18719 * tree-parloops.c (build_new_reduction): Bail out if
18720 reduction_code isn't one of the standard OpenMP reductions.
18721 Move the details printing after that decision.
18723 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18725 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
18726 related to reload_in_progress.
18727 (splat_input_operand): Likewise.
18728 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
18730 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
18732 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
18733 (TARGET_INSTANTIATE_DECLS): Likewise.
18734 (legitimate_indexed_address_p): Delete reload_in_progress code.
18735 (rs6000_debug_legitimate_address_p): Likewise.
18736 (rs6000_eliminate_indexed_memrefs): Likewise.
18737 (rs6000_emit_le_vsx_store): Likewise.
18738 (rs6000_emit_move_si_sf_subreg): Likewise.
18739 (rs6000_emit_move): Likewise.
18740 (register_to_reg_type): Likewise.
18741 (rs6000_pre_atomic_barrier): Likewise.
18742 (rs6000_machopic_legitimize_pic_address): Likewise.
18743 (rs6000_allocate_stack_temp): Likewise.
18744 (rs6000_address_for_fpconvert): Likewise.
18745 (rs6000_address_for_altivec): Likewise.
18746 (rs6000_secondary_memory_needed_rtx): Delete function.
18747 (rs6000_check_sdmode): Likewise.
18748 (rs6000_alloc_sdmode_stack_slot): Likewise.
18749 (rs6000_instantiate_decls): Likewise.
18750 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
18751 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
18752 Delete reload_in_progress.
18753 (*vec_reload_and_plus_<mptrsize>): Likewise.
18754 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
18755 (vsx_div_v2di): Likewise.
18756 (vsx_udiv_v2di): Likewise.
18758 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18760 * config/rs6000/rs6000.opt (mlra): Replace with stub.
18761 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
18762 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
18763 (rs6000_debug_reg_global): Delete print of LRA status.
18764 (rs6000_option_override_internal): Delete dead LRA related code.
18765 (rs6000_lra_p): Delete function.
18766 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
18768 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
18770 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
18771 * config/riscv/rtems.h: New file.
18773 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18774 Sudakshina Das <sudi.das@arm.com>
18776 * config/aarch64/aarch64.md
18777 (define_split for and<mode>3nr_compare): Move
18778 non aarch64_logical_operand to a register.
18779 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
18780 register immediate operand to a register.
18781 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
18783 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
18785 PR middle-end/81564
18786 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
18788 2017-07-27 Richard Biener <rguenther@suse.de>
18790 PR tree-optimization/81573
18791 PR tree-optimization/81494
18792 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
18793 multi defuse cycle case.
18795 2017-07-27 Richard Biener <rguenther@suse.de>
18797 PR tree-optimization/81571
18798 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
18801 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
18803 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
18804 earlier and only if MASK_FPU is set. Adjust formatting.
18806 2017-07-27 Martin Liska <mliska@suse.cz>
18808 * opt-functions.awk: Add validation of value of Init.
18809 * optc-gen.awk: Pass new argument.
18811 2017-07-27 Martin Liska <mliska@suse.cz>
18813 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
18814 Fix wrong condition.
18816 2017-07-27 Martin Liska <mliska@suse.cz>
18818 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
18819 BBs and edges seen by autoFDO.
18821 2017-07-27 Richard Biener <rguenther@suse.de>
18823 PR tree-optimization/81502
18824 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
18825 with incompatible but same sized type.
18826 (execute_update_addresses_taken): Likewise.
18828 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
18830 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
18831 flag_tree_loop_vectorize rather than flag_tree_vectorize.
18833 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18836 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
18837 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
18838 Change s_operand to memory_operand.
18840 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
18842 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
18843 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
18844 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
18845 Emit instructions rather than returning an expression. Handle TFmode
18846 and KFmode by casting to TImode.
18847 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
18848 (rs6000_emit_le_vsx_store): Likewise.
18849 * config/rs6000/vsx.md (VSX_TI): New iterator.
18850 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
18851 (*vsx_le_undo_permute_<mode>): Likewise.
18852 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
18853 emit the split sequence.
18854 (*vsx_le_perm_store_<mode>): Likewise.
18856 2017-07-27 Jakub Jelinek <jakub@redhat.com>
18858 PR tree-optimization/81555
18859 PR tree-optimization/81556
18860 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
18861 if true, force CHANGED for the recursive invocation.
18862 (reassociate_bb): Remember original length of ops array, pass
18863 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
18865 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
18866 attributes for noipa attribute. For naked attribute use
18867 lookup_attribute first before lookup_attribute_spec.
18868 * final.c (rest_of_handle_final): Disable IPA RA for functions with
18870 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
18871 for functions with noipa attribute.
18872 (cgraph_externally_visible_p): Return true for functions with noipa
18874 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
18875 for functions with noipa attribute.
18876 * doc/extend.texi: Document noipa function attribute.
18877 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
18878 also for functions with noipa attribute.
18879 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
18881 2017-07-26 Andrew Pinski <apinski@cavium.com>
18883 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
18884 vec_unalign_load_cost and vec_unalign_store_cost.
18886 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
18888 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
18889 -mvsx-small-integer option.
18890 (ISA_3_0_MASKS_IEEE): Likewise.
18891 (OTHER_VSX_VECTOR_MASKS): Likewise.
18892 (POWERPC_MASKS): Likewise.
18893 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
18894 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
18895 code, only testing for DImode being allowed in non-VSX floating
18897 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
18898 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
18900 (rs6000_option_override_internal): Delete -mvsx-small-integer.
18901 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
18902 TARGET_P8_VECTOR test.
18903 (rs6000_secondary_reload_simple_move): Likewise.
18904 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
18905 since TARGET_P9_VECTOR was already tested.
18906 (rs6000_opt_masks): Remove -mvsx-small-integer.
18907 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
18908 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18910 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
18911 test for TARGET_VEXTRACTUB was used, and that uses
18913 (p9 extract splitter): Likewise.
18914 (vsx_extract_<mode>_di_p9): Likewise.
18915 (vsx_extract_<mode>_store_p9): Likewise.
18916 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
18917 for TARGET_P9_VECTOR was used. Delete code that is now dead with
18918 the elimination of TARGET_VSX_SMALL_INTEGER.
18919 (vsx_extract_<mode>_p8): Likewise.
18920 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
18921 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
18922 (vsx_set_<mode>_p9): Likewise.
18923 (vsx_set_v4sf_p9): Likewise.
18924 (vsx_set_v4sf_p9_zero): Likewise.
18925 (vsx_insert_extract_v4sf_p9): Likewise.
18926 (vsx_insert_extract_v4sf_p9_2): Likewise.
18927 * config/rs6000/rs6000.md (sign extend splitter): Change
18928 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
18929 (floatsi<mode>2_lfiwax_mem): Likewise.
18930 (floatunssi<mode>2_lfiwzx_mem): Likewise.
18931 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
18932 since a test for TARGET_P9_VECTOR was used.
18933 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18934 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
18935 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
18936 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18937 TARGET_P8_VECTOR test.
18938 (fix_trunc<mode>si2_stfiwx): Likewise.
18939 (fix_trunc<mode>si2_internal): Likewise.
18940 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
18941 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18943 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18944 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
18945 TARGET_P8_VECTOR test.
18946 (fixuns_trunc<mode>si2_stfiwx): Likewise.
18947 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
18948 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
18950 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
18951 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
18952 since a test for TARGET_P9_VECTOR was used.
18953 (splitter for loading small constants): Likewise.
18955 2017-07-26 Andrew Pinski <apinski@cavium.com>
18957 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
18960 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
18963 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
18964 (fp_valid_at): Likewise.
18966 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18968 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
18969 (qdf24xx_addrcost_table): Likewise.
18970 (cortexa57_tunings): Update to use generic_branch_cost.
18971 (cortexa72_tunings): Likewise.
18972 (cortexa73_tunings): Likewise.
18973 (qdf24xx_tunings): Likewise.
18975 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
18977 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
18978 (thunderx2t99_branch_cost): Likewise.
18979 (cortexa35_tunings): Update to use generic_branch_cost.
18980 (cortexa53_tunings): Likewise.
18981 (cortexa57_tunings): Likewise.
18982 (cortexa72_tunings): Likewise.
18983 (cortexa73_tunings): Likewise.
18984 (thunderx2t99_tunings): Likewise.
18986 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
18988 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
18989 (sparc_option_override): Honour MASK_FSMULD.
18990 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
18991 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
18992 * config/sparc/sparc.opt (mfsmuld): New option.
18993 * doc/invoke.texi (mfsmuld): Document option.
18995 2017-07-26 Marek Polacek <polacek@redhat.com>
18997 PR middle-end/70992
18998 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
19000 2017-07-26 Richard Biener <rguenther@suse.de>
19002 * gimple-match-head.c (do_valueize): Return OP if valueize
19004 (get_def): New helper to get at the def stmt of a SSA name
19005 if valueize allows.
19006 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
19007 do_valueize to get at the def stmt.
19008 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
19010 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
19012 PR middle-end/46932
19013 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
19015 2017-07-26 Martin Liska <mliska@suse.cz>
19018 * function.c (expand_function_start): Make expansion of
19019 nonlocal_goto_save_area after parm_birth_insn.
19021 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
19023 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
19024 from all CPU target flags enable members.
19026 2017-07-26 Richard Biener <rguenther@suse.de>
19028 * genmatch.c (dt_simplify::gen): Make iterator vars const.
19029 (decision_tree::gen): Make 'type' const.
19030 (write_predicate): Likewise.
19032 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19034 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
19035 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
19036 (rs6000_option_override_internal): Likewise.
19037 (rs6000_expand_vector_set): Likewise.
19038 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
19039 (TARGET_UPPER_REGS_SF): Likewise.
19040 (TARGET_UPPER_REGS_DI): Likewise.
19041 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
19042 (TARGET_DIRECT_MOVE_64BIT): Likewise.
19043 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
19044 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
19045 (Splitters for DI constants in Altivec registers): Likewise.
19046 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
19047 (vsx_set_v4sf_p9): Likewise.
19048 (vsx_set_v4sf_p9_zero): Likewise.
19049 (vsx_insert_extract_v4sf_p9): Likewise.
19050 (vsx_insert_extract_v4sf_p9_2): Likewise.
19052 2017-07-25 Carl Love <cel@us.ibm.com>
19054 * doc/extend.texi: Update the built-in documentation file for the
19055 existing built-in functions
19056 vector signed char vec_cnttz (vector signed char);
19057 vector unsigned char vec_cnttz (vector unsigned char);
19058 vector signed short vec_cnttz (vector signed short);
19059 vector unsigned short vec_cnttz (vector unsigned short);
19060 vector signed int vec_cnttz (vector signed int);
19061 vector unsigned int vec_cnttz (vector unsigned int);
19062 vector signed long long vec_cnttz (vector signed long long);
19063 vector unsigned long long vec_cnttz (vector unsigned long long);
19065 2017-07-25 Andrew Pinski <apinski@cavium.com>
19067 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
19068 accesses where the use is for the first operand of a BIT_INSERT.
19070 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19073 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19074 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19076 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
19078 * config/i386/gstabs.h: Delete.
19079 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
19081 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
19083 * config/i386/i386.c (ix86_decompose_address): Do not check for
19084 register RTX when looking at index_reg or base_reg.
19085 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
19087 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
19089 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
19090 to update EH info here.
19092 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19094 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
19096 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19098 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
19100 2017-07-25 Torsten Duwe <duwe@suse.de>
19102 * common.opt: Introduce -fpatchable-function-entry
19103 command line option, and its variables function_entry_patch_area_size
19104 and function_entry_patch_area_start.
19105 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
19106 including a two-value parser.
19107 * target.def (print_patchable_function_entry): New target hook.
19108 * targhooks.h (default_print_patchable_function_entry): New function.
19109 * targhooks.c (default_print_patchable_function_entry): Likewise.
19110 * toplev.c (process_options): Switch off IPA-RA if
19111 patchable function entries are being generated.
19112 * varasm.c (assemble_start_function): Look at the
19113 patchable-function-entry command line switch and current
19114 function attributes and maybe generate NOP instructions by
19115 calling the print_patchable_function_entry hook.
19116 * doc/extend.texi: Document patchable_function_entry attribute.
19117 * doc/invoke.texi: Document -fpatchable_function_entry
19118 command line option.
19119 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
19121 * doc/tm.texi: Re-generate.
19123 2017-07-25 Jakub Jelinek <jakub@redhat.com>
19126 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
19127 TARGET_AVX512DQ rather than TARGET_AVX512BW.
19129 2017-07-25 Tamar Christina <tamar.christina@arm.com>
19131 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
19133 2017-07-25 Richard Biener <rguenther@suse.de>
19135 PR tree-optimization/81455
19136 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
19137 not walk in cycles when looking for guards.
19139 2017-07-25 Richard Biener <rguenther@suse.de>
19141 PR tree-optimization/81529
19142 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
19143 when optimizing backedge uses.
19145 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
19147 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
19149 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
19150 to dl_section_ref. On AIX, append an expression to subtract
19151 the size of the section length to dl_section_ref.
19153 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
19155 * configure.ac: If any of the config.* scripts fail, exit 1.
19156 * configure: Regenerate.
19158 2017-07-25 Richard Biener <rguenther@suse.de>
19160 PR middle-end/81546
19161 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
19162 of immediate uses, be more verbose on errors.
19164 2017-07-25 Richard Biener <rguenther@suse.de>
19166 PR tree-optimization/81510
19167 * tree-vect-loop.c (vect_is_simple_reduction): When the
19168 reduction stmt is not inside the loop bail out.
19170 2017-07-25 Richard Biener <rguenther@suse.de>
19172 PR tree-optimization/81303
19173 * tree-vect-loop-manip.c (vect_loop_versioning): Build
19174 profitability check against LOOP_VINFO_NITERSM1.
19176 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
19178 * domwalk.c (cmp_bb_postorder): Simplify.
19179 (sort_bbs_postorder): New function. Use it...
19180 (dom_walker::walk): ...here to optimize common cases.
19182 2017-07-25 Martin Liska <mliska@suse.cz>
19185 * ipa-visibility.c (function_and_variable_visibility): Make the
19186 redirection just on target that supports aliasing.
19187 Fix GNU coding style.
19189 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
19192 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
19194 * config/arm/rtems.h: Likewise.
19195 * config/bfin/rtems.h: Likewise.
19196 * config/i386/rtemself.h: Likewise.
19197 * config/lm32/rtems.h: Likewise.
19198 * config/m32c/rtems.h: Likewise.
19199 * config/m68k/rtemself.h: Likewise.
19200 * config/microblaze/rtems.h: Likewise.
19201 * config/mips/rtems.h: Likewise.
19202 * config/moxie/rtems.h: Likewise.
19203 * config/nios2/rtems.h: Likewise.
19204 * config/powerpcspe/rtems.h: Likewise.
19205 * config/rs6000/rtems.h: Likewise.
19206 * config/rtems.h: Likewise.
19207 * config/sh/rtems.h: Likewise.
19208 * config/sh/rtemself.h: Likewise.
19209 * config/sparc/rtemself.h: Likewise.
19211 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19214 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
19215 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
19216 * tree-ssa-structalias.c (alias_get_name): Same.
19218 2017-07-25 Bin Cheng <bin.cheng@arm.com>
19221 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
19222 instructions if no du chain is found.
19224 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
19226 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
19228 2017-07-25 Richard Biener <rguenther@suse.de>
19230 PR middle-end/81505
19231 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
19234 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
19236 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
19237 upper-regs options.
19238 (ISA_2_7_MASKS_SERVER): Likewise.
19239 (ISA_3_0_MASKS_IEEE): Likewise.
19240 (OTHER_P8_VECTOR_MASKS): Likewise.
19241 (OTHER_VSX_VECTOR_MASKS): Likewise.
19242 (POWERPC_MASKS): Likewise.
19243 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
19244 duplicate list of options.
19245 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
19246 explicit -mupper-regs options.
19247 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
19248 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
19249 alias for -mupper-regs-df.
19250 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19251 (rs6000_init_hard_regno_mode_ok): Likewise.
19252 (rs6000_option_override_internal): Likewise.
19253 (rs6000_opt_masks): Likewise.
19254 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
19255 options in terms of whether -mvsx or -mpower8-vector was used.
19256 (TARGET_UPPER_REGS_DI): Likewise.
19257 (TARGET_UPPER_REGS_SF): Likewise.
19258 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
19259 -mupper-regs-* options.
19261 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19263 * passes.c (emergency_dump_function): Print some empty lines and a
19264 header before the RTL dump.
19266 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
19268 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
19270 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
19273 * config/aarch64/aarch64.c (aarch64_classify_symbol):
19274 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
19276 2017-07-24 Carl Love <cel@us.ibm.com>
19278 * config/rs6000/rs6000-c.c: Add support for built-in functions
19279 vector float vec_extract_fp32_from_shorth (vector unsigned short);
19280 vector float vec_extract_fp32_from_shortl (vector unsigned short);
19281 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
19282 vec_extract_fp_from_shortl): Add defines for the two builtins.
19283 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
19284 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
19286 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
19287 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
19288 * doc/extend.texi: Update the built-in documentation file for the
19289 new built-in function.
19291 2017-07-24 Jakub Jelinek <jakub@redhat.com>
19294 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
19296 * doc/generic.texi: Likewise.
19297 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
19298 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
19300 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
19302 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
19303 (aarch64_mls_elt_merge<mode>): Likewise.
19305 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
19307 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
19308 having __cxa_atexit.
19310 2017-07-23 Michael Collison <michael.collison@arm.com>
19312 * config/arm/arm.c (arm_option_override): Deprecate
19313 use of -mstructure-size-boundary.
19314 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
19315 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
19317 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19320 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19321 Reduce cost estimate for direct moves.
19323 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
19326 * config/i386/i386.c (ix86_option_override_internal): Disable
19327 BMI, BMI2 and TBM instructions for -m16.
19329 2017-07-21 Carl Love <cel@us.ibm.com>
19331 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19332 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19333 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19334 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19335 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19336 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19337 VMULOSW): New enum "unspec" values.
19338 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19339 altivec_vmulosw): New patterns.
19340 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19341 VMULOSW): Add definitions.
19343 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
19345 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
19346 (qdf24xx): Likewise.
19347 * config/aarch64/aarch64-options-extensions.def (rdma); New.
19348 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
19349 (AARCH64_FL_V8_1): Renumber.
19350 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
19351 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
19352 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
19353 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
19354 rdma to feature modifiers list.
19356 2017-07-21 Yury Gribov <tetra2005@gmail.com>
19358 PR middle-end/56727
19359 * ipa-visibility (function_and_variable_visibility): Convert
19360 recursive PLT call to direct call if appropriate.
19362 2017-07-21 Andrew Pinski <apinski@cavium.com>
19364 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
19365 operand 1 to see if the types precision matches.
19366 * fold-const.c (operand_equal_p): Likewise.
19368 2017-07-21 Richard Biener <rguenther@suse.de>
19370 PR tree-optimization/81303
19371 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
19372 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
19373 (vect_peeling_hash_get_lowest_cost): Adjust.
19374 (vect_enhance_data_refs_alignment): Likewise. Use
19375 vect_get_peeling_costs_all_drs to compute the penalty for no
19376 peeling to match up costs.
19378 2017-07-21 Richard Biener <rguenther@suse.de>
19380 PR tree-optimization/81500
19381 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
19382 we didn't identify a reduction path.
19384 2017-07-21 Tom de Vries <tom@codesourcery.com>
19385 Cesar Philippidis <cesar@codesourcery.com>
19387 PR gcov-profile/81442
19388 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
19391 2017-07-21 Tom de Vries <tom@codesourcery.com>
19394 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
19396 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
19397 nvptx_override_options_after_change.
19399 2017-07-21 Ulrich Drepper <drepper@redhat.com>
19401 * dwarf2out.c (output_file_names): Avoid double testing for
19402 dwarf_version >= 5.
19404 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
19406 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
19408 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19410 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
19412 (try_crossjump_to_edge): Do not punt on partitioned functions.
19414 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
19416 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
19417 Put all BBs reachable only via paths crossing cold region to cold
19419 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
19421 2016-07-21 Richard Biener <rguenther@suse.de>
19423 PR tree-optimization/81303
19424 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
19425 into account prologue and epilogue iterations when raising
19426 min_profitable_iters to sth at least covering one vector iteration.
19428 2017-07-21 Tamar Christina <tamar.christina@arm.com>
19430 * config/arm/arm.c (arm_test_cpu_arch_dat):
19433 2017-07-20 Nathan Sidwell <nathan@acm.org>
19435 Remove TYPE_METHODS.
19436 * tree.h (TYPE_METHODS): Delete.
19437 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
19438 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
19439 (dbxout_type_methods): Scan TYPE_FIELDS.
19440 (dbxout_type): Don't check TYPE_METHODS here.
19441 * function.c (use_register_for_decl): Always ignore register for
19442 class types when not optimizing.
19443 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
19444 * tree.c (free_lang_data_in_type): Stitch out member functions and
19445 templates from TYPE_FIELDS.
19446 (build_distinct_type_copy, verify_type_variant,
19447 verify_type): Member fns are on TYPE_FIELDS.
19448 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
19449 * tree-pretty-print.c (dump_generic_node): Likewise.
19451 2017-07-20 Jakub Jelinek <jakub@redhat.com>
19454 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
19455 V2TImode and V4TImode.
19456 (ix86_expand_vector_extract): Likewise.
19457 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
19458 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
19459 (ssescalarmode): Handle V4TImode and V2TImode.
19460 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
19461 (*vec_extractv2ti, *vec_extractv4ti): New insns.
19462 (VEXTRACTI128_MODE): New mode iterator.
19463 (splitter for *vec_extractv?ti first element): New.
19464 (VEC_INIT_MODE): New mode iterator.
19465 (vec_init<mode>): Consolidate 3 expanders into one using
19466 VEC_INIT_MODE mode iterator.
19468 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19470 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
19471 non_spilled_static_chain_regno_p.
19473 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
19475 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
19477 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
19479 * bb-reorder.c (connect_traces): Allow copying of blocks within
19482 2017-07-20 Richard Biener <rguenther@suse.de>
19484 * gimple.h (gimple_phi_result): Add gphi * overload.
19485 (gimple_phi_result_ptr): Likewise.
19486 (gimple_phi_arg): Likewise. Adjust index assert to only
19487 allow actual argument accesses rather than all slots available
19489 (gimple_phi_arg_def): Add gphi * overload.
19490 * tree-phinodes.c (make_phi_node): Initialize only actual
19492 (resize_phi_node): Clear memory not covered by old node,
19493 do not initialize excess argument slots.
19494 (reserve_phi_args_for_new_edge): Initialize new argument slot
19497 2017-07-20 Bin Cheng <bin.cheng@arm.com>
19499 PR tree-optimization/81388
19501 2016-07-21 Bin Cheng <bin.cheng@arm.com>
19503 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
19504 by removing computation of may_be_zero.
19506 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19507 Tom de Vries <tom@codesourcery.com>
19509 PR middle-end/81030
19510 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
19511 when gimple level profile disagrees with what RTL expander did.
19513 2017-07-20 Richard Biener <rguenther@suse.de>
19515 PR tree-optimization/61171
19516 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
19517 (vect_analyze_stmt): Add slp instance parameter.
19518 (vectorizable_reduction): Likewise.
19519 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
19520 (vect_is_simple_reduction): Deal with chains not detected
19521 as SLP reduction chain, specifically not properly associated
19522 chains containing a mix of plus/minus.
19523 (get_reduction_op): Remove.
19524 (get_initial_defs_for_reduction): Simplify, pass in whether
19525 this is a reduction chain, pass in the SLP node for the PHIs.
19526 (vect_create_epilog_for_reduction): Get the SLP instance as
19528 (vectorizable_reduction): Get the SLP instance as arg.
19529 During analysis remember the SLP node with the PHIs in the
19530 instance. Simplify getting at the vectorized reduction PHIs.
19531 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
19532 through SLP instance.
19533 (vect_slp_analyze_operations): Likewise.
19534 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
19535 (vect_transform_stmt): Likewise.
19537 2017-07-20 Tom de Vries <tom@codesourcery.com>
19539 PR tree-optimization/81489
19540 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
19541 read of phi arg location to before loop that modifies phi.
19543 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19545 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
19548 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19550 PR middle-end/81331
19551 * except.c (execute): Fix ordering issue.
19553 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19555 PR rtl-optimization/81423
19556 * combine.c (make_compound_operation_int): Don't try to optimize
19557 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
19559 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
19561 PR rtl-optimization/81423
19562 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
19563 with a constant that is -1 in the truncated to mode.
19565 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19567 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
19568 (determine_unlikely_bbs): ... here.
19569 * predict.h (propagate_unlikely_bbs_forward): Declare.
19570 * cfgexpand.c (pass_expand::execute): Use it.
19571 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
19573 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
19574 propagate_unlikely_bbs_forward.
19576 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
19578 PR middle-end/81331
19579 * except.c (maybe_add_nop_after_section_switch): New function.
19582 2017-07-19 Tom de Vries <tom@codesourcery.com>
19584 * gimple.h (gimple_phi_set_arg): Make assert more strict.
19586 2017-07-19 Tom de Vries <tom@codesourcery.com>
19588 * gimple.h (gimple_phi_arg): Make assert more strict.
19590 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
19592 * config.gcc (powerpc*-*-*): Add mmintrin.h.
19593 * config/rs6000/mmintrin.h: New file.
19594 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
19596 2017-07-19 Jakub Jelinek <jakub@redhat.com>
19598 PR tree-optimization/81346
19599 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
19601 2017-07-19 Tom de Vries <tom@codesourcery.com>
19603 * config/nvptx/nvptx.md (VECIM): Add V2DI.
19605 2017-07-19 Tom de Vries <tom@codesourcery.com>
19607 * config/nvptx/nvptx-modes.def: Add V2DImode.
19608 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
19609 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
19610 (nvptx_output_mov_insn): Handle lack of mov.b128.
19611 (nvptx_print_operand): Handle 'H' and 'L' codes.
19612 (nvptx_vector_mode_supported): Allow V2DImode.
19613 (nvptx_preferred_simd_mode): New function.
19614 (nvptx_data_alignment): New function.
19615 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
19616 nvptx_preferred_simd_mode.
19617 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
19619 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
19621 2017-07-19 Tom de Vries <tom@codesourcery.com>
19623 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
19624 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
19625 (nvptx_vector_mode_supported): New function. Allow V2SImode.
19626 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
19627 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
19628 (mov<VECIM>_insn): New define_insn.
19629 (define_expand "mov<VECIM>): New define_expand.
19631 2017-07-19 Tom de Vries <tom@codesourcery.com>
19633 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
19635 2017-07-19 Jakub Jelinek <jakub@redhat.com>
19637 PR tree-optimization/81346
19638 * fold-const.h (fold_div_compare, range_check_type): Declare.
19639 * fold-const.c (range_check_type): New function.
19640 (build_range_check): Use range_check_type.
19641 (fold_div_compare): No longer static, rewritten into
19642 a match.pd helper function.
19643 (fold_comparison): Don't call fold_div_compare here.
19644 * match.pd (X / C1 op C2): New optimization using fold_div_compare
19645 as helper function.
19647 2017-07-19 Nathan Sidwell <nathan@acm.org>
19649 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
19650 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
19651 * tree.c (find_decls_types_r, verify_type): Use
19652 TYPE_{MIN,MAX}_VALUE_RAW.
19653 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
19654 (hash_tree): Likewise.
19655 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19657 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19660 2017-07-18 Tom de Vries <tom@codesourcery.com>
19662 PR middle-end/81464
19663 * omp-expand.c (expand_omp_for_static_chunk): Handle
19664 equal-argument loop exit phi.
19666 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
19669 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
19670 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
19671 operand 2 predicate.
19672 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
19673 operand 2 predicate.
19674 (ror,rol -> rorx splitters): Use const_int_operand as
19675 operand 2 predicate.
19677 2017-06-18 Richard Biener <rguenther@suse.de>
19679 PR tree-optimization/81410
19680 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
19681 the gap in the ! slp_perm SLP case after each group.
19683 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19685 PR middle-end/81463
19686 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
19689 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
19691 PR middle-end/81462
19692 * predict.c (set_even_probabilities): Cleanup; do not affect
19693 probabilities that are already known.
19694 (combine_predictions_for_bb): Call even when count is set.
19696 2017-07-18 Nathan Sidwell <nathan@acm.org>
19698 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
19701 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19704 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
19705 optimization for loop niter analysis.
19707 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
19710 * config/avr/avr.c (avr_optimize_casesi): Don't use
19711 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
19713 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
19715 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
19716 body_cost_vec from _vect_peel_extended_info.
19717 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
19718 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
19721 2017-07-18 Bin Cheng <bin.cheng@arm.com>
19723 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
19725 2017-07-18 Richard Biener <rguenther@suse.de>
19727 PR tree-optimization/80620
19728 PR tree-optimization/81403
19729 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
19730 info when re-using a VN table entry.
19732 2017-07-18 Richard Biener <rguenther@suse.de>
19734 PR tree-optimization/81418
19735 * tree-vect-loop.c (vectorizable_reduction): Properly compute
19736 vectype_in. Verify that with lane-reducing reduction operations
19737 we have a single def-use cycle.
19739 2017-07-17 Carl Love <cel@us.ibm.com>
19741 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
19743 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19744 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19745 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19746 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19747 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19748 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19749 VMULOSW): New enum "unspec" values.
19750 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19751 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19752 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19753 altivec_vmulosw): New patterns.
19754 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19755 VMULOSW): Add definitions.
19756 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
19758 * config/alpha/alpha.c: Include predict.h.
19760 2017-07-17 Yury Gribov <tetra2005@gmail.com>
19762 * tree-vrp.c (compare_assert_loc): Fix comparison function
19763 to return predictable results.
19765 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19767 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
19769 (subdi3): Likewise.
19770 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
19771 * doc/invoke.texi (mexpand-adddi): Update text.
19773 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19775 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
19776 that also clobbers the CC register. The old expand code is moved
19778 (*arc_clzsi2): ... here.
19779 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
19780 the CC register. The old expand code is moved to ...
19781 (arc_ctzsi2): ... here.
19783 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19785 * config/arc/arc.opt (mindexed-loads): Use initial value
19786 TARGET_INDEXED_LOADS_DEFAULT.
19787 (mauto-modify-reg): Use initial value
19788 TARGET_AUTO_MODIFY_REG_DEFAULT.
19789 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19790 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19791 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
19792 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
19794 2017-07-17 Martin Liska <mliska@suse.cz>
19797 * opts.c (finish_options): Do not allow -fgnu-tm
19798 w/ -fsanitize={kernel-,}address. Say sorry.
19800 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19803 * tree-loop-distribution.c (classify_partition): Only assert on
19804 numer of iterations.
19805 (merge_dep_scc_partitions): Delete prameter. Update function call.
19806 (distribute_loop): Remove code handling loop with unknown niters.
19807 (pass_loop_distribution::execute): Skip loop with unknown niters.
19809 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19812 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
19813 function sort_partitions_by_post_order.
19815 2017-07-17 Bin Cheng <bin.cheng@arm.com>
19817 PR tree-optimization/81374
19818 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
19819 the max index of basic blocks, rather than number of basic blocks.
19821 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19823 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
19825 (arc_legitimate_pic_operand_p): Likewise.
19826 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
19828 (arc_needs_pcl_p): Likewise.
19829 (arc_legitimate_pc_offset_p): Likewise.
19830 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
19831 function is also used in constrains.md.
19832 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
19833 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
19834 PLUS. Only return true/false in known cases, otherwise assert.
19835 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
19836 is already called in arc_legitimate_constant_p.
19837 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
19839 (LEGITIMATE_PIC_OPERAND_P): Use
19840 arc_raw_symbolic_reference_mentioned_p function.
19841 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
19846 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
19847 Andrew Burgess <andrew.burgess@embecosm.com>
19849 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
19850 (arc_return_address_register): New function.
19851 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
19852 (arc_handle_fndecl_attribute): Add naked attribute.
19853 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19854 (TARGET_WARN_FUNC_RETURN): Likewise.
19855 (arc_allocate_stack_slots_for_args): New function.
19856 (arc_warn_func_return): Likewise.
19857 (machine_function): Change type fn_type.
19858 (arc_compute_function_type): Consider new naked function type,
19859 change function return type.
19860 (arc_must_save_register): Adapt to handle new
19861 arc_compute_function_type's return type.
19862 (arc_expand_prologue): Likewise.
19863 (arc_expand_epilogue): Likewise.
19864 (arc_return_address_regs): Delete.
19865 (arc_return_address_register): New function.
19866 (arc_epilogue_uses): Use above function.
19867 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
19868 (arc_function_type): Change encoding, add naked type.
19869 (ARC_INTERRUPT_P): Change to handle the new encoding.
19870 (ARC_FAST_INTERRUPT_P): Likewise.
19871 (ARC_NORMAL_P): Define.
19872 (ARC_NAKED_P): Likewise.
19873 (arc_compute_function_type): Delete prototype.
19874 * config/arc/arc.md (in_ret_delay_slot): Use
19875 arc_return_address_register function.
19876 (simple_return): Likewise.
19877 (p_return_i): Likewise.
19879 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19881 PR tree-optimization/81428
19882 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
19883 can't be built for those types.
19885 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19887 Remove stuff dead since r239246.
19889 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
19890 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
19891 (avr_inform_devices): Remove dead stuff.
19893 2017-07-17 Tamar Christina <tamar.christina@arm.com>
19895 * config/arm/arm_neon.h: Fix softp typo.
19897 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19899 PR tree-optimization/81365
19900 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
19901 aggregate moves onto bb predecessor edges, make sure there are no
19902 loads that could alias the lhs in between the start of bb and the
19905 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
19908 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
19909 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
19910 [LSHIFTRT, outer_code = TRUNCATE]: Same.
19912 2017-07-17 Jakub Jelinek <jakub@redhat.com>
19914 PR tree-optimization/81396
19915 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
19916 (init_symbolic_number): Initialize it to 1.
19917 (perform_symbolic_merge): Add n_ops from both operands into the new
19919 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
19920 without base_addr as useless if they need more than one operation.
19921 (bswap_replace): Handle !bswap case for NULL base_addr.
19923 2017-07-17 Tom de Vries <tom@codesourcery.com>
19926 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
19929 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
19931 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
19932 conditional builtin define __FIX_LEON3FT_B2BST.
19934 2017-07-17 Daniel Cederman <cederman@gaisler.com>
19936 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
19937 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
19940 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
19942 PR rtl-optimization/81424
19943 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
19944 to remove potential trapping from operands if -fnon-call-exceptions.
19946 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19948 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
19949 profile_proability for scalling.
19950 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
19952 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19954 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
19956 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19958 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
19959 fixpoint arithmetics.
19961 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19963 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
19964 fixpoint arithmetics.
19966 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19968 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
19969 fixpoint arithmetics.
19971 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
19973 * profile-count.h (profile_probability::from_reg_br_prob_note,
19974 profile_probability::to_reg_br_prob_note): New functions.
19975 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
19976 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
19977 * predict.c (probability_reliable_p): Update.
19978 (edge_probability_reliable_p): Update.
19979 (br_prob_note_reliable_p): Update.
19980 (invert_br_probabilities): Update.
19981 (add_reg_br_prob_note): New function.
19982 (combine_predictions_for_insn): Update.
19983 * asan.c (asan_clear_shadow): Update.
19984 * cfgbuild.c (compute_outgoing_frequencies): Update.
19985 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
19986 (update_br_prob_note): Update.
19987 (rtl_verify_edges): Update.
19988 (purge_dead_edges): Update.
19989 (fixup_reorder_chain): Update.
19990 * emit-rtl.c (try_split): Update.
19991 * ifcvt.c (cond_exec_process_insns): Update.
19992 (cond_exec_process_if_block): Update.
19993 (dead_or_predicable): Update.
19994 * internal-fn.c (expand_addsub_overflow): Update.
19995 (expand_neg_overflow): Update.
19996 (expand_mul_overflow): Update.
19997 * loop-doloop.c (doloop_modify): Update.
19998 * loop-unroll.c (compare_and_jump_seq): Update.
19999 * optabs.c (emit_cmp_and_jump_insn_1): Update.
20000 * predict.h: Update.
20001 * reorg.c (mostly_true_jump): Update.
20003 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
20004 * config/alpha/alpha.c (emit_unlikely_jump): Update.
20005 * config/arc/arc.c: (emit_unlikely_jump): Update.
20006 * config/arm/arm.c: (emit_unlikely_jump): Update.
20007 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
20008 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
20009 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
20010 (ix86_print_operand): Update.
20011 (ix86_split_fp_branch): Update.
20012 (predict_jump): Update.
20013 * config/ia64/ia64.c (ia64_print_operand): Update.
20014 * config/mmix/mmix.c (mmix_print_operand): Update.
20015 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
20016 (rs6000_expand_split_stack_prologue): Update.
20017 * config/rs6000/rs6000.c: Update.
20018 * config/s390/s390.c (s390_expand_vec_strlen): Update.
20019 (s390_expand_vec_movstr): Update.
20020 (s390_expand_cs_tdsi): Update.
20021 (s390_expand_split_stack_prologue): Update.
20022 * config/sh/sh.c (sh_print_operand): Update.
20023 (expand_cbranchsi4): Update.
20024 (expand_cbranchdi4): Update.
20025 * config/sparc/sparc.c (output_v9branch): Update.
20026 * config/spu/spu.c (get_branch_target): Update.
20027 (ea_load_store_inline): Update.
20028 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
20029 * config/tilepro/tilepro.c: Update.
20031 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
20033 * gimplify.c (mostly_copy_tree_r): Revert latest change.
20034 (gimplify_save_expr): Likewise.
20036 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20038 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
20040 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20042 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
20044 * timevar.def (TV_IPA_FNSUMMARY): Define.
20046 2017-07-16 Daniel Cederman <cederman@gaisler.com>
20048 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
20049 to back store errata sensitive sequence from being generated.
20050 (sqrtdf2_fix): Likewise.
20052 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20054 * tree-ssa-threadupdate.c (compute_path_counts,
20055 update_joiner_offpath_counts): Use profile_probability.
20057 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
20060 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20062 * config/arm/arm-c.c (arm_cpu_builtins): Define
20063 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20065 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
20067 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20068 array entries to represent __ieee128 versions of the
20069 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
20070 scalar_extract_sig, and scalar_insert_exp built-in functions.
20071 (altivec_resolve_overloaded_builtin): Add special case handling
20072 for the __builtin_scalar_insert_exp function, as represented by
20073 the P9V_BUILTIN_VEC_VSIEDP constant.
20074 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
20075 exponent support for __ieee128 argument.
20076 (VSESQP): Add scalar extract signature support for __ieee128
20078 (VSTDCNQP): Add scalar test negative support for __ieee128
20080 (VSIEQP): Add scalar insert exponent support for __int128 argument
20081 with __ieee128 result.
20082 (VSIEQPF): Add scalar insert exponent support for __ieee128
20083 argument with __ieee128 result.
20084 (VSTDCQP): Add scalar test data class support for __ieee128
20086 (VSTDCNQP): Add overload support for scalar test negative with
20087 __ieee128 argument.
20088 (VSTDCQP): Add overload support for scalar test data class
20089 __ieee128 argument.
20090 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
20091 UNSPEC_VSX_SXSIGDP.
20092 (UNSPEC_VSX_SIEXPQP): New constant.
20093 (xsxexpqp): New insn for VSX scalar extract exponent quad
20095 (xsxsigqp): New insn for VSX scalar extract significand quad
20097 (xsiexpqpf): New insn for VSX scalar insert exponent quad
20098 precision with floating point argument.
20099 (xststdcqp): New expand for VSX scalar test data class quad
20101 (xststdcnegqp): New expand for VSX scalar test negative quad
20103 (xststdcqp): New insn to match expansions for VSX scalar test data
20104 class quad precision and VSX scalar test negative quad precision.
20105 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
20106 special case operand checking to enforce that second operand of
20107 VSX scalar test data class with quad precision argument is a 7-bit
20109 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20110 prototypes and descriptions of __ieee128 versions of
20111 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
20112 scalar_test_data_class, and scalar_test_neg built-in functions.
20114 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20116 PR tree-optimization/81162
20117 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
20118 replace a negate with an add.
20120 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
20122 * doc/invoke.texi (arm/-mcpu): Document +crypto.
20124 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20126 * config/arm/arm-c.c (arm_cpu_builtins): Define
20127 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
20129 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20131 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
20132 (armv8-r): Set ARM Cortex-R52 as default CPU.
20133 * config/arm/arm-tables.opt: Regenerate.
20134 * config/arm/arm-tune.md: Regenerate.
20135 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
20137 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
20138 extension for -mcpu=cortex-r52.
20140 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
20142 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
20143 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
20144 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
20145 (fp-armv8): Define it as FP_ARMv8 only.
20146 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
20147 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
20149 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
20150 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
20151 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
20152 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
20153 than TARGET_FPU_ARMV8.
20154 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
20155 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
20156 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
20158 * config/arm/neon.md (neon_vrint): Likewise.
20159 (neon_vcvt): Likewise.
20160 (neon_<fmaxmin_op><mode>): Likewise.
20161 (<fmaxmin><mode>3): Likewise.
20162 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
20163 * config/arm/predicates.md (arm_cond_move_operator): Check against
20164 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
20166 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
20168 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
20169 to top of function.
20171 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20173 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
20174 loop in comment with memset.
20176 2017-07-14 Martin Liska <mliska@suse.cz>
20178 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
20179 * dwarf2out.c (is_java): Remove the function.
20180 (output_pubname): Remove usage of the function.
20181 (lower_bound_default): Remove usage of DW_LANG_Java.
20182 (gen_compile_unit_die): Likewise.
20183 * gcc.c: Remove compiler defaults for .java and .zip files.
20184 * gimple-expr.c (remove_suffix): Change as there's no longer
20185 extension than 4-letter one.
20186 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
20187 (gimplify_save_expr): Likewise.
20188 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
20189 as it's possible even for other languages than Java.
20190 * langhooks.h (struct lang_hooks): Remove Java from a comment.
20191 * lto-opts.c (lto_write_options): Remove reference to Java.
20192 * opts.c (strip_off_ending): Update file extension handling.
20193 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
20194 * tree-eh.c (lower_resx): Likewise.
20195 * tree.c (free_lang_data_in_type): Remove dead code.
20196 (find_decls_types_r): Likewise.
20197 (build_common_builtin_nodes): Remove Java from a comment.
20198 (verify_type): Remove dead code.
20199 * varasm.c (assemble_external): Remove Java from a comment.
20201 2017-07-14 Martin Liska <mliska@suse.cz>
20203 * opts.c (finish_options): Add quotes.
20204 (common_handle_option): Likewise.
20206 2017-07-14 Martin Liska <mliska@suse.cz>
20208 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
20209 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
20210 Remove N_SO_PASCAL.
20211 * dwarf2out.c (lower_bound_default): Do not handle
20213 (gen_compile_unit_die): Likewise.
20214 * gcc.c: Remove default extension binding for GNU Pascal.
20215 * stmt.c: Remove Pascal language from a comment.
20216 * xcoffout.c: Likewise.
20218 2017-07-13 David Malcolm <dmalcolm@redhat.com>
20221 * diagnostic-show-locus.c (fixit_cmp): New function.
20222 (layout::layout): Sort m_fixit_hints.
20223 (column_range::column_range): Assert that the values are valid.
20224 (struct char_span): New struct.
20225 (correction::overwrite): New method.
20226 (struct source_line): New struct.
20227 (line_corrections::add_hint): Add assertions. Reimplement memcpy
20228 calls in terms of classes source_line and char_span, and
20229 correction::overwrite.
20230 (selftest::test_overlapped_fixit_printing_2): New function.
20231 (selftest::diagnostic_show_locus_c_tests): Call it.
20233 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
20235 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
20236 early if there is no lhs.
20238 2017-07-13 Martin Liska <mliska@suse.cz>
20240 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
20241 (gen_reference_type_die): Likewise.
20242 * stor-layout.c: Remove Pascal-related comment.
20244 2017-07-13 Martin Liska <mliska@suse.cz>
20246 * opts.c (finish_options): Add quotes to error messages.
20247 (parse_sanitizer_options): Likewise.
20249 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
20251 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
20253 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
20255 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
20257 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
20259 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
20261 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
20263 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
20266 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
20267 provides the hardware capability bits, define the macro
20268 __BUILTIN_CPU_SUPPORTS__.
20269 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
20270 if GLIBC does not provide the hardware capability bits. Add a
20271 gcc_unreachable call if the built-in cpu function is neither
20272 __builtin_cpu_is nor __builtin_cpu_supports.
20273 (rs6000_get_function_versions_dispatcher): Change the warning
20274 that an old GLIBC is used which does not export the capability
20275 bits to be an error.
20276 * doc/extend.texi (target_clones attribute): Document the
20277 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
20278 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
20279 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
20280 the macros defined by GCC if the newer GLIBC is available.
20282 2017-07-12 Jeff Law <law@redhat.com>
20284 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
20285 remaining includes slightly.
20286 * config/riscv/riscv-builtins.c: Include profile-count.h.
20288 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20291 * config/avr/avr.c (avr_set_current_function): In diagnostic
20292 messages: Quote keywords and (parts of) identifiers.
20293 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
20296 2017-07-12 Carl Love <cel@us.ibm.com>
20298 * config/rs6000/rs6000-c.c: Add support for built-in functions
20299 vector bool char vec_revb (vector bool char);
20300 vector bool short vec_revb (vector short char);
20301 vector bool int vec_revb (vector bool int);
20302 vector bool long long vec_revb (vector bool long long);
20303 * doc/extend.texi: Update the built-in documentation file for the
20304 new built-in functions.
20306 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20308 * config/s390/s390.md: Remove movcc splitter.
20310 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20312 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
20313 load/store on condition.
20315 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
20318 * config/avr/avr.c (avr_encode_section_info)
20319 [progmem && !TREE_READONLY]: Error if progmem object needs
20322 2017-07-11 Michael Collison <michael.collison@arm.com>
20324 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
20327 2017-07-11 Carl Love <cel@us.ibm.com>
20329 * config/rs6000/rs6000-c.c: Add support for builtins
20330 vector unsigned int vec_parity_lsbb (vector signed int);
20331 vector unsigned int vec_parity_lsbb (vector unsigned int);
20332 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
20333 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
20334 vector unsigned long long vec_parity_lsbb (vector signed long long);
20335 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
20336 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
20337 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
20338 * doc/extend.texi: Update the built-in documentation file for the
20339 new built-in functions.
20341 2017-07-11 David Malcolm <dmalcolm@redhat.com>
20343 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
20344 (layout::m_primary_loc): New field.
20345 (layout::layout): Initialize new field. Move location filtering
20346 logic from here to...
20347 (layout::maybe_add_location_range): ...this new method. Add
20348 support for filtering to just the lines already specified by other
20350 (layout::will_show_line_p): New method.
20351 (gcc_rich_location::add_location_if_nearby): New method.
20352 (selftest::test_add_location_if_nearby): New test function.
20353 (selftest::diagnostic_show_locus_c_tests): Call it.
20354 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
20357 2017-07-11 Tom de Vries <tom@codesourcery.com>
20359 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
20360 (bb_first_real_insn): New function.
20361 (nvptx_single): Add extra initialization of broadcasted condition
20364 2017-07-11 Nathan Sidwell <nathan@acm.org>
20366 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
20368 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
20370 * doc/extend.texi (AVR Function Attributes): Remove weblink to
20371 Binutils doc as TEXI will mess them up.
20372 * doc/invoke.texi (AVR Options): Same here.
20374 2017-07-11 Daniel Cederman <cederman@gaisler.com>
20376 * config/sparc/sparc.opt (mfix-ut700): New option.
20377 (mfix-gr712rc): Likewise.
20378 (sparc_fix_b2bst): New variable.
20379 * doc/invoke.texi (SPARC options): Document them.
20380 (ARM options): Fix warnings.
20381 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
20382 instructions to prevent sequences that can trigger the store-store
20383 errata for certain LEON3FT processors.
20384 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
20385 (sparc_option_override): Set sparc_fix_b2bst appropriately.
20386 * config/sparc/sparc.md (fix_b2bst): New attribute.
20387 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
20389 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
20392 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
20394 (*rsqrtsf2_sse): Ditto.
20396 (div<mode>3): Macroize insn from divdf3 and divsf3
20397 using MODEF mode iterator.
20399 2017-07-10 Martin Sebor <msebor@redhat.com>
20401 PR tree-optimization/80397
20402 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
20403 instead of testing for equality to INTEGER_TYPE.
20405 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
20407 * config.gcc: Remove uclibc from arc target spec.
20409 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
20411 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
20413 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
20416 * lto-wrapper.c (remove_option): New function.
20417 (merge_and_complain): Merge PIC/PIE options more realistically.
20419 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20421 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
20425 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
20426 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
20427 * config.in: Regenerate.
20428 * configure: Regenerate.
20429 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
20430 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
20431 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
20432 (TARGET_GASISR_PROLOGUES): ...target mask.
20433 * common/config/avr/avr-common.c
20434 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
20435 Set -mgas-isr-prologues.
20436 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
20437 INSERT_PASS_BEFORE for it.
20438 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
20439 * config/avr/avr.c (avr_option_override)
20440 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
20441 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
20442 (avr_attribute_table) <no_gccisr>: Add new function attribute.
20443 (avr_set_current_function) <is_no_gccisr>: Init machine field.
20444 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
20446 (make_avr_pass_pre_proep): New function.
20447 (emit_push_sfr) <treg>: Add argument to function and use it
20448 instead of TMP_REG.
20449 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
20450 and set machine->gasisr.yes.
20451 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
20452 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
20453 __gcc_isr.n_pushed to .L__stack_usage.
20454 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
20455 (avr_asm_final_postscan_insn): ...this new static function.
20456 * config/avr/avr.h (machine_function)
20457 <is_no_gccisr, use_L__stack_usage>: New fields.
20458 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
20459 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
20460 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
20461 (gasisr, *gasisr): New expander and insn.
20462 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20463 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
20464 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
20466 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
20468 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
20471 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
20473 Move jump-tables out of .text again.
20476 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
20477 (ASM_OUTPUT_ADDR_VEC): New function.
20478 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
20479 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
20480 INSN_ADDRESSes as asm comment.
20481 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
20482 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
20483 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
20484 * config/avr/avr.md (*tablejump): Adjust comment.
20485 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
20486 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
20488 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
20489 (avr_output_addr_vec): New proto.
20490 (avr_log_t) <insn_addresses>: New field.
20492 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
20495 * config/i386/i386.c (ix86_function_arg_advance): Set
20496 outgoing_args_on_stack to true if there are outgoing arguments
20498 (ix86_function_arg): Likewise.
20499 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
20500 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
20501 * config/i386/i386.h (machine_function): Add
20502 outgoing_args_on_stack.
20504 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
20506 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20507 supporting pthreds.
20508 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
20510 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
20512 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
20513 (REAL_H): Remove $(MACHMODE_H).
20514 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
20516 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
20517 $(MACHMODE_H) and double-int.h.
20518 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
20520 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
20523 2017-07-07 Andrew Pinski <apinski@cavium.com>
20525 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
20526 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
20528 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20530 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
20531 Add warning if GCC was not configured to link against a GLIBC that
20532 exports the hardware capability bits.
20533 (make_resolver_func): Make resolver function private and not a
20534 COMDAT function. Create the name with clone_function_name instead
20535 of make_unique_name.
20538 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
20539 correct operand in doing the split.
20541 2017-07-07 Carl Love <cel@us.ibm.com>
20543 * config/rs6000/rs6000-c: Add support for built-in function
20544 vector unsigned short vec_pack_to_short_fp32 (vector float,
20546 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
20547 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
20548 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
20549 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
20550 (convert_4f32_8i16): Add define_expand.
20551 * doc/extend.texi: Update the built-in documentation file for the
20552 new built-in function.
20554 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20556 * config/sparc/m8.md: New file.
20557 * config/sparc/sparc.md: Include m8.md.
20559 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20561 * config/sparc/sparc.opt: New option -mvis4b.
20562 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
20563 (sparc_option_override): Handle VIS4B.
20564 (enum sparc_builtins): Define
20565 SPARC_BUILTIN_DICTUNPACK{8,16,32},
20566 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
20567 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
20568 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
20569 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
20570 (check_constant_argument): New function.
20571 (sparc_vis_init_builtins): Define builtins
20572 __builtin_vis_dictunpack{8,16,32},
20573 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
20574 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
20575 __builtin_vis_fpcmpde{8,16,32}shl and
20576 __builtin_vis_fpcmpur{8,16,32}shl.
20577 (sparc_expand_builtin): Check that the constant operands to
20578 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
20579 constant and in range.
20580 * config/sparc/sparc-c.c (sparc_target_macros): Handle
20582 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
20583 (SPARC_IMM5_P): Likewise.
20584 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
20585 (enabled): Handle vis4b.
20586 (UNSPEC_DICTUNPACK): New unspec.
20587 (UNSPEC_FPCMPSHL): Likewise.
20588 (UNSPEC_FPUCMPSHL): Likewise.
20589 (UNSPEC_FPCMPDESHL): Likewise.
20590 (UNSPEC_FPCMPURSHL): Likewise.
20591 (cpu_feature): New CPU feature `vis4b'.
20592 (dictunpack{8,16,32}): New insns.
20593 (FPCSMODE): New mode iterator.
20594 (fpcscond): New code iterator.
20595 (fpcsucond): Likewise.
20596 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
20597 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
20598 (fpcmpde{8,16,32}{si,di}shl): Likewise.
20599 (fpcmpur{8,16,32}{si,di}shl): Likewise.
20600 * config/sparc/constraints.md: Define constraints `q' for unsigned
20601 2-bit integer constants and `t' for unsigned 5-bit integer
20603 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
20605 (imm5_operand_dictunpack16): Likewise.
20606 (imm5_operand_dictunpack32): Likewise.
20607 (imm2_operand): Likewise.
20608 * doc/invoke.texi (SPARC Options): Document -mvis4b.
20609 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
20610 ditunpack* and fpcmp*shl builtins.
20612 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20614 * config.gcc: Handle m8 in --with-{cpu,tune} options.
20615 * config.in: Add HAVE_AS_SPARC6 define.
20616 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
20618 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
20620 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
20621 (CPP_CPU_SPEC): Handle m8.
20622 (ASM_CPU_SPEC): Likewise.
20623 * config/sparc/sparc-opts.h (enum processor_type): Add
20625 * config/sparc/sparc.c (m8_costs): New struct.
20626 (sparc_option_override): Handle TARGET_CPU_m8.
20627 (sparc32_initialize_trampoline): Likewise.
20628 (sparc64_initialize_trampoline): Likewise.
20629 (sparc_issue_rate): Likewise.
20630 (sparc_register_move_cost): Likewise.
20631 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
20632 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
20633 (ASM_CPU64_DEFAULT_SPEC): Likewise.
20634 (CPP_CPU_SPEC): Handle M8.
20635 (ASM_CPU_SPEC): Likewise.
20636 (AS_M8_FLAG): Define.
20637 * config/sparc/sparc.md: Add m8 to the cpu attribute.
20638 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
20639 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
20641 * configure: Regenerate.
20642 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
20645 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20647 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
20649 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
20650 ("*movdi_insn_sp32"): Do not set v3pipe.
20651 ("*movsi_insn"): Likewise.
20652 ("*movdi_insn_sp64"): Likewise.
20653 ("*movsf_insn"): Likewise.
20654 ("*movdf_insn_sp32"): Likewise.
20655 ("*movdf_insn_sp64"): Likewise.
20656 ("*zero_extendsidi2_insn_sp64"): Likewise.
20657 ("*sign_extendsidi2_insn"): Likewise.
20658 ("*mov<VM32:mode>_insn"): Likewise.
20659 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20660 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20661 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20662 ("<vlop:code><VL:mode>3"): Likewise.
20663 ("*not_<vlop:code><VL:mode>3"): Likewise.
20664 ("*nand<VL:mode>_vis"): Likewise.
20665 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
20666 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
20667 ("one_cmpl<VL:mode>2"): Likewise.
20668 ("faligndata<VM64:mode>_vis"): Likewise.
20669 ("alignaddrsi_vis"): Likewise.
20670 ("alignaddrdi_vis"): Likweise.
20671 ("alignaddrlsi_vis"): Likewise.
20672 ("alignaddrldi_vis"): Likewise.
20673 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20674 ("bmaskdi_vis"): Likewise.
20675 ("bmasksi_vis"): Likewise.
20676 ("bshuffle<VM64:mode>_vis"): Likewise.
20677 ("cmask8<P:mode>_vis"): Likewise.
20678 ("cmask16<P:mode>_vis"): Likewise.
20679 ("cmask32<P:mode>_vis"): Likewise.
20680 ("pdistn<P:mode>_vis"): Likewise.
20681 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20683 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20685 * config/sparc/sparc.md ("subtype"): New insn attribute.
20686 ("*wrgsr_sp64"): Set insn subtype.
20687 ("*rdgsr_sp64"): Likewise.
20688 ("alignaddrsi_vis"): Likewise.
20689 ("alignaddrdi_vis"): Likewise.
20690 ("alignaddrlsi_vis"): Likewise.
20691 ("alignaddrldi_vis"): Likewise.
20692 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
20693 ("fexpand_vis"): Likewise.
20694 ("fpmerge_vis"): Likewise.
20695 ("faligndata<VM64:mode>_vis"): Likewise.
20696 ("bshuffle<VM64:mode>_vis"): Likewise.
20697 ("cmask8<P:mode>_vis"): Likewise.
20698 ("cmask16<P:mode>_vis"): Likewise.
20699 ("cmask32<P:mode>_vis"): Likewise.
20700 ("fchksm16_vis"): Likewise.
20701 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
20702 ("fmean16_vis"): Likewise.
20703 ("fp<plusminus_insn>64_vis"): Likewise.
20704 ("<plusminus_insn>v8qi3"): Likewise.
20705 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
20706 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
20707 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
20708 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
20709 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
20710 ("*movqi_insn"): Likewise.
20711 ("*movhi_insn"): Likewise.
20712 ("*movsi_insn"): Likewise.
20713 ("movsi_pic_gotdata_op"): Likewise.
20714 ("*movdi_insn_sp32"): Likewise.
20715 ("*movdi_insn_sp64"): Likewise.
20716 ("movdi_pic_gotdata_op"): Likewise.
20717 ("*movsf_insn"): Likewise.
20718 ("*movdf_insn_sp32"): Likewise.
20719 ("*movdf_insn_sp64"): Likewise.
20720 ("*zero_extendhisi2_insn"): Likewise.
20721 ("*zero_extendqihi2_insn"): Likewise.
20722 ("*zero_extendqisi2_insn"): Likewise.
20723 ("*zero_extendqidi2_insn"): Likewise.
20724 ("*zero_extendhidi2_insn"): Likewise.
20725 ("*zero_extendsidi2_insn_sp64"): Likewise.
20726 ("ldfsr"): Likewise.
20727 ("prefetch_64"): Likewise.
20728 ("prefetch_32"): Likewise.
20729 ("tie_ld32"): Likewise.
20730 ("tie_ld64"): Likewise.
20731 ("*tldo_ldub_sp32"): Likewise.
20732 ("*tldo_ldub1_sp32"): Likewise.
20733 ("*tldo_ldub2_sp32"): Likewise.
20734 ("*tldo_ldub_sp64"): Likewise.
20735 ("*tldo_ldub1_sp64"): Likewise.
20736 ("*tldo_ldub2_sp64"): Likewise.
20737 ("*tldo_ldub3_sp64"): Likewise.
20738 ("*tldo_lduh_sp32"): Likewise.
20739 ("*tldo_lduh1_sp32"): Likewise.
20740 ("*tldo_lduh_sp64"): Likewise.
20741 ("*tldo_lduh1_sp64"): Likewise.
20742 ("*tldo_lduh2_sp64"): Likewise.
20743 ("*tldo_lduw_sp32"): Likewise.
20744 ("*tldo_lduw_sp64"): Likewise.
20745 ("*tldo_lduw1_sp64"): Likewise.
20746 ("*tldo_ldx_sp64"): Likewise.
20747 ("*mov<VM32:mode>_insn"): Likewise.
20748 ("*mov<VM64:mode>_insn_sp64"): Likewise.
20749 ("*mov<VM64:mode>_insn_sp32"): Likewise.
20751 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20753 * config/sparc/sparc.md ("type"): New insn type viscmp.
20754 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
20756 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
20757 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
20758 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
20759 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
20761 ("n7_vis_logical_11cycle"): Likewise.
20762 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
20763 * config/sparc/niagara2.md ("niag3_vis": Likewise.
20764 * config/sparc/niagara.md ("niag_vis"): Likewise.
20765 * config/sparc/ultra3.md ("us3_fga"): Likewise.
20766 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
20768 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
20770 * config/sparc/sparc.md: New instruction type `bmask'.
20771 (bmaskdi_vis): Use the `bmask' type.
20772 (bmasksi_vis): Likewise.
20773 * config/sparc/ultra3.md (us3_array): Likewise.
20774 * config/sparc/niagara7.md (n7_array): Likewise.
20775 * config/sparc/niagara4.md (n4_array): Likewise.
20776 * config/sparc/niagara2.md (niag2_vis): Likewise.
20777 (niag3_vis): Likewise.
20778 * config/sparc/niagara.md (niag_vis): Likewise.
20780 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20782 * ipa-comdats.c: Remove optimize check from gate.
20783 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
20784 for functions not optimized.
20785 (ipa_fn_summary_read): Skip optimize check.
20786 (ipa_fn_summary_write): Likewise.
20787 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
20789 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
20791 (can_inline_edge_p): Check flag_pcc_struct_return for match.
20792 (check_callers): Give up on caller which is not optimized.
20793 (inline_small_functions): Likewise.
20794 (ipa_inline): Do not give up when not optimizing.
20795 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
20796 away unoptimizes cdtors.
20797 (whole_program_function_and_variable_visibility): Do
20798 ipa_discover_readonly_nonaddressable_vars in LTO mode.
20799 * ipa.c (process_references): Do not check optimize.
20800 (symbol_table::remove_unreachable_nodes): Update optimize check.
20801 (set_writeonly_bit): Update optimize check.
20802 (pass_ipa_cdtor_merge::gate): Do not check optimize.
20803 (pass_ipa_single_use::gate): Remove.
20805 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20807 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
20808 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
20809 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
20810 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
20811 permute_load, permute_store, adjust_extract, adjust_splat,
20812 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
20813 replace_swap_with_copy, dump_swap_insn_table,
20814 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
20815 recombine_lvx_pattern, recombine_stvx_pattern,
20816 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
20817 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
20818 to file rs6000-p8swap.c.
20819 * config/rs6000/rs6000-p8swap.c: New file.
20820 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
20821 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
20822 and rs6000*-*-* targets.
20824 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20826 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
20828 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20830 * lto-wrapper.c (merge_and_complain): Do not merge
20831 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
20832 fsigned_zeros, ftrapping_math, fwrapv.
20833 (append_compiler_options): Do not track these options.
20834 (append_linker_options): Likewie
20836 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20838 * cgraphunit.c (cgraph_node::finalize_function): When
20839 !flag_toplevel_reorde set no_reorder flag.
20840 (varpool_node::finalize_decl): Likewise.
20841 (symbol_table::compile): Drop no toplevel reorder path.
20843 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20845 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
20846 edges; zero probability is not better than uninitialized.
20848 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20850 * asan.h (asan_sanitize_allocas_p): Declare.
20851 * asan.c (asan_sanitize_allocas_p): New function.
20852 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
20853 (handle_builtin_alloca): Likewise.
20854 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
20855 if !asan_sanitize_allocas_p.
20856 * params.def (asan-instrument-allocas): Add new option.
20857 * params.h (ASAN_PROTECT_ALLOCAS): Define.
20858 * opts.c (common_handle_option): Disable allocas sanitization for
20861 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
20863 * asan.c: Include gimple-fold.h.
20864 (get_last_alloca_addr): New function.
20865 (handle_builtin_stackrestore): Likewise.
20866 (handle_builtin_alloca): Likewise.
20867 (asan_emit_allocas_unpoison): Likewise.
20868 (get_mem_refs_of_builtin_call): Add new parameter, remove const
20869 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
20870 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
20871 (instrument_builtin_call): Pass gimple iterator to
20872 get_mem_refs_of_builtin_call.
20873 (last_alloca_addr): New global.
20874 * asan.h (asan_emit_allocas_unpoison): Declare.
20875 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
20876 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
20877 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
20878 if function calls alloca.
20879 * gimple-fold.c (replace_call_with_value): Remove static keyword.
20880 * gimple-fold.h (replace_call_with_value): Declare.
20881 * internal-fn.c: Include asan.h.
20882 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
20883 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
20885 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20887 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
20888 (C_SELFTEST_FLAGS): New.
20889 (CPP_SELFTEST_FLAGS): New.
20890 (SELFTEST_DEPS): New, from deps of s-selftest.
20891 (C_SELFTEST_DEPS): New, from deps of s-selftest.
20892 (CPP_SELFTEST_DEPS): New.
20893 (selftest): Add dependency on s-selftest-c++.
20894 (s-selftest): Rename to...
20895 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
20896 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
20897 than SELFTEST_FLAGS.
20898 (selftest-gdb): Rename to...
20899 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
20901 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
20902 (selftest-valgrind): Rename to...
20903 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
20905 (selftest-valgrind): Reintroduce as an alias for
20906 selftest-c-valgrind.
20907 (s-selftest-c++): New.
20908 (selftest-c++-gdb): New.
20909 (selftest-c++-valgrind): New.
20911 2017-07-06 Olivier Hainque <hainque@adacore.com>
20913 * gcc.c (process_command): When deciding if undefined variables
20914 should be ignored when processing specs, accept "gcc -v" as well.
20916 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
20918 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
20919 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
20921 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20923 * config/arm/arm-cpus.in (armv8-r): Add new entry.
20924 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
20925 * config/arm/arm-tables.opt: Regenerate.
20926 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
20928 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
20930 2017-07-06 Carl Love <cel@us.ibm.com>
20932 * ChangeLog: Clean up from mid air collision
20934 2017-07-06 Carl Love <cel@us.ibm.com>
20936 * config/rs6000/rs6000-c.c: Add support for built-in functions
20937 vector signed int vec_subc (vector signed int, vector signed int);
20938 vector signed __int128 vec_subc (vector signed __int128,
20939 vector signed __int128);
20940 vector unsigned __int128 vec_subc (vector unsigned __int128,
20941 vector unsigned __int128);
20942 vector signed int vec_sube (vector signed int, vector signed int,
20943 vector signed int);
20944 vector unsigned int vec_sube (vector unsigned int,
20945 vector unsigned int,
20946 vector unsigned int);
20947 vector signed __int128 vec_sube (vector signed __int128,
20948 vector signed __int128,
20949 vector signed__int128);
20950 vector unsigned __int128 vec_sube (vector unsigned __int128,
20951 vector unsigned __int128,
20952 vector unsigned __int128);
20953 vector signed int vec_subec (vector signed int, vector signed int,
20954 vector signed int);
20955 vector unsigned int vec_subec (vector unsigned int,
20956 vector unsigned int,
20957 vector unsigned int);
20958 vector signed __int128 vec_subec (vector signed __int128,
20959 vector signed __int128,
20960 vector signed__int128);
20961 vector unsigned __int128 vec_subec (vector unsigned __int128,
20962 vector unsigned __int128,
20963 vector unsigned __int128);
20964 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
20965 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
20966 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
20967 BU_ALTIVEC_OVERLOAD_X definitions.
20968 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
20969 * doc/extend.texi: Update the built-in documentation file for the new
20970 built-in functions.
20972 2017-07-06 David Malcolm <dmalcolm@redhat.com>
20975 * diagnostic-show-locus.c (layout::layout): Use start and finish
20976 spelling location for the start and finish of each range.
20977 * genmatch.c (linemap_client_expand_location_to_spelling_point):
20978 Add unused aspect param.
20979 * input.c (expand_location_1): Add "aspect" param, and use it
20980 to access the correct part of the location.
20981 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
20983 (expand_location_to_spelling_point): Likewise.
20984 (linemap_client_expand_location_to_spelling_point): Add "aspect"
20985 param, and pass it to expand_location_1.
20987 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
20989 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
20990 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
20991 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
20992 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
20993 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
20994 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
20995 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
20996 _mm_maskz_getmant_ss): New intrinsics.
20997 (__builtin_ia32_getexpss128_mask): Changed to ...
20998 __builtin_ia32_getexpss128_round ... this.
20999 (__builtin_ia32_getexpsd128_mask): Changed to ...
21000 __builtin_ia32_getexpsd128_round ... this.
21001 * config/i386/i386-builtin-types.def
21002 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
21003 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
21004 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
21005 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
21006 __builtin_ia32_getmantss_mask_round): New builtins.
21007 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
21008 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
21009 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
21010 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
21011 * config/i386/sse.md
21012 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
21013 avx512f_sgetexp<mode><mask_scalar_name>
21014 <round_saeonly_scalar_name> ... this.
21015 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
21016 %0, %1, %2<round_saeonly_op3>}): Changed to ...
21017 vgetexp<ssescalarmodesuffix>
21018 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21019 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
21020 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
21021 avx512f_vgetmant<mode><mask_scalar_name>
21022 <round_saeonly_scalar_name> ... this.
21023 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
21024 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
21025 vgetmant<ssescalarmodesuffix>
21026 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
21027 %0<mask_scalar_operand4>, %1, %2
21028 <round_saeonly_scalar_mask_op4>, %3} ... this.
21029 * config/i386/subst.md (mask_scalar_operand4,
21030 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
21031 round_saeonly_scalar_nimm_predicate): New subst attributes.
21033 2017-07-06 Julia Koval <julia.koval@intel.com>
21035 * config/i386/i386.c (ix86_erase_embedded_rounding):
21036 Remove code for old rounding pattern.
21038 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
21040 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
21042 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
21044 * doc/sourcebuild.texi (Test Directives, Variants of
21045 dg-require-support): Add documentation for dg-require-stack-check.
21047 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
21049 * config/i386/subst.md (mask_scalar, round_scalar,
21050 round_saeonly_scalar): New meta-templates.
21051 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
21052 round_scalar_mask_operand3, round_scalar_mask_op3,
21053 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
21054 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
21055 round_saeonly_scalar_constraint,
21056 round_saeonly_scalar_prefix): New subst attribute.
21057 * config/i386/sse.md
21058 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
21059 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
21060 <round_scalar_name> ... this.
21061 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
21062 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
21063 <round_scalar_name> ... this.
21064 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
21065 <sse>_vm<code><mode>3<mask_scalar_name>
21066 <round_saeonly_scalar_name> ... this.
21067 (v<plusminus_mnemonic><ssescalarmodesuffix>
21068 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21069 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21070 v<plusminus_mnemonic><ssescalarmodesuffix>
21071 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21072 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21073 (v<multdiv_mnemonic><ssescalarmodesuffix>
21074 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
21075 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
21076 v<multdiv_mnemonic><ssescalarmodesuffix>
21077 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21078 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
21079 (v<maxmin_float><ssescalarmodesuffix>
21080 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
21081 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
21082 v<maxmin_float><ssescalarmodesuffix>
21083 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
21084 %0<mask_scalar_operand3>, %1, %<iptr>2
21085 <round_saeonly_scalar_mask_op3>} ... this.
21087 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
21089 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
21090 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
21092 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
21093 Alan Hayward <alan.hayward@arm.com>
21094 David Sherwood <david.sherwood@arm.com>
21096 * combine.c (simplify_if_then_else): Remove "enum" before
21098 * compare-elim.c (can_eliminate_compare): Likewise.
21099 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
21101 (aarch64_lookup_simd_builtin_type): Likewise.
21102 (aarch64_simd_builtin_type): Likewise.
21103 (aarch64_init_simd_builtin_types): Likewise.
21104 (aarch64_simd_expand_args): Likewise.
21105 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
21107 (aarch64_reverse_mask): Likewise.
21108 (aarch64_simd_emit_reg_reg_move): Likewise.
21109 (aarch64_gen_adjusted_ldpstp): Likewise.
21110 (aarch64_ccmp_mode_to_code): Likewise.
21111 (aarch64_operands_ok_for_ldpstp): Likewise.
21112 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21113 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
21115 (aarch64_min_divisions_for_recip_mul): Likewise.
21116 (aarch64_reassociation_width): Likewise.
21117 (aarch64_get_condition_code_1): Likewise.
21118 (aarch64_simd_emit_reg_reg_move): Likewise.
21119 (aarch64_simd_attr_length_rglist): Likewise.
21120 (aarch64_reverse_mask): Likewise.
21121 (aarch64_operands_ok_for_ldpstp): Likewise.
21122 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
21123 (aarch64_gen_adjusted_ldpstp): Likewise.
21124 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
21126 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
21127 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
21128 (arm_lookup_simd_builtin_type): Likewise.
21129 (arm_simd_builtin_type): Likewise.
21130 (arm_init_simd_builtin_types): Likewise.
21131 (arm_expand_builtin_args): Likewise.
21132 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
21133 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
21134 (ft32_setup_incoming_varargs): Likewise.
21135 (ft32_function_arg): Likewise.
21136 (ft32_function_arg_advance): Likewise.
21137 (ft32_pass_by_reference): Likewise.
21138 (ft32_arg_partial_bytes): Likewise.
21139 (ft32_valid_pointer_mode): Likewise.
21140 (ft32_addr_space_pointer_mode): Likewise.
21141 (ft32_addr_space_legitimate_address_p): Likewise.
21142 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
21144 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
21145 (ix86_emit_outlined_ms2sysv_restore): Likewise.
21146 (iamcu_alignment): Likewise.
21147 (canonicalize_vector_int_perm): Likewise.
21148 (ix86_noce_conversion_profitable_p): Likewise.
21149 (ix86_mpx_bound_mode): Likewise.
21150 (ix86_operands_ok_for_move_multiple): Likewise.
21151 * config/microblaze/microblaze-protos.h
21152 (microblaze_expand_conditional_branch_reg): Likewise.
21153 * config/microblaze/microblaze.c
21154 (microblaze_expand_conditional_branch_reg): Likewise.
21155 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
21157 (rs6000_reassociation_width): Likewise.
21158 (rs6000_invalid_binary_op): Likewise.
21159 (fusion_p9_p): Likewise.
21160 (emit_fusion_p9_load): Likewise.
21161 (emit_fusion_p9_store): Likewise.
21162 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
21164 (riscv_hard_regno_mode_ok_p): Likewise.
21165 (riscv_address_insns): Likewise.
21166 (riscv_split_symbol): Likewise.
21167 (riscv_legitimize_move): Likewise.
21168 (riscv_function_value): Likewise.
21169 (riscv_hard_regno_nregs): Likewise.
21170 (riscv_expand_builtin): Likewise.
21171 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
21172 (riscv_build_integer): Likewise.
21173 (riscv_split_integer): Likewise.
21174 (riscv_legitimate_constant_p): Likewise.
21175 (riscv_cannot_force_const_mem): Likewise.
21176 (riscv_regno_mode_ok_for_base_p): Likewise.
21177 (riscv_valid_base_register_p): Likewise.
21178 (riscv_valid_offset_p): Likewise.
21179 (riscv_valid_lo_sum_p): Likewise.
21180 (riscv_classify_address): Likewise.
21181 (riscv_legitimate_address_p): Likewise.
21182 (riscv_address_insns): Likewise.
21183 (riscv_load_store_insns): Likewise.
21184 (riscv_force_binary): Likewise.
21185 (riscv_split_symbol): Likewise.
21186 (riscv_force_address): Likewise.
21187 (riscv_legitimize_address): Likewise.
21188 (riscv_move_integer): Likewise.
21189 (riscv_legitimize_const_move): Likewise.
21190 (riscv_legitimize_move): Likewise.
21191 (riscv_address_cost): Likewise.
21192 (riscv_subword): Likewise.
21193 (riscv_output_move): Likewise.
21194 (riscv_canonicalize_int_order_test): Likewise.
21195 (riscv_emit_int_order_test): Likewise.
21196 (riscv_function_arg_boundary): Likewise.
21197 (riscv_pass_mode_in_fpr_p): Likewise.
21198 (riscv_pass_fpr_single): Likewise.
21199 (riscv_pass_fpr_pair): Likewise.
21200 (riscv_get_arg_info): Likewise.
21201 (riscv_function_arg): Likewise.
21202 (riscv_function_arg_advance): Likewise.
21203 (riscv_arg_partial_bytes): Likewise.
21204 (riscv_function_value): Likewise.
21205 (riscv_pass_by_reference): Likewise.
21206 (riscv_setup_incoming_varargs): Likewise.
21207 (riscv_print_operand): Likewise.
21208 (riscv_elf_select_rtx_section): Likewise.
21209 (riscv_save_restore_reg): Likewise.
21210 (riscv_for_each_saved_reg): Likewise.
21211 (riscv_register_move_cost): Likewise.
21212 (riscv_hard_regno_mode_ok_p): Likewise.
21213 (riscv_hard_regno_nregs): Likewise.
21214 (riscv_class_max_nregs): Likewise.
21215 (riscv_memory_move_cost): Likewise.
21216 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
21217 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
21218 (rl78_addr_space_address_mode): Likewise.
21219 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21221 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
21222 (rs6000_reassociation_width): Likewise.
21223 (rs6000_invalid_binary_op): Likewise.
21224 (fusion_p9_p): Likewise.
21225 (emit_fusion_p9_load): Likewise.
21226 (emit_fusion_p9_store): Likewise.
21227 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
21228 (ok_for_simple_move_operands): Likewise.
21229 (ok_for_simple_move_strict_operands): Likewise.
21230 (ok_for_simple_arith_logic_operands): Likewise.
21231 (visium_legitimize_reload_address): Likewise.
21232 (visium_select_cc_mode): Likewise.
21233 (output_cbranch): Likewise.
21234 (visium_split_double_move): Likewise.
21235 (visium_expand_copysign): Likewise.
21236 (visium_expand_int_cstore): Likewise.
21237 (visium_expand_fp_cstore): Likewise.
21238 * config/visium/visium.c (visium_pass_by_reference): Likewise.
21239 (visium_function_arg): Likewise.
21240 (visium_function_arg_advance): Likewise.
21241 (visium_libcall_value): Likewise.
21242 (visium_setup_incoming_varargs): Likewise.
21243 (visium_legitimate_constant_p): Likewise.
21244 (visium_legitimate_address_p): Likewise.
21245 (visium_legitimize_address): Likewise.
21246 (visium_secondary_reload): Likewise.
21247 (visium_register_move_cost): Likewise.
21248 (visium_memory_move_cost): Likewise.
21249 (prepare_move_operands): Likewise.
21250 (ok_for_simple_move_operands): Likewise.
21251 (ok_for_simple_move_strict_operands): Likewise.
21252 (ok_for_simple_arith_logic_operands): Likewise.
21253 (visium_function_value_1): Likewise.
21254 (rtx_ok_for_offset_p): Likewise.
21255 (visium_legitimize_reload_address): Likewise.
21256 (visium_split_double_move): Likewise.
21257 (visium_expand_copysign): Likewise.
21258 (visium_expand_int_cstore): Likewise.
21259 (visium_expand_fp_cstore): Likewise.
21260 (visium_split_cstore): Likewise.
21261 (visium_select_cc_mode): Likewise.
21262 (visium_split_cbranch): Likewise.
21263 (output_cbranch): Likewise.
21264 (visium_print_operand_address): Likewise.
21265 * expmed.c (flip_storage_order): Likewise.
21266 * expmed.h (emit_cstore): Likewise.
21267 (flip_storage_order): Likewise.
21268 * genrecog.c (validate_pattern): Likewise.
21269 * hsa-gen.c (gen_hsa_addr): Likewise.
21270 * internal-fn.c (expand_arith_overflow): Likewise.
21271 * ira-color.c (allocno_copy_cost_saving): Likewise.
21272 * lra-assigns.c (find_hard_regno_for_1): Likewise.
21273 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
21274 (process_invariant_for_inheritance): Likewise.
21275 * lra-eliminations.c (move_plus_up): Likewise.
21276 * omp-low.c (lower_oacc_reductions): Likewise.
21277 * simplify-rtx.c (simplify_subreg): Likewise.
21278 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
21279 (TARGET_CHKP_BOUND_MODE): Likewise..
21280 * targhooks.c (default_chkp_bound_mode): Likewise.
21281 (default_setup_incoming_vararg_bounds): Likewise.
21282 * targhooks.h (default_chkp_bound_mode): Likewise.
21283 (default_setup_incoming_vararg_bounds): Likewise.
21284 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
21285 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
21286 (have_whole_vector_shift): Likewise.
21287 * tree-vect-stmts.c (vectorizable_load): Likewise.
21288 * doc/tm.texi: Regenerate.
21290 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21292 Graceful degrade if Binutils PR21472 is not available.
21295 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
21296 .rodata in flash test fails.
21297 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
21298 * confgure: Regenerate.
21299 * config.in: Regenerate.
21300 * config/avr/avr.c (avr_asm_named_section)
21301 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
21302 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
21303 (avr_asm_init_sections): Same.
21305 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21307 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
21308 (fma<VH:mode>4_intrinsic): Likewise.
21309 (*fmsub<VCVTF:mode>4): Likewise.
21310 (*fmsub<VH:mode>4_intrinsic): Likewise.
21312 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
21315 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
21316 Don't depend on "optimize > 0".
21317 (out_movhi_r_mr, out_movqi_mr_r): Same.
21318 (out_movhi_mr_r, out_movqi_r_mr): Same.
21319 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
21320 io_address_operand on "optimize > 0".
21322 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21324 * tree-loop-distribution.c: Add general explanantion on the pass.
21325 (generate_loops_for_partition): Mark distributed loop.
21326 (pg_add_dependence_edges): New parameter. Handle alias data
21327 dependence specially and record it in the parameter if asked.
21328 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
21329 (init_partition_graph_vertices, add_partition_graph_edge): New.
21330 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
21331 (free_partition_graph_vdata, build_partition_graph): New.
21332 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
21333 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
21334 (data_ref_segment_size, latch_dominated_by_data_ref): New.
21335 (compute_alias_check_pairs, version_loop_by_alias_check): New.
21336 (version_for_distribution_p, finalize_partitions): New.
21337 (distribute_loop): Handle alias data dependence specially. Factor
21338 out loop fusion code as functions and call these functions.
21340 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21342 * tree-loop-distribution.c (classify_partition): New parameter and
21343 better handle reduction statement.
21344 (rdg_build_partitions): Revise comment.
21345 (distribute_loop): Compute statements in all partitions and pass it
21346 to classify_partition.
21348 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21350 * tree-loop-distribution.c (enum partition_type): New.
21351 (struct partition): New field type.
21352 (partition_merge_into): Add parameter. Update partition type.
21353 (data_dep_in_cycle_p, update_type_for_merge): New functions.
21354 (build_rdg_partition_for_vertex): Compute partition type.
21355 (rdg_build_partitions): Dump partition type.
21356 (distribute_loop): Update calls to partition_merge_into.
21358 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21360 * tree-loop-distribution.c (struct ddr_hasher): New.
21361 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
21363 (classify_partition): Call get_data_dependence.
21364 (pg_add_dependence_edges): Ditto.
21365 (distribute_loop): Release data dependence hash table.
21367 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21369 * tree-loop-distribution.c (ref_base_address): Delete.
21370 (similar_memory_accesses): Rename ...
21371 (share_memory_accesses): ... to this. Check if partitions access
21372 the same memory reference.
21373 (distribute_loop): Call share_memory_accesses.
21375 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21377 * tree-loop-distribution.c (struct partition): New field recording
21378 its data reference.
21379 (partition_alloc, partition_free): Init and release data refs.
21380 (partition_merge_into): Merge data refs.
21381 (build_rdg_partition_for_vertex): Collect data refs for partition.
21382 (pg_add_dependence_edges): Change parameters from vector to bitmap.
21384 (distribute_loop): Remve data refs from vertice data of partition
21387 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21389 * tree-loop-distribution.c (params.h): Include header file.
21390 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
21391 (datarefs_vec): New global var.
21392 (create_rdg_vertices): Use datarefs_vec directly.
21393 (free_rdg): Don't free data references.
21394 (build_rdg): Update use. Don't free data references.
21395 (distribute_loop): Compute global variable for data references.
21396 Bail out if there are too many data references.
21398 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21400 * tree-loop-distribution.c (loop_nest): New global var.
21401 (build_rdg): Use loop directly, rather than loop nest.
21402 (pg_add_dependence_edges): Remove loop nest parameter. Use global
21404 (distribute_loop): Compute global variable loop nest. Update use.
21406 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21408 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
21409 (partition_merge_into): New parameter. Dump reason for fusion.
21410 (distribute_loop): Update use of partition_merge_into.
21412 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21414 * tree-loop-distribution.c (bb_top_order_index): New.
21415 (bb_top_order_index_size, bb_top_order_cmp): New.
21416 (stmts_from_loop): Use topological order.
21417 (pass_loop_distribution::execute): Compute and release topological
21418 order for basic blocks.
21420 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21422 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
21425 2017-07-05 Bin Cheng <bin.cheng@arm.com>
21427 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
21428 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
21429 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
21430 * internal-fn.def (LOOP_DIST_ALIAS): New.
21431 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
21432 (fold_loop_internal_call): ... this.
21433 (vect_loop_dist_alias_call): New function.
21434 (set_uid_loop_bbs): Call fold_loop_internal_call.
21435 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
21438 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21441 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
21442 Require dead FLAGS_REG at the beginning of a peephole.
21444 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
21447 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
21448 arguments in the call to __builtin_ia32_sbb_u32.
21449 (_subborrow_u64): Swap _X and _Y arguments in the call to
21450 __builtin_ia32_sbb_u64.
21452 2017-07-04 Jakub Jelinek <jakub@redhat.com>
21455 * tree-vrp.c (compare_assert_loc): Turn into a function template
21456 with stable template parameter. Only test if a->e is NULL,
21457 !a->e == !b->e has been verified already. Use e == NULL or
21458 e != NULL instead of e or ! e tests. If stable is true, don't use
21459 iterative_hash_expr, on the other side allow a or b or both NULL
21460 and sort the NULLs last.
21461 (process_assert_insertions): Sort using compare_assert_loc<false>
21462 instead of compare_assert_loc, later sort using
21463 compare_assert_loc<true> before calling process_assert_insertions_for
21464 in a loop. Use break instead of continue once seen NULL pointer.
21466 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21468 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21469 Cortex-R7 and Cortex-R8 processors.
21471 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21473 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
21474 uninitialized while src is not.
21476 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
21478 * common/config/arm/arm-common.c: Adjust include path for
21480 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
21481 (arm-cpu.h): Create in build directory. Adjust dependency rules.
21482 (arm-cpu-data.h): Likewise.
21483 (arm-cpu-cdata.h): Likewise.
21484 * config/arm/arm-cpu.h: Delete.
21485 * config/arm/arm-cpu-cdata.h: Delete.
21486 * config/arm/arm-cpu-data.h: Delete.
21488 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
21490 * config/arm/arm-cpus.in (cortex-a55): New.
21491 (cortex-a75): Likewise.
21492 (cortex-a75.cortex-a55): Likewise.
21493 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
21495 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
21496 * config/arm/arm-cpu-cdata.h: Regenerate.
21497 * config/arm/arm-cpu-data.h: Regenerate.
21498 * config/arm/arm-cpu.h: Regenerate.
21499 * config/arm/arm-tables.opt: Regenerate.
21500 * config/arm/arm-tune.md: Regenerate.
21502 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21504 * haifa-sched.c (sched_create_recovery_edges): Update profile.
21506 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
21508 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
21511 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
21513 PR tree-optimization/81292
21514 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
21515 full_string_p, also call adjust_related_strinfos if the adjustment
21516 is simple, otherwise invalidate related strinfos.
21518 2017-07-04 Martin Liska <mliska@suse.cz>
21521 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
21522 newly created variable as DECL_IGNORED_P.
21524 2017-07-04 Martin Liska <mliska@suse.cz>
21527 * ipa-inline.c (inline_small_functions):
21528 Use xstrdup_for_dump.
21530 2017-07-04 Tom de Vries <tom@codesourcery.com>
21532 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
21534 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21537 * config/darwin.c (darwin_function_switched_text_sections):
21540 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
21542 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
21544 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21546 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
21548 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21550 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
21551 min_profitable_iters, and th as inclusive lower bounds.
21552 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
21553 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
21554 for min_profitable_iters and min_profitable_estimate.
21555 (vect_transform_loop): Treat th as an inclusive lower bound.
21556 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
21558 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
21561 * config/darwin.c (darwin_function_switched_text_sections):
21562 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
21563 in two pieces, and suppress the use of buf.
21565 2017-07-03 Nathan Sidwell <nathan@acm.org>
21567 * hash-table.h (hash_table_mod1): Fix indentation.
21569 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21571 PR middle-end/81290
21572 * predict.c (force_edge_cold): Be more careful about propagation
21574 * profile-count.h (profile_probability::guessed,
21575 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
21577 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
21579 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
21581 * doc/invoke.texi (rcpc architecture extension): Document it.
21583 2017-07-03 Richard Biener <rguenther@suse.de>
21585 PR tree-optimization/60510
21586 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
21587 the scalar reduction PHI and use it.
21588 (vectorizable_reduction): Properly guard the single_defuse_cycle
21589 path for non-SLP reduction chains where we cannot use it.
21590 Rework reduc_def/index and vector type deduction. Rework
21591 vector operand gathering during reduction op code-gen.
21592 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
21593 chains dissolve the chain and leave it to non-SLP reduction
21596 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21598 * tree-data-ref.h (dr_alignment): Declare.
21599 * tree-data-ref.c (dr_alignment): New function.
21600 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
21601 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
21603 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
21605 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21607 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
21608 and base_misalignment fields.
21609 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
21610 * tree-data-ref.c: Include builtins.h.
21611 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
21612 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
21613 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
21614 * tree-vect-data-refs.c: Include tree-cfg.h.
21615 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
21616 fields instead of calculating an alignment here.
21617 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
21618 innermost_loop_behavior fields.
21620 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21622 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
21624 (DR_STEP_ALIGNMENT): New macro.
21625 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
21626 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
21627 (create_data_ref): Print it.
21628 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
21629 to tell whether the step preserves vector (mis)alignment.
21630 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
21631 Move the check for an integer step and generalise to all INTEGER_CST.
21632 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
21633 Print the outer step alignment.
21635 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21637 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
21638 with offset_alignment.
21639 (DR_ALIGNED_TO): Delete.
21640 (DR_OFFSET_ALIGNMENT): New macro.
21641 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
21642 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
21643 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
21644 (create_data_ref): Likewise.
21645 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
21646 (vect_analyze_data_refs): Likewise.
21647 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
21648 creating dummy innermost behavior.
21650 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21652 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
21653 with a "innermost_loop_behavior *" and refeence tree.
21654 * tree-data-ref.c (dr_analyze_innermost): Likewise.
21655 (create_data_ref): Update call accordingly.
21656 * tree-predcom.c (find_looparound_phi): Likewise.
21658 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21660 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
21661 fields with dr_wrt_vec_loop.
21662 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
21663 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
21664 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
21665 (vect_dr_behavior): New function.
21666 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21667 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
21668 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
21669 track whether the step preserves the misalignment.
21670 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
21671 Use vect_dr_behavior.
21672 (vect_setup_realignment): Update call accordingly.
21673 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
21674 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
21675 call to vect_create_addr_base_for_vector_ref.
21676 (vect_create_cond_for_align_checks): Likewise.
21677 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
21678 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
21679 (vect_recog_mask_conversion_pattern): Likewise.
21680 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
21681 (new_stmt_vec_info): Remove redundant zeroing.
21683 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
21685 * common/config/arm/arm-common.c (arm_be8_option): New function.
21686 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
21687 (ISA_ARMv6): Add isa_bit_be8.
21688 * config/arm/arm.h (arm_be8_option): Add prototype.
21689 (BE8_SPEC_FUNCTION): New define.
21690 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
21691 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
21692 (mlittle-endian): Similarly.
21693 (mbe8, mbe32): New options.
21694 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
21695 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
21697 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21699 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
21701 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21703 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
21704 (cleanup_tree_cfg_bb): Use it.
21705 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
21707 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
21709 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21712 * loop-doloop.c (add_test): Update profile.
21714 2017-07-03 Martin Liska <mliska@suse.cz>
21717 * sanopt.c (rewrite_usage_of_param): New function.
21718 (sanitize_rewrite_addressable_params): Likewise.
21719 (pass_sanopt::execute): Call rewrite_usage_of_param.
21721 2017-07-03 Richard Biener <rguenther@suse.de>
21723 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
21724 back to using VIEW_CONVERT_EXPR.
21726 2017-07-03 Martin Liska <mliska@suse.cz>
21729 * doc/extend.texi: Document when a resolver function is
21730 generated for target_clones.
21732 2017-07-03 Martin Liska <mliska@suse.cz>
21734 * asan.c (asan_emit_stack_protection): Unpoison just red zones
21735 and shadow memory of auto variables which are subject of
21736 use-after-scope sanitization.
21737 (asan_expand_mark_ifn): Add do set only when is_poison.
21739 2016-07-03 Richard Biener <rguenther@suse.de>
21741 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
21743 (vect_force_simple_reduction): Record reduction def -> phi mapping.
21744 (vectorizable_reduction): Perform reduction PHI creation when
21745 visiting a reduction PHI and adjust and simplify code generation
21746 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
21747 (vect_transform_loop): Visit reduction PHIs.
21748 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
21749 defs into the SLP tree.
21750 (vect_build_slp_tree): Reduction defs terminate the recursion.
21751 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
21753 (vect_get_vec_defs_for_stmt_copy): Export.
21754 (vect_get_vec_defs): Likewise.
21755 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
21757 (vect_get_vec_defs_for_stmt_copy): Declare.
21758 (vect_get_vec_defs): Likewise.
21760 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
21762 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
21763 parameter with a "loop" parameter and use it instead of the
21764 loop containing DR_STMT. Don't check simple_iv when doing
21765 BB analysis. Describe the two analysis modes in the comment.
21767 2017-07-03 Tom de Vries <tom@codesourcery.com>
21769 PR tree-optimization/69468
21770 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
21771 (find_same_succ_bb): Handle ignore_edge_flags.
21773 2017-07-03 Tom de Vries <tom@codesourcery.com>
21775 PR tree-optimization/81192
21776 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
21778 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
21780 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
21782 2017-07-03 Tom de Vries <tom@codesourcery.com>
21784 PR tree-optimization/81192
21785 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
21786 BB_SAME_SUCC (bb) == NULL.
21788 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21790 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
21793 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21795 * dumpfile.c: Include profile-count.h
21796 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
21798 (insert_cond_bb): Update profile.
21799 * tree-cfg.h (insert_cond_bb): Update prototype.
21800 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
21801 * tree-dump.c: Do not include tree-cfg.
21803 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21805 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
21807 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21809 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
21812 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
21814 * tree-complex.c (expand_complex_div_wide): update profile.
21816 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21817 Alan Hayward <alan.hayward@arm.com>
21818 David Sherwood <david.sherwood@arm.com>
21820 * Makefile.in (MACHMODE_H): Remove insn-modes.h
21821 (CORETYPES_H): New define.
21822 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
21823 (insn-modes-inline.h, s-modes-inline-h): New rules.
21824 (generated_files): Add insn-modes-inline.h.
21825 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
21826 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
21827 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
21828 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
21829 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
21830 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
21831 (build/gencodes.o, build/genconditions.o): Likewise.
21832 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
21833 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
21834 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
21835 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
21836 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
21837 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
21838 * coretypes.h: Include everything up to real.h for generators.
21839 Include insn-modes.h first. Include wide-int-print.h after
21840 wide-int.h. Include insn-modes-inline.h and then machmode.h.
21841 * machmode.h: Don't include insn-modes.h here.
21842 * function-tests.c: Remove includes of signop.h, machmode.h,
21843 double-int.h and wide-int.h.
21845 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
21847 * optc-save-gen.awk: Likewise.
21848 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
21849 * godump.c: Remove include of wide-int-print.h.
21850 * pretty-print.h: Likewise.
21851 * wide-int-print.cc: Likewise.
21852 * wide-int.cc: Likewise.
21853 * hash-map-tests.c: Remove include of signop.h.
21854 * hash-set-tests.c: Likewise.
21855 * rtl-tests.c: Likewise.
21856 * mkconfig.sh: Remove include of machmode.h.
21857 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
21859 (emit_insn_modes_inline_h): ...this new function. Emit the code
21860 into an insn-modes-inline.h header file, adding appropriate
21861 include guards and end comments.
21862 (emit_insn_modes_c_header): Remove include of machmode.h.
21863 (emit_min_insn_modes_c_header): Include coretypes.h rather than
21865 (main): Handle -i flag and call emit_insn_modes_inline_h when
21868 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21870 * tree-ssa-strlen.c (strinfo): Rename the length field to
21871 nonzero_chars. Add a full_string_p field.
21872 (compare_nonzero_chars, zero_length_string_p): New functions.
21873 (get_addr_stridx): Add an offset_out parameter.
21874 Use compare_nonzero_chars.
21875 (get_stridx): Update accordingly. Use compare_nonzero_chars.
21876 (new_strinfo): Update after above changes to strinfo.
21877 (set_endptr_and_length): Set full_string_p.
21878 (get_string_length): Update after above changes to strinfo.
21879 (unshare_strinfo): Update call to new_strinfo.
21880 (maybe_invalidate): Likewise.
21881 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
21882 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
21883 as a uhwi instead of an shwi. Update after above changes to
21884 strinfo and new_strinfo.
21885 (zero_length_string): Assert that chainsi contains full strings.
21886 Use zero_length_string_p. Update call to new_strinfo.
21887 (adjust_related_strinfos): Update after above changes to strinfo.
21888 Copy full_string_p from origsi.
21889 (adjust_last_stmt): Use zero_length_string_p.
21890 (handle_builtin_strlen): Update after above changes to strinfo and
21891 new_strinfo. Install the lhs as the string length if the previous
21892 entry didn't describe a full string.
21893 (handle_builtin_strchr): Update after above changes to strinfo
21895 (handle_builtin_strcpy): Likewise.
21896 (handle_builtin_strcat): Likewise.
21897 (handle_builtin_malloc): Likewise.
21898 (handle_pointer_plus): Likewise.
21899 (handle_builtin_memcpy): Likewise. Track nonzero characters
21900 that aren't necessarily followed by a nul terminator.
21901 (handle_char_store): Likewise.
21903 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21905 PR tree-optimization/80769
21906 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
21907 for malloc and calloc. Document the new invariant that all related
21908 strinfos have delayed lengths or none do.
21909 (verify_related_strinfos): Move earlier in file.
21910 (set_endptr_and_length): New function, split out from...
21911 (get_string_length): ...here. Also set the lengths of related
21913 (zero_length_string): Assert that chainsi has known (rather than
21915 (adjust_related_strinfos): Likewise.
21917 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
21919 PR tree-optimization/81136
21920 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
21921 assert that two references with the same misalignment have the same
21922 compile-time misalignment if those compile-time misalignments
21925 2017-07-01 Andi Kleen <ak@linux.intel.com>
21927 * print-tree.c (print_node): Print all attributes.
21929 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21931 * cfg.c (scale_bbs_frequencies): New function.
21932 * cfg.h (scale_bbs_frequencies): Declare it.
21933 * cfgloopanal.c (single_likely_exit): Cleanup.
21934 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
21936 (scale_loop_profile): Likewise.
21937 (loop_version): Likewise.
21938 (create_empty_loop_on_edge): Update.
21939 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
21940 scale_loop_frequencies, scale_loop_profile, loopify,
21941 loop_version): Update prototypes.
21942 * modulo-sched.c (sms_schedule): Update.
21943 * predict.c (unlikely_executed_edge_p): Also check probability.
21944 (probably_never_executed_edge_p): Fix typo.
21945 * tree-if-conv.c (version_loop_for_if_conversion): Update.
21946 * tree-parloops.c (gen_parallel_loop): Update.
21947 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
21948 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21949 * tree-ssa-loop-split.c (split_loop): Update.
21950 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
21951 * tree-vect-loop-manip.c (vect_do_peeling): Update.
21952 (vect_loop_versioning): Update.
21953 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21955 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21957 * trans-mem.c (split_bb_make_tm_edge): Update profile.
21959 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21961 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
21962 to keep profile consistent.
21964 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21966 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
21967 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
21968 * profile-count.h (max_safe_multiplier): Make unsigned.
21969 (profile_count::guessed_zero): New.
21971 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21973 * bb-reorder.c (fix_up_crossing_landing_pad,
21974 fix_crossing_conditional_branches): Use make_single_succ_edge
21975 to keep profile consistent.
21977 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
21979 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
21982 2017-07-01 Jakub Jelinek <jakub@redhat.com>
21985 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
21986 the right scopes, make sure cond_jump isn't preserved between multiple
21987 iterations. Search for fallthru edge whenever there are 3+ edges and
21988 use find_fallthru_edge for it.
21990 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21992 Patch by Alexander Monakov <amonakov@ispras.ru>
21993 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
21994 probabilities consistently.
21996 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
21998 * pa.c (pa_expand_compare_and_swap_loop): Update call of
21999 emit_cmp_and_jump_insns.
22001 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22004 * tree-inline.c (expand_call_inline): Combine profile statuses.
22006 2017-06-30 Andrew Pinski <apinski@cavium.com>
22008 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
22009 fold_stmt returned true.
22011 2017-06-30 Nathan Sidwell <nathan@acm.org>
22013 * ggc.h (empty_string): Delete.
22014 * cfgexpand.c (expand_asm_stmt): Use plain "".
22015 * optabs.c (expand_asm_memory_barrier): Likewise.
22016 * stringpool.c (empty_string): Delete.
22017 (digit_vector, digit_string): Delete.
22018 (ggc_alloc_string): Use plain "", don't optimize single digit
22019 strings. Use ggc_alloc_atomic.
22021 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
22023 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
22024 comparison set and one other set, use the cost of the non-comparison
22027 2017-06-30 Nathan Sidwell <nathan@acm.org>
22029 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
22032 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
22034 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
22035 loops. Remove now unneeded calls to gimple_switch_set_label() that
22036 just set removed labels to NULL_TREE.
22038 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
22040 * tree-ssanames.c (set_range_info_raw): Abstract from ...
22041 (set_range_info): ...here. Only call set_range_info_raw if domain
22043 (set_nonzero_bits): Call set_range_info_raw.
22044 * tree-ssanames.h (set_range_info_raw): New.
22046 2017-06-30 Jakub Jelinek <jakub@redhat.com>
22049 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
22050 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
22051 of nonimmediate_operand and <store_mask_constraint> instead of m
22052 for the input operand. For V8FI iterator, always split if input
22053 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
22054 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
22055 <store_mask_predicate> instead of register_operand and
22056 <store_mask_constraint> instead of v for the input operand. Make
22057 sure both operands aren't MEMs for if not <mask_applied>.
22059 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
22061 * lto-wrapper.c (copy_file) Close both file descriptors before
22064 2017-06-30 Martin Liska <mliska@suse.cz>
22067 * multiple_target.c (create_dispatcher_calls): Make ifunc
22068 also for function that don't have calls or are not referenced.
22070 2017-06-30 Richard Biener <rguenther@suse.de>
22072 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
22073 analyze the first scalar stmt. Move vector type computation
22074 for the BB case here from ...
22075 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
22076 live operation processing in the SLP case properly.
22078 2017-06-30 Richard Biener <rguenther@suse.de>
22080 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
22082 2017-06-30 Martin Liska <mliska@suse.cz>
22085 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
22086 before BUILT_IN_UNWIND_RESUME when ASAN is used.
22088 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
22090 * doc/invoke.texi (AArch64): Add missing options and remove redundant
22093 2017-06-30 Richard Biener <rguenther@suse.de>
22095 PR tree-optimization/81249
22096 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
22097 condition reduction result to original scalar type.
22099 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22101 * profile-count.h (enum profile_quality): Fix typos and whitespace
22104 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22106 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
22107 type for branch probabilities.
22109 2017-06-29 Julian Brown <julian@codesourcery.com>
22110 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22112 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
22113 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
22114 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
22115 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
22117 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22119 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
22120 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
22121 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
22122 CC usage from generic code to here.
22123 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
22124 CC usage into the target macros.
22126 2017-06-29 Maya Rashish <coypu@sdf.org>
22128 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
22131 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22133 * arm/arm-builtins.c: Include profile-count.h
22134 * except.c (sjlj_emit_function_enter): Use
22135 profile_probability::unlikely.
22137 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22139 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
22140 and tocrel_offset be pointer args rather than implicitly using
22142 (legitimate_constant_pool_address_p, rs6000_emit_move,
22143 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
22144 tocrel_offset and use in toc_relative_expr_p call.
22145 (print_operand, print_operand_address): Use static tocrel_base_oac
22146 and tocrel_offset_oac.
22147 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
22150 2017-06-29 Maya Rashish <coypu@sdf.org>
22152 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
22154 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
22156 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
22157 objects, take into account only the alignment of 'op0' and 'mode1' if
22160 2017-06-29 Steve Ellcey <sellcey@cavium.com>
22162 * ccmp.c (ccmp_tree_comparison_p): New function.
22163 (ccmp_candidate_p): Update to use above function.
22164 (get_compare_parts): New function.
22165 (expand_ccmp_next): Update to use new functions.
22166 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
22168 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
22169 take mode as argument.
22170 * ccmp.h (expand_ccmp_expr): Add mode as argument.
22171 * expr.c (expand_expr_real_1): Pass mode as argument.
22173 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
22175 * combine.c (combine_instructions): Print insns to dump_file, together
22178 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
22180 * asan.c (asan_emit_stack_protection): Update.
22181 (create_cond_insert_point): Update.
22182 * auto-profile.c (afdo_propagate_circuit): Update.
22183 * basic-block.h (struct edge_def): Turn probability to
22184 profile_probability.
22185 (EDGE_FREQUENCY): Update.
22186 * bb-reorder.c (find_traces_1_round): Update.
22187 (better_edge_p): Update.
22188 (sanitize_hot_paths): Update.
22189 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
22190 (make_single_succ_edge): Update.
22191 (check_bb_profile): Update.
22192 (dump_edge_info): Update.
22193 (update_bb_profile_for_threading): Update.
22194 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
22195 probabilitycount to 0.
22196 * cfgbuild.c (compute_outgoing_frequencies): Update.
22197 * cfgcleanup.c (try_forward_edges): Update.
22198 (outgoing_edges_match): Update.
22199 (try_crossjump_to_edge): Update.
22200 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
22201 (expand_gimple_tailcall): Update.
22202 (construct_init_block): Use make_single_succ_edge.
22203 (construct_exit_block): Use make_single_succ_edge.
22204 * cfghooks.c (verify_flow_info): Update.
22205 (redirect_edge_succ_nodup): Update.
22206 (split_edge): Update.
22207 (account_profile_record): Update.
22208 * cfgloopanal.c (single_likely_exit): Update.
22209 * cfgloopmanip.c (scale_loop_profile): Update.
22210 (set_zero_probability): Remove.
22211 (duplicate_loop_to_header_edge): Update.
22212 * cfgloopmanip.h (loop_version): Update prototype.
22213 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
22214 (force_nonfallthru_and_redirect): Update.
22215 (update_br_prob_note): Update.
22216 (rtl_verify_edges): Update.
22217 (purge_dead_edges): Update.
22218 (rtl_lv_add_condition_to_bb): Update.
22219 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
22220 * cgraphunit.c (init_lowered_empty_function): Update.
22221 (cgraph_node::expand_thunk): Update.
22222 * cilk-common.c: Include profile-count.h
22223 * dojump.c (inv): Remove.
22224 (jumpifnot): Update.
22225 (jumpifnot_1): Update.
22226 (do_jump_1): Update.
22228 (do_jump_by_parts_greater_rtx): Update.
22229 (do_compare_rtx_and_jump): Update.
22230 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
22231 do_jump_1. do_compare_rtx_and_jump): Update prototype.
22232 * dwarf2cfi.c: Include profile-count.h
22233 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
22234 (sjlj_emit_dispatch_table): Likewise.
22235 * explow.c: Include profile-count.h
22236 * expmed.c (emit_store_flag_force): Update.
22237 (do_cmp_and_jump): Update.
22238 * expr.c (compare_by_pieces_d::generate): Update.
22239 (compare_by_pieces_d::finish_mode): Update.
22240 (emit_block_move_via_loop): Update.
22241 (store_expr_with_bounds): Update.
22242 (store_constructor): Update.
22243 (expand_expr_real_2): Update.
22244 (expand_expr_real_1): Update.
22245 * expr.h (try_casesi, try_tablejump): Update prototypes.
22246 * gimple-pretty-print.c (dump_probability): Update.
22247 (dump_profile): New.
22248 (dump_gimple_label): Update.
22249 (dump_gimple_bb_header): Update.
22250 * graph.c (draw_cfg_node_succ_edges): Update.
22251 * hsa-gen.c (convert_switch_statements): Update.
22252 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
22253 (find_if_case_1): Update.
22254 (find_if_case_2): Update.
22255 * internal-fn.c (expand_arith_overflow_result_store): Update.
22256 (expand_addsub_overflow): Update.
22257 (expand_neg_overflow): Update.
22258 (expand_mul_overflow): Update.
22259 (expand_vector_ubsan_overflow): Update.
22260 * ipa-cp.c (good_cloning_opportunity_p): Update.
22261 * ipa-split.c (split_function): Use make_single_succ_edge.
22262 * ipa-utils.c (ipa_merge_profiles): Update.
22263 * loop-doloop.c (add_test): Update.
22264 (doloop_modify): Update.
22265 * loop-unroll.c (compare_and_jump_seq): Update.
22266 (unroll_loop_runtime_iterations): Update.
22267 * lra-constraints.c (lra_inheritance): Update.
22268 * lto-streamer-in.c (input_cfg): Update.
22269 * lto-streamer-out.c (output_cfg): Update.
22270 * mcf.c (adjust_cfg_counts): Update.
22271 * modulo-sched.c (sms_schedule): Update.
22272 * omp-expand.c (expand_omp_for_init_counts): Update.
22273 (extract_omp_for_update_vars): Update.
22274 (expand_omp_ordered_sink): Update.
22275 (expand_omp_for_ordered_loops): Update.
22276 (expand_omp_for_generic): Update.
22277 (expand_omp_for_static_nochunk): Update.
22278 (expand_omp_for_static_chunk): Update.
22279 (expand_cilk_for): Update.
22280 (expand_omp_simd): Update.
22281 (expand_omp_taskloop_for_outer): Update.
22282 (expand_omp_taskloop_for_inner): Update.
22283 * omp-simd-clone.c (simd_clone_adjust): Update.
22284 * optabs.c (expand_doubleword_shift): Update.
22285 (expand_abs): Update.
22286 (emit_cmp_and_jump_insn_1): Update.
22287 (expand_compare_and_swap_loop): Update.
22288 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
22289 * predict.c (predictable_edge_p): Update.
22290 (edge_probability_reliable_p): Update.
22291 (set_even_probabilities): Update.
22292 (combine_predictions_for_insn): Update.
22293 (combine_predictions_for_bb): Update.
22294 (propagate_freq): Update.
22295 (estimate_bb_frequencies): Update.
22296 (force_edge_cold): Update.
22297 * profile-count.c (profile_count::dump): Add missing space into dump.
22298 (profile_count::debug): Add newline.
22299 (profile_count::differs_from_p): Explicitly convert to unsigned.
22300 (profile_count::stream_in): Update.
22301 (profile_probability::dump): New member function.
22302 (profile_probability::debug): New member function.
22303 (profile_probability::differs_from_p): New member function.
22304 (profile_probability::differs_lot_from_p): New member function.
22305 (profile_probability::stream_in): New member function.
22306 (profile_probability::stream_out): New member function.
22307 * profile-count.h (profile_count_quality): Rename to ...
22308 (profile_quality): ... this one.
22309 (profile_probability): New.
22310 (profile_count): Update.
22311 * profile.c (compute_branch_probabilities): Update.
22312 * recog.c (peep2_attempt): Update.
22313 * sched-ebb.c (schedule_ebbs): Update.
22314 * sched-rgn.c (find_single_block_region): Update.
22315 (compute_dom_prob_ps): Update.
22316 (schedule_region): Update.
22317 * sel-sched-ir.c (compute_succs_info): Update.
22318 * stmt.c (struct case_node): Update.
22319 (do_jump_if_equal): Update.
22320 (get_outgoing_edge_probs): Update.
22321 (conditional_probability): Update.
22322 (emit_case_dispatch_table): Update.
22323 (expand_case): Update.
22324 (expand_sjlj_dispatch_table): Update.
22325 (emit_case_nodes): Update.
22326 * targhooks.c: Update.
22327 * tracer.c (better_p): Update.
22328 (find_best_successor): Update.
22329 * trans-mem.c (expand_transaction): Update.
22330 * tree-call-cdce.c: Update.
22331 * tree-cfg.c (gimple_split_edge): Upate.
22332 (move_sese_region_to_fn): Upate.
22333 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
22334 * tree-eh.c (lower_resx): Upate.
22335 (cleanup_empty_eh_move_lp): Upate.
22336 * tree-if-conv.c (version_loop_for_if_conversion): Update.
22337 * tree-inline.c (copy_edges_for_bb): Update.
22338 (copy_cfg_body): Update.
22339 * tree-parloops.c (gen_parallel_loop): Update.
22340 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
22341 (gimple_gen_time_profiler): Update.
22342 * tree-ssa-dce.c (remove_dead_stmt): Update.
22343 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22344 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
22345 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
22346 (unloop_loops): Update.
22347 (try_peel_loop): Update.
22348 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22349 * tree-ssa-loop-split.c (connect_loops): Update.
22350 (split_loop): Update.
22351 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
22352 (hoist_guard): Update.
22353 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
22354 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
22355 (value_replacement): Update.
22356 * tree-ssa-reassoc.c (branch_fixup): Update.
22357 * tree-ssa-tail-merge.c (replace_block_by): Update.
22358 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
22359 (create_edge_and_update_destination_phis): Update.
22360 (compute_path_counts): Update.
22361 (recompute_probabilities): Update.
22362 (update_joiner_offpath_counts): Update.
22363 (freqs_to_counts_path): Update.
22364 (duplicate_thread_path): Update.
22365 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
22366 (struct switch_conv_info): Update.
22367 (gen_inbound_check): Update.
22368 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22369 (vect_do_peeling): Update.
22370 (vect_loop_versioning): Update.
22371 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22372 (optimize_mask_stores): Update.
22373 * ubsan.c (ubsan_expand_null_ifn): Update.
22374 * value-prof.c (gimple_divmod_fixed_value): Update.
22375 (gimple_divmod_fixed_value_transform): Update.
22376 (gimple_mod_pow2): Update.
22377 (gimple_mod_pow2_value_transform): Update.
22378 (gimple_mod_subtract): Update.
22379 (gimple_mod_subtract_transform): Update.
22380 (gimple_ic): Update.
22381 (gimple_stringop_fixed_value): Update.
22382 (gimple_stringops_transform): Update.
22383 * value-prof.h: Update.
22385 2017-06-29 Carl Love <cel@us.ibm.com>
22387 * config/rs6000/rs6000-c.c: Add support for built-in functions
22388 vector signed int vec_signed (vector float);
22389 vector signed long long vec_signed (vector double);
22390 vector signed int vec_signed2 (vector double, vector double);
22391 vector signed int vec_signede (vector double);
22392 vector signed int vec_signedo (vector double);
22393 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
22394 instruction generator.
22395 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22396 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
22397 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
22399 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
22400 vunsignede_v2df): Add define_expands.
22401 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
22402 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
22403 VEC_UNSIGNEDO): Add definitions.
22404 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
22405 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
22406 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
22407 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
22408 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
22409 * config/rs6000/altivec.h (vec_signed, vec_signed2,
22410 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
22411 vec_unsignede, vec_unsignedo): Add builtin defines.
22412 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
22414 * doc/extend.texi: Update the built-in documentation file for the
22415 new built-in functions.
22417 2017-06-29 Richard Biener <rguenther@suse.de>
22419 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
22420 reduction chains to LOOP_VINFO_REDUCTIONS.
22421 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
22422 SLP reductions after processing reduction chains.
22424 2017-06-29 Nathan Sidwell <nathan@acm.org>
22426 * builtins.c (fold_builtin_FUNCTION): Use
22427 lang_hooks.decl_printable_name.
22429 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
22431 PR middle-end/81194
22432 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
22433 with only one label.
22434 * stmt.c (expand_case): Assert NCASES is greater than one.
22436 2017-06-29 Richard Biener <rguenther@suse.de>
22438 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
22440 (group_case_labels): Likewise.
22441 (find_taken_edge): Push sanity checking on val to workers...
22442 (find_taken_edge_cond_expr): ... here
22443 (find_taken_edge_switch_expr): ... and here, handle cases
22444 with just a default label.
22445 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
22446 (group_case_labels): Likewise.
22447 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
22448 group_case_labels does anything cleanup the CFG again.
22450 2017-06-29 Bin Cheng <bin.cheng@arm.com>
22452 PR tree-optimization/81196
22453 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
22454 exit condition comparing two IVs.
22456 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
22458 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
22459 profile to the dummy entry at the end of the list of architectures.
22460 * config/arm/arm-cpu-cdata.h: Regenerated.
22462 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22463 Michael Collison <michael.collison@arm.com>
22466 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
22468 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
22469 (*aarch64_reg_<mode>3_minus_mask): New pattern.
22470 (*aarch64_<optab>_reg_di3_mask2): New pattern.
22471 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
22472 of shift when the shift amount is masked with constant equal to
22473 the size of the mode.
22474 * config/aarch64/predicates.md (subreg_lowpart_operator): New
22477 2017-06-29 Martin Liska <mliska@suse.cz>
22479 * config/i386/i386.opt: Change range from [1,5] to [0,5].
22481 2017-06-29 Yury Gribov <tetra2005@gmail.com>
22484 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
22487 (edge_growth_cache_entry::edge_growth_cache_entry): New
22489 (reset_edge_growth_cache): Update to use constructor.
22491 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22493 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
22494 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22495 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
22497 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
22499 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
22500 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
22502 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
22504 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
22505 (*-linux-uclibc*): Add t-uclibc tmake_file.
22506 * config/t-musl: New.
22507 * config/t-uclibc: New.
22509 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
22511 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
22513 (gen_comm_data): Emit architectural setting of arch_prof.
22514 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
22516 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
22517 (armv8-m.base, armv8-m.main): Likewise.
22518 * arm-protos.h (arm_build_target): Add profile field.
22519 (arch_option): Likewise.
22520 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
22522 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
22523 arm_active_target.profile.
22525 2017-06-28 Richard Biener <rguenther@suse.de>
22527 PR middle-end/81227
22528 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
22529 TYPE_OVERFLOW_WRAPS.
22530 * match.pd (negate_expr_p): Likewise.
22531 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
22532 fold_build2, not fold_binary.
22534 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22536 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
22537 Convert memory address to Pmode.
22538 (aarch64_print_operand): Assert MEM operands are always Pmode.
22540 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
22543 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
22544 Remove redundant if.
22545 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
22546 * config/arm/aarch-common-protos.h
22547 (aarch_forward_to_shift_is_not_shifted_re): Remove.
22548 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
22550 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
22553 * multiple_target.c (create_dispatcher_calls): Set the default
22554 clone to be static, not public.
22556 2017-06-28 Richard Biener <rguenther@suse.de>
22558 * tree-vect-loop.c (vectorizable_reduction): Move special
22559 cond reduction IV var creation ...
22560 (vect_create_epilog_for_reduction): ... here. Remove induction_index
22561 parameter. Use STMT_VINFO_VECTYPE.
22562 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
22565 2017-06-28 Martin Liska <mliska@suse.cz>
22568 * ipa-visibility.c (non_local_p): Handle visibility.
22570 2017-06-28 Martin Liska <mliska@suse.cz>
22573 * common.opt: Add IntegerRange to various options.
22574 * opt-functions.awk (integer_range_info): New function.
22575 * optc-gen.awk: Add integer_range_info to cl_options struct.
22576 * opts-common.c (decode_cmdline_option): Handle
22577 CL_ERR_INT_RANGE_ARG.
22578 (cmdline_handle_error): Likewise.
22579 * opts.c (print_filtered_help): Show valid interval in
22580 when --help is provided.
22581 * opts.h (struct cl_option): Add range_min and range_max fields.
22582 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
22584 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
22586 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
22587 (x * C EQ/NE y * C): New transformation.
22589 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
22591 * genmultilib (combination_space): Accept '+' in option names.
22593 2017-06-28 Martin Liska <mliska@suse.cz>
22596 * asan.c (instrument_derefs): Bail out inner references
22597 that are hard register variables.
22599 2017-06-28 Jakub Jelinek <jakub@redhat.com>
22602 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
22603 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
22605 2017-06-28 Richard Biener <rguenther@suse.de>
22607 * tree-vectorizer.h (vect_get_vec_defs): Remove.
22608 (vect_get_slp_defs): Adjust.
22609 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
22611 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
22613 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
22614 get_initial_defs_for_reduction instead of vect_get_vec_defs.
22615 (vectorizable_reduction): Adjust.
22616 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
22618 (vect_get_slp_defs): Likewise.
22619 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
22620 (vectorizable_bswap): Adjust.
22621 (vectorizable_call): Likewise.
22622 (vectorizable_conversion): Likewise.
22623 (vectorizable_assignment): Likewise.
22624 (vectorizable_shift): Likewise.
22625 (vectorizable_operation): Likewise.
22626 (vectorizable_store): Likewise.
22627 (vectorizable_condition): Likewise.
22628 (vectorizable_comparison): Likewise.
22630 2017-06-28 Michael Collison <michael.collison@arm.com>
22633 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
22635 (arm_gen_movmemqi): Removed unused variable 'i'.
22636 Convert 'for' loop into 'while' loop.
22637 (arm_expand_prologue): Remove last unnecessary set of insn.
22638 (thumb_pop): Remove unused variable 'pushed_words'.
22639 (thumb_exit): Remove last unnecessary set of regs_to_pop.
22641 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22643 * config/s390/predicates.md: Use s390_rel_address_ok_p.
22644 * config/s390/s390-protos.h: Add prototype of
22645 s390_rel_address_ok_p.
22646 * config/s390/s390.c (s390_got_symbol): New function.
22647 (s390_rel_address_ok_p): New function.
22648 (legitimize_pic_address): Use s390_rel_address_ok_p.
22649 (s390_load_got): Use s390_got_symbol.
22650 (s390_option_override): Issue error if
22651 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
22652 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
22654 * config/s390/s390.opt: New option mpic-data-is-text-relative.
22656 2017-06-27 Andrew Pinski <apinski@cavium.com>
22658 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
22659 (X * copysign (1.0, X)): New pattern.
22660 (X * copysign (1.0, -X)): New pattern.
22661 (copysign (-1.0, CST)): New pattern.
22663 2017-06-27 Joseph Myers <joseph@codesourcery.com>
22665 * genmultilib (combination_space): Remove variable.
22666 Validate reuse rules against regular expression for any sequence
22667 of multilib options in any order.
22669 2017-06-27 Michael Collison <michael.collison@arm.com>
22671 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
22672 call aarch64_split_simd_combine.
22673 * (aarch64_combine_internal<mode>): Delete pattern.
22674 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
22675 Allow register and subreg operands.
22677 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22679 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
22680 specific need, just fallback on defaults.
22681 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
22683 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22684 Olivier Hainque <hainque@adacore.com>
22686 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
22688 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
22689 targets. Pick a default if no particular attempt applied.
22690 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
22693 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22695 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
22696 (x86_64-wrs-vxworks7): Likewise.
22698 2017-06-27 Marek Polacek <polacek@redhat.com>
22701 * ubsan.c (instrument_null): Check get_base_address's result for null.
22703 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22705 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
22707 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
22709 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
22710 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
22711 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
22712 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
22713 New function types.
22714 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
22715 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
22716 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
22717 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
22718 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
22719 BUILT_IN_FEUPDATEENV): New builtins.
22720 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
22721 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
22722 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
22723 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
22725 (builtin_structptr_types): Adjust size.
22726 * tree.c (builtin_structptr_types): Add four entries.
22728 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22729 Olivier Hainque <hainque@adacore.com>
22731 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
22732 (TLS_SYM): New local macro, forcing reference to __tls__ on
22733 link command lines for VxWorks 7 RTPs, triggering initialization
22735 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
22736 OS features TLS support, true for RTPs on VxWorks 7.
22737 * config/vxworks.c (vxworks_override_options): Setup emutls
22740 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22742 * predict.c (test_prediction_value_range): Use -1U instead of -1
22743 to avoid narrowing conversion warning.
22744 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
22745 to avoid narrowing conversion warning.
22746 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
22748 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
22750 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
22752 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
22753 64bit configurations.
22754 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
22755 (SIZE_TYPE): Likewise.
22756 * config/vxworks.c (vxworks_emutls_var_fields): Use
22757 long_unsigned_type_node instead of unsigned_type_node as the offset
22758 field type, which is "pointer" mode in emutls.c.
22760 2017-06-27 Jakub Jelinek <jakub@redhat.com>
22763 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
22765 PR middle-end/81207
22766 * gimple-fold.c (replace_call_with_call_and_fold): Handle
22767 gimple_vuse copying separately from gimple_vdef copying.
22769 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22771 * value-prof.c (free_hist): Remove call to memset and the enclosing if
22774 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22775 Olivier Hainque <hainque@adacore.com>
22777 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
22778 for all vxworks7 targets.
22779 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
22780 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
22781 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
22782 variations for VX6/VX7 and 32/64bits later on in ...
22783 (VXWORKS_LIB_SPEC): Leverage new macros.
22784 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
22785 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
22787 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
22789 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
22790 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
22792 2017-06-26 Carl Love <cel@us.ibm.com>
22794 * config/rs6000/rs6000-c.c: Add support for built-in functions
22795 vector bool char vec_reve (vector bool char);
22796 vector signed char vec_reve (vector signed char);
22797 vector unsigned char vec_reve (vector unsigned char);
22798 vector bool int vec_reve (vector bool int);
22799 vector signed int vec_reve (vector signed int);
22800 vector unsigned int vec_reve (vector unsigned int);
22801 vector bool long long vec_reve (vector bool long long);
22802 vector signed long long vec_reve (vector signed long long);
22803 vector unsigned long long vec_reve (vector unsigned long long);
22804 vector bool short vec_reve (vector bool short);
22805 vector signed short vec_reve (vector signed short);
22806 vector double vec_reve (vector double);
22807 vector float vec_reve (vector float);
22808 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
22809 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
22810 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
22811 (altivec_vreve): New pattern.
22812 * config/rs6000/altivec.h (vec_reve): New define.
22813 * doc/extend.texi (vec_rev): Update the built-in documentation file
22814 for the new built-in functions.
22816 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22818 PR tree-optimization/71815
22819 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
22821 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
22823 (slsr_process_phi): Likewise.
22824 (replace_uncond_cands_and_profitable_phis): Don't replace a
22825 multiply candidate with a stride of 1 (copy or cast).
22826 (phi_incr_cost): Call uses_consumed_by_stmt rather than
22828 (lowest_cost_path): Likewise.
22829 (total_savings): Likewise.
22831 2017-06-26 Richard Biener <rguenther@suse.de>
22834 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
22835 Use def_builtin_pure for all gather builtins.
22837 2017-06-26 Richard Biener <rguenther@suse.de>
22839 PR tree-optimization/81203
22840 * tree-tailcall.c (find_tail_calls): Do not move stmts into
22841 non-dominating BBs.
22843 2017-06-26 Marek Polacek <polacek@redhat.com>
22846 * doc/invoke.texi: Document -Wmultistatement-macros.
22848 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
22850 * doc/sourcebuild.texi (ARM-specific attributes): Document new
22851 arm_neon_ok_no_float_abi effective target.
22853 2017-06-26 Richard Biener <rguenther@suse.de>
22855 PR tree-optimization/80928
22856 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
22857 (copy_bbs): Set BB_DUPLICATED flag early.
22858 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
22860 (execute_on_shrinking_pred): Likewise.
22861 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
22862 BB_DUPLICATED blocks.
22863 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
22864 iterate over all PHIs considering removal of *gsi.
22866 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
22868 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
22871 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22873 * config/rs6000/rs6000-string.c: (expand_block_clear,
22874 do_load_for_compare, select_block_compare_mode,
22875 compute_current_alignment, expand_block_compare,
22876 expand_strncmp_align_check, expand_strn_compare,
22877 expand_block_move, rs6000_output_load_multiple)
22878 Move functions related to string/block move/compare
22879 to a separate file.
22880 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
22881 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
22882 for this function which is now used in two files.
22883 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
22884 * config.gcc: Add rs6000-string.o to extra_objs for
22885 targets powerpc*-*-* and rs6000*-*-*.
22887 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
22890 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
22891 32-bit, since indexed is not valid for DImode.
22892 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
22893 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
22894 (define_peephole2 for Altivec d-form load): Add 32-bit support.
22895 (define_peephole2 for Altivec d-form store): Likewise.
22898 * multiple_target.c (create_dispatcher_calls): Only create the
22899 dispatcher call if the function is the default clone of a
22900 versioned function.
22902 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
22904 PR middle-end/80902
22905 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
22906 a call, force the call to not be a tail call.
22908 2017-06-23 Jeff Law <law@redhat.com>
22910 * doc/contrib.texi: Add entry for Steven Pemberton's work on
22913 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
22915 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
22916 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
22917 handling for early expansion of vector shifts (sl,sr,sra,rl).
22918 (builtin_function_type): Add vector shift right instructions
22919 to the unsigned argument list.
22921 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
22923 rtl-optimizatoin/79286
22924 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
22925 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
22926 trap. PIC register plus a const unspec without offset can never trap.
22928 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
22930 * tree.h (builtin_structptr_type): New type.
22931 (builtin_structptr_types): Declare new array.
22932 * tree.c (builtin_structptr_types): New array.
22933 (free_lang_data, build_common_tree_nodes): Use it.
22935 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
22938 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
22941 2017-06-22 Matt Turner <mattst88@gmail.com>
22943 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
22944 Lake models to skylake case. Assume skylake for unknown
22945 models with clflushopt.
22947 2017-06-22 Jeff Law <law@redhat.com>
22949 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
22950 frame sizes that do not satisfy aarch64_uimm12_shift.
22952 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
22954 * profile-count.h (apply_probability,
22955 apply_scale, probability_in): Fix checks for zero.
22957 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22959 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
22960 * doc/cppdiropts.texi (-I @var{dir}): Document it.
22962 2016-06-22 Richard Biener <rguenther@suse.de>
22964 * tree-vect-loop.c (vect_model_reduction_cost): Handle
22965 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
22966 REDUC_MAX_EXPR support.
22967 (vectorizable_reduction): Likewise.
22968 (vect_create_epilog_for_reduction): Likewise.
22970 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
22972 * match.pd (A / (1 << B) -> A >> B): New.
22973 * generic-match-head.c: Include optabs-tree.h.
22974 * gimple-match-head.c: Likewise.
22975 * optabs-tree.h (target_supports_op_p): New.
22976 * optabs-tree.c (target_supports_op_p): New.
22978 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22980 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
22981 $gcc_cv_ld --help output.
22982 (gcc_cv_ld_demangle): Likewise.
22983 (gcc_cv_ld_eh_frame_hdr): Likewise.
22984 (gcc_cv_ld_pie): Likewise.
22985 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
22986 (gcc_cv_ld_buildid): Likewise.
22987 (gcc_cv_ld_sysroot): Likewise.
22988 (ld_bndplt_support): Likewise.
22989 (ld_pushpopstate_support): Likewise.
22990 * configure: Regenerate.
22991 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
22993 2017-06-21 Jakub Jelinek <jakub@redhat.com>
22996 * config/i386/sse.md (round<mode>2): Renumber match_dup and
22997 operands indexes to avoid gap between operands and match_dups.
22999 2017-06-21 Andrew Pinski <apinski@cavium.com>
23001 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
23002 Increment Arith_shift and Arith_shift_reg by 1.
23003 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
23005 * config/aarch64/aarch64.c (thunderx_tunings): Enable
23006 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
23007 (aarch64_strip_extend): Add new argument and test for it.
23008 (aarch64_cheap_mult_shift_p): New function.
23009 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
23010 add a cost if it is true.
23011 Update calls to aarch64_strip_extend.
23012 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
23014 2017-06-21 Andrew Pinski <apinski@cavium.com>
23016 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
23018 (thunderxt88): Likewise.
23019 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
23020 (thunderx_prefetch_tune): New variable.
23021 (thunderx2t99_prefetch_tune): Update for the correct values.
23022 (thunderxt88_tunings): New variable.
23023 (thunderx_tunings): Use thunderx_prefetch_tune instead of
23024 generic_prefetch_tune.
23025 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
23027 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23029 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
23030 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
23031 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
23032 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
23033 (aarch64_atomic_cas<mode>, GPI): Likewise.
23035 2017-06-21 Martin Liska <mliska@suse.cz>
23037 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
23038 statements on cold and hot labels.
23039 * predict.c (tree_estimate_probability_bb): Remove the
23040 prediction from this place.
23042 2017-06-21 Martin Liska <mliska@suse.cz>
23044 PR tree-optimization/79489
23045 * gimplify.c (maybe_add_early_return_predict_stmt): New
23047 (gimplify_return_expr): Call the function.
23048 * predict.c (tree_estimate_probability_bb): Remove handling
23050 * predict.def: Update comment about early return predictor.
23051 * gimple-predict.h (is_gimple_predict): New function.
23052 * predict.def: Change default value of early return to 66.
23053 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
23055 * passes.def: Put pass_strip_predict_hints to the beginning of
23058 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
23060 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
23061 FUNCTION_DECL declarations.
23062 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
23064 (dwaf2out_decl): Likewise.
23065 * godump.c (go_early_global_decl): Skip call to the real debug hook
23066 for FUNCTION_DECL declarations.
23067 * passes.c (rest_of_decl_compilation): Skip call to the
23068 early_global_decl debug hook for FUNCTION_DECL declarations, unless
23069 -fdump-go-spec is passed.
23071 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23073 * config/i386/i386.c (struct builtin_isa): New field pure_p.
23074 Reorder for compactness.
23075 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
23076 (def_builtin_pure, def_builtin_pure2): New functions.
23077 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
23079 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
23081 * match.pd (nop_convert): New predicate.
23082 ((A +- CST1) +- CST2): Allow some NOP conversions.
23084 2017-06-21 Jakub Jelinek <jakub@redhat.com>
23087 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
23088 with ctors/dtors if GOVD_SHARED is set.
23090 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23092 * config/aarch64/aarch64.md (movti_aarch64):
23093 Emit mov rather than orr.
23094 (movtf_aarch64): Likewise.
23095 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
23096 Emit mov rather than orr.
23098 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23100 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
23101 Swap alternatives, make integer dup more expensive.
23103 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
23105 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23106 Return true for non-tls symbols.
23108 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
23110 * config/aarch64/aarch64-cores.def (cortex-a55): New.
23111 (cortex-a75): Likewise.
23112 (cortex-a75.cortex-a55): Likewise.
23113 * config/aarch64/aarch64-tune.md: Regenerate.
23114 * doc/invoke.texi (-mtune): Document new values for -mtune.
23116 2017-06-21 Tom de Vries <tom@codesourcery.com>
23118 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
23119 stack_size feature.
23120 (Effective-Target Keywords, Other attributes): Suggest using
23121 dg-add-options stack_size feature to get stack limit in stack_size
23122 effective target documentation.
23124 2017-06-21 Julian Brown <julian@codesourcery.com>
23125 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23127 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
23128 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
23129 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
23131 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
23132 attribute type list for neon_multiply.
23133 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
23134 attribute type list for neon_multiply.
23135 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
23136 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
23137 attribute type list for neon_multiply.
23138 * config/arm/types.md (crypto_pmull): Add.
23139 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
23140 attribute type list.
23142 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
23144 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
23147 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23149 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
23150 to make sure not to dereference a NULL cost_classes_ptr pointer.
23152 2017-06-20 Carl Love <cel@us.ibm.com>
23154 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23155 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23156 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23157 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23158 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23159 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23160 VMULOSW): New enum "unspec" values.
23161 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23162 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23163 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23164 altivec_vmulosw): New patterns.
23165 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23166 VMULOSW): Add definitions.
23168 2017-06-20 Julia Koval <julia.koval@intel.com>
23170 * config/i386/i386.c: Fix rounding expand for new pattern.
23171 * config/i386/subst.md: Fix pattern (parallel -> unspec).
23173 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23175 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
23176 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
23178 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23180 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
23183 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
23185 * config/aarch64/aarch64-cores.def: Rearrange to sort by
23186 architecture, then by implementer ID.
23187 * config/aarch64/aarch64-tune.md: Regenerate.
23189 2017-06-20 Richard Biener <rguenther@suse.de>
23191 PR middle-end/81097
23192 * fold-const.c (split_tree): Fold to type before negating.
23194 2017-06-20 David Malcolm <dmalcolm@redhat.com>
23196 * diagnostic-show-locus.c
23197 (selftest::test_fixit_deletion_affecting_newline): New function.
23198 (selftest::diagnostic_show_locus_c_tests): Call it.
23200 2017-06-20 Andreas Schwab <schwab@suse.de>
23203 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
23206 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23208 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
23209 __ARM_FEATURE_COPROC according to support.
23211 2017-06-20 Jakub Jelinek <jakub@redhat.com>
23213 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
23214 Rewritten to avoid overflow for > 32-bit pointers.
23217 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
23218 by removing enum keyword.
23219 (ubsan_type_descriptor): Likewise. Formatting fix.
23222 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
23223 splitter): Require TARGET_SSE2 in the condition.
23225 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
23228 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
23229 for doing vector set of SFmode on ISA 3.0.
23230 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
23231 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
23233 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
23234 SFmode value into a V4SF variable that was extracted from another
23235 V4SF variable without converting the element to double precision
23236 and back to single precision vector format.
23237 (vsx_insert_extract_v4sf_p9_2): Likewise.
23239 2017-06-19 Jakub Jelinek <jakub@redhat.com>
23241 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
23242 in UWHI to avoid undefined overflow.
23245 * ubsan.h (enum ubsan_encode_value_phase): New.
23246 (ubsan_encode_value): Change second argument to
23247 enum ubsan_encode_value_phase with default value of
23248 UBSAN_ENCODE_VALUE_GENERIC.
23249 * ubsan.c (ubsan_encode_value): Change second argument to
23250 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
23251 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
23252 create_tmp_var_raw instead of create_tmp_var and use a
23254 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
23255 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
23256 ubsan_encode_value callers.
23259 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
23260 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
23261 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
23263 2017-06-19 Richard Biener <rguenther@suse.de>
23265 PR middle-end/81118
23266 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
23267 estimates if we changed anything.
23269 2017-06-19 Richard Biener <rguenther@suse.de>
23271 PR tree-optimization/80887
23272 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
23273 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
23274 simplified lookups, then reset mprts_hook.
23275 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
23277 (try_to_simplify): Likewise.
23279 2017-06-19 Martin Liska <mliska@suse.cz>
23282 * gimplify.c (gimplify_switch_expr):
23283 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
23285 2017-06-19 Martin Liska <mliska@suse.cz>
23287 * doc/install.texi: Document that PGO runs in 4 stages.
23289 2017-06-19 Martin Liska <mliska@suse.cz>
23292 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
23293 to dispatcher function name.
23294 * multiple_target.c (replace_function_decl): New function.
23295 (create_dispatcher_calls): Redirect both edges and references.
23297 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
23299 * profile-count.c (profile_count::dump): Dump quality.
23300 (profile_count::differs_from_p): Update for unsigned val.
23301 * profile-count.h (profile_count_quality): New enum.
23302 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
23304 2017-06-19 Richard Biener <rguenther@suse.de>
23306 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
23307 struct function as arg.
23308 (estimate_numbers_of_iterations): Export overload with loop arg.
23309 (free_numbers_of_iterations_estimates_loop): Use an overload of
23310 free_numbers_of_iterations_estimates instead.
23311 * tree-cfg.c (remove_bb): Adjust.
23312 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
23313 * tree-parloops.c (gen_parallel_loop): Likewise.
23314 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
23316 (tree_unroll_loops_completely): Likewise.
23317 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
23318 Use an overload instead and export.
23319 (estimated_loop_iterations): Adjust.
23320 (max_loop_iterations): Likewise.
23321 (likely_max_loop_iterations): Likewise.
23322 (estimate_numbers_of_iterations): Take struct function as arg
23324 (loop_exits_before_overflow): Adjust.
23325 (free_numbers_of_iterations_estimates_loop): Use an overload.
23326 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
23327 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
23329 2017-06-19 Richard Biener <rguenther@suse.de>
23332 * ipa-prop.c (find_constructor_constant_at_offset): Handle
23333 RANGE_EXPR conservatively.
23335 2017-06-16 Carl Love <cel@us.ibm.com>
23337 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23338 definitions for vec_float, vec_float2, vec_floato,
23339 vec_floate built-ins.
23340 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
23341 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
23343 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
23344 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
23345 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
23346 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
23347 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
23348 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
23349 vec_floato): Add builtin defines.
23350 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
23351 Update the built-in documentation file for the new built-in
23354 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23356 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
23357 (mthumb): Mark as the negative of -marm.
23359 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23361 * doc/invoke.texi (ARM Options, -mcpu): Document supported
23363 (ARM Options, -mtune): Document that this accepts the same
23364 extension options as -mcpu.
23365 (ARM Options, -mfpu): Document addition of -mfpu=auto.
23367 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23369 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
23370 permitted extensions.
23372 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23374 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
23375 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
23376 (armv8-m.main): Add option +nodsp.
23377 * config/arm/arm-cpu-cdata.h: Regenerated.
23379 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23381 * config/arm/t-fuchsia: New file.
23382 * config.gcc (arm*-*-fuchsia*): Use it.
23384 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23386 * config/arm/t-symbian: Rewrite for new option infrastructure.
23388 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23390 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
23391 (MULTILIB_REQUIRED): Likewise.
23393 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23395 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
23396 (MULTILIB_RESUE): Likewise.
23397 (MULTILIB_MATCHES): Likewise.
23398 (MULTLIB_REQUIRED): Likewise.
23400 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23402 * config/arm/t-rtems: Rewrite for new option framework.
23404 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23406 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
23407 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
23408 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
23409 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
23410 * config/arm/t-multilib: ... here.
23411 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
23412 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
23413 armv7-a and armv8*-a when A-profile libraries have not been built.
23414 * config/arm/t-rmprofile: Rewrite.
23416 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23418 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
23419 with a backslash. Remove the backslash after substituting unescaped
23421 * doc/fragments.texi (MULTILIB_REUSE): Document it.
23423 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23425 * config.gcc: (arm*-*-*): When building a-profile libraries, force
23426 the driver to pass through the default setting of -mfloat-abi.
23427 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
23429 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
23430 (all_feat_combs): New rule.
23431 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
23433 * config/arm/t-aprofile: Rewrite.
23435 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23437 * config/arm/arm.h (FPUTYPE_AUTO): Define.
23438 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
23439 fpu is not specified by the user/command-line.
23440 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
23441 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
23442 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
23443 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
23444 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
23445 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
23447 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23449 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
23450 * config/arm/t-arm-elf: Rewritten.
23452 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23454 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
23455 have some floating-point instructions.
23456 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
23457 (TARGET_MAYBE_HARD_FLOAT): New macro.
23458 * config/arm/arm-builtins.c (arm_init_builtins): Use
23459 TARGET_MAYBE_HARD_FLOAT.
23460 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
23462 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23464 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
23465 (configargs.h): Include it.
23466 (arm_print_hint_for_fpu_option): New function.
23467 (arm_parse_fpu_option): New function.
23468 (candidate_extension): New class.
23469 (arm_canon_for_multilib): New function.
23470 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
23471 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
23472 (ARCH_CANONICAL_SPECS): New macro.
23473 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
23475 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23477 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
23478 are set after handling multilib fragments. Set target_cpu_default2
23481 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23483 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
23485 (arm*-*-*): Set target_cpu_default2 to a quoted string.
23486 * config/arm/parsecpu.awk (check_cpu): Validate any extension
23488 (check_arch): Likewise.
23489 * config/arm/arm.c (arm_configure_build_target): Handle
23490 TARGET_CPU_DEFAULT being a string constant. Scan any feature
23491 options in the default.
23493 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23495 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
23496 when an option is an alias of another.
23497 * config/arm/parsecpu.awk (optalias): New parser token.
23498 (gen_comm_data): Mark non-alias options as such. Emit entries
23499 for extension aliases.
23500 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
23501 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
23502 (armv6kz, armv6zk, armv6t2): Likewise.
23503 (armv7): Make vfpv3-d16 an alias.
23504 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
23506 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
23507 Sort in canonical order.
23508 (armv8-a): Sort in canonical order.
23509 (armv8.1-a, armv8.2-a): Likewise.
23510 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
23512 (cortex-a9): Sort in canonical order.
23513 * config/arm/arm.c (selftests.h): Include it.
23514 (arm_test_cpu_arch_data): New function.
23515 (arm_run_self_tests): New function.
23516 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
23517 (targetm): Move declaration to the end of the file.
23518 * arm-cpu-cdata.h: Regenerated.
23520 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23522 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
23523 call to target_mode_check describing the type of option passed.
23524 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
23525 (arm_target_thumb_only): Use arm_parse_arch_option_name or
23526 arm_parse_cpu_option_name to match parameters against list of
23528 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
23529 arm_arch_core_flags data structure.
23530 * config/arm/arm-cpu_cdata.h: Regenerated.
23532 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23534 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
23536 (arm_print_hint_for_cpu_option): Likewise.
23537 (arm_print_hint_for_arch_option): Likewise.
23538 (arm_parse_cpu_option_name): Likewise.
23539 (arm_parse_arch_option_name): Likewise.
23540 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
23541 of entries in the all_fpus list.
23542 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
23543 (arm_parse_cpu_option_name): Declare.
23544 (arm_parse_arch_option_name): Declare.
23545 (arm_parse_option_features): Declare.
23546 (arm_intialize_isa): Declare.
23547 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
23549 (gen_comm_data): ... here. Make definitions non-static.
23550 * config/arm/arm-cpu-data.h: Regenerated.
23551 * config/arm/arm-cpu-cdata.h: Regenerated.
23553 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23555 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
23556 (cpu_arch_extension): New structure.
23557 (cpu_arch_option, arch_option, cpu_option): New structures.
23558 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
23559 architecture types.
23560 (gen_data): Generate new format data tables.
23561 * config/arm/arm.c (cpu_tune): New structure.
23562 (cpu_option, processors): Delete.
23563 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
23564 (arm_print_hint_for_cpu_option): ... this and ...
23565 (arm_print_hint_for_arch_option): ... this.
23566 (arm_parse_arch_cpu_name): Delete. Replace with ...
23567 (arm_parse_cpu_option_name): ... this and ...
23568 (arm_parse_arch_option_name): ... this.
23569 (arm_unrecognized_feature): Change type of target parameter to
23571 (arm_parse_arch_cpu_features): Delete. Replace with ...
23572 (arm_parse_option_features): ... this.
23573 (arm_configure_build_target): Rework to use new configuration data
23575 (arm_print_tune_info): Rework for new configuration data tables.
23576 * config/arm/arm-cpu-data.h: Regenerated.
23577 * config/arm/arm-cpu.h: Regenerated.
23579 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23581 * Makefile.in (OBJS): Move sbitmap.o from here ...
23582 (OBJS-libcommon): ... to here.
23584 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23586 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
23587 (ISA_ALL_CRYPTO): New macro.
23588 (ISA_ALL_SIMD): New macro
23589 (ISA_ALL_FP): New macro.
23590 * config/arm/arm.c (fpu_bitlist): Update initializer.
23591 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
23593 (arm9e): Add fpu. Add option for nofp
23594 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
23595 (arm926ej-s, arm1026ej-s): Likewise.
23596 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
23597 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
23598 neon-fp16, neon-vfpv4, nofp and nosimd.
23599 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
23600 (cortex-a8): Add fpu. Add option for nofp.
23601 (cortex-a9): Add fpu. Add options for nosimd and nofp.
23602 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
23603 (cortex-r4f): Add fpu.
23604 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
23605 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
23607 (cortex-r8): Likewise.
23608 (cortex-m4): Add fpu. Add option for nofp.
23609 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
23610 (cortex-a17.cortex-a7): Likewise.
23611 (cortex-a32): Add fpu. Add options for crypto and nofp.
23612 (cortex-a35, cortex-a53): Likewise.
23613 (cortex-a57): Add fpu. Add option for crypto.
23614 (cortex-a72, cortex-a73): Likewise.
23615 (exynos-m1): Likewise.
23616 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
23617 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
23618 (cortex-m33): Add fpu. Add option for nofp.
23619 * config/arm/arm-cpu-cdata.h: Regenerated
23620 * config/arm/arm-cpu-data.h: Regenerated.
23622 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23624 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
23625 (armv5te, armv5tej): Likewise.
23626 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
23627 (armv7): Add options fp and vfpv3-d16.
23628 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
23629 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
23631 (armv7ve): Likewise.
23632 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
23633 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
23634 (armv8-a): Add nocrypto option.
23635 (armv8.1-a, armv8.2-a): Likewise.
23636 (armv8-m.main): add options fp, fp.dp and nofp.
23638 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23640 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
23642 (armv8-a+crc): Delete.
23643 (armv8.1-a): Add options simd, crypto and nofp.
23644 (armv8.2-a): Add options fp16, simd, crypto and nofp.
23645 (armv8.2-a+fp16): Delete.
23646 (armv8-m.main): Add option dsp.
23647 (armv8-m.main+dsp): Delete.
23648 (cortex-a8): Add fpu. Add nofp option.
23649 (cortex-a9): Add fpu. Add nofp and nosimd options.
23650 * config/arm/parsecpu.awk (gen_data): Generate option tables and
23651 link to main cpu and architecture data structures.
23652 (gen_comm_data): Only put isa attributes from the main architecture
23654 (option): New statement for architecture and CPU entries.
23655 * arm.c (struct cpu_option): New structure.
23656 (struct processors): Add entry for options.
23657 (arm_unrecognized_feature): New function.
23658 (arm_parse_arch_cpu_name): Ignore any characters after the first
23660 (arm_parse_arch_cpu_feature): New function.
23661 (arm_configure_build_target): Separate out any CPU and architecture
23662 features and parse separately. Don't error out if -mfpu=auto is
23663 used with only an architecture string.
23664 (arm_print_asm_arch_directives): New function.
23665 (arm_file_start): Call it.
23666 * config/arm/arm-cpu-cdata.h: Regenerated.
23667 * config/arm/arm-cpu-data.h: Likewise.
23668 * config/arm/arm-tables.opt: Likewise.
23670 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23672 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
23673 assembler when it is not -mfpu=auto.
23675 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23677 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
23678 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
23679 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
23680 (ASM_CPU_SPEC): Rewrite.
23681 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
23682 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
23683 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
23684 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
23685 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
23686 copied string is NUL-terminated. Also strip any characters prefixed
23688 (arm_rewrite_selected_arch): New function.
23689 (arm_rewrite_march): New function.
23691 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
23693 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
23694 (x_arm_cpu_string, x_arm_tune_string): Likewise.
23695 (march, mcpu, mtune): Convert to string-based options.
23696 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
23697 (arm_parse_arch_cpu_name): New function.
23698 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
23699 identify selected architecture or CPU.
23700 (arm_option_save): New function.
23701 (TARGET_OPTION_SAVE): Redefine.
23702 (arm_option_restore): Restore string options.
23703 (arm_option_print): Print string options.
23705 2017-06-16 Martin Sebor <msebor@redhat.com>
23707 PR tree-optimization/80933
23708 PR tree-optimization/80934
23709 * builtins.c (fold_builtin_3): Do not handle bcmp here.
23710 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
23711 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
23712 (gimple_fold_builtin): Call them.
23714 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23716 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
23717 as unlikely; update profile.
23719 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23721 * predict.c (force_edge_cold): Handle declaring edges impossible
23724 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23726 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
23728 (try_unroll_loop_completely): Fix reporting.
23730 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23732 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
23734 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
23737 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
23738 if given a non-constant argument for an intrinsic which requires a
23741 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23743 * profile.c (compare_freqs): New function.
23744 (branch_prob): Sort edge list.
23745 (find_spanning_tree): Assume that the list is priority sorted.
23747 2017-06-16 Richard Biener <rguenther@suse.de>
23749 PR tree-optimization/81090
23750 * passes.def (pass_record_bounds): Remove.
23751 * tree-pass.h (make_pass_record_bounds): Likewise.
23752 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
23753 make_pass_record_bounds): Likewise.
23754 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
23755 not free niter estimates at the beginning but at the end.
23756 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
23758 2017-06-16 Richard Biener <rguenther@suse.de>
23760 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
23761 initializer to workaround ICE in host GCC 4.8.
23763 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
23765 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
23767 (clone_inlined_nodes): Update.
23769 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23771 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
23772 prefetch settings, and enable prefetching by default at -O3.
23774 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23776 * config/aarch64/aarch64.c (aarch64_override_options_internal):
23777 Set flag_prefetch_loop_arrays according to tuning data.
23779 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
23781 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
23782 New tune structure.
23783 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
23784 [Unrelated to main purpose of the patch] Place the pointer field last
23785 to enable type checking errors when tune structure are wrongly merged.
23786 * config/aarch64/aarch64.c (generic_prefetch_tune,)
23787 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
23788 (thunderx2t99_prefetch_tune): New tune constants.
23789 (tune_params *_tunings): Update all tunings (no functional change).
23790 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
23791 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
23792 from tunings structures.
23794 2017-06-16 Jakub Jelinek <jakub@redhat.com>
23797 * ubsan.c (instrument_null): Add T argument, use it instead
23798 of computing it based on IS_LHS.
23799 (instrument_object_size): Likewise.
23800 (pass_ubsan::execute): Adjust instrument_null and
23801 instrument_object_size callers to pass gimple_get_lhs or
23802 gimple_assign_rhs1 result to it. Use instrument_null instead of
23803 calling get_base_address and instrument_mem_ref. Handle
23804 aggregate call arguments for object-size sanitization.
23806 2017-06-16 Yury Gribov <tetra2005@gmail.com>
23808 PR tree-optimization/81089
23809 * tree-vrp.c (is_masked_range_test): Validate operands of
23812 2017-06-15 Martin Sebor <msebor@redhat.com>
23815 * dumpfile.c (dump_register): Avoid calling memset to initialize
23816 a class with a default ctor.
23817 * gcc.c (struct compiler): Remove const qualification.
23818 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
23819 * hash-table.h: Ditto.
23820 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
23822 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
23823 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
23825 * params.h (struct param_info): Make struct members non-const.
23826 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
23827 with default initialization.
23828 * vec.h (vec_copy_construct, vec_default_construct): New helper
23830 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
23831 with vec_copy_construct.
23832 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
23833 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
23834 * doc/invoke.texi (-Wclass-memaccess): Document.
23836 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23838 * emit-rtl.h (is_leaf): Update comment about local
23839 register allocator.
23841 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23844 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
23845 for a variable to have a section before checking if the section has a
23847 Set section to.persistent if persistent attribute is set.
23848 Warn if .persistent attribute is used on an automatic variable.
23850 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
23852 PR rtl-optimization/80474
23853 * reorg.c (update_block): Do not ignore instructions in a delay slot.
23855 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
23857 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
23860 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
23862 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
23863 (casesi): Emit bounds checking as RTL.
23864 (casesi_internal_mips16_<mode>): Remove bounds checking.
23866 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
23868 * config/xtensa/xtensa.c (xtensa_option_override): Append
23869 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
23870 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
23871 xtensa_doloop_hooks): Define unconditionally.
23872 (xtensa_reorg_loops): Only call reorg_loops in the presence of
23874 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
23875 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
23876 for it in xtensa_option_override.
23877 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
23878 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
23880 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
23882 * doc/cppopts.texi: Document '-' special value to -MF.
23884 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
23886 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
23887 (cortex_a53_fconst): Likewise.
23888 (cortex_a53_fpmul): Likewise.
23889 (cortex_a53_f_load_64): Likewise.
23890 (cortex_a53_f_load_many): Likewise.
23891 (cortex_a53_advsimd_alu): Likewise.
23892 (cortex_a53_advsimd_alu_q): Likewise.
23893 (cortex_a53_advsimd_mul): Likewise.
23894 (cortex_a53_advsimd_mul_q): Likewise.
23895 (fpmac bypass): Add new bypass for fpmac-fpmac case.
23896 Add missing fmul, r2f_cvt and fconst cases.
23898 2017-06-14 Richard Biener <rguenther@suse.de>
23900 PR middle-end/81088
23901 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
23903 (fold_binary_loc): When associating do not treat pre-existing
23904 TREE_OVERFLOW on literal constants as a reason to allow
23905 TREE_OVERFLOW on associated literal constants.
23907 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23909 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
23910 (MASK_FEATURES): New macro.
23911 * config/sparc/sparc.c (sparc_option_override): Remove the special
23912 handling of -mfpu and generalize it to all MASK_FEATURES switches.
23914 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23916 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
23917 a division of 0 if non-call exceptions are enabled.
23919 2017-06-14 Andrew Pinski <apinski@cavium.com>
23920 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23923 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
23924 Improve vector initialization code gen for only variable case.
23926 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
23928 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
23930 2017-06-14 Richard Biener <rguenther@suse.de>
23932 PR tree-optimization/81083
23933 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
23936 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23938 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
23939 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
23940 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
23941 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
23942 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
23943 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
23945 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23947 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
23948 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
23950 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23952 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
23954 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23956 * config/rs6000/t-rtems: Don't handle SPE.
23958 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23960 * config/rs6000/t-linux: Don't handle SPE.
23962 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23964 * config/rs6000/eabispe.h: Delete file.
23966 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23968 * config/rs6000/t-spe: Delete file.
23970 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23972 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
23973 (rs6000_legitimate_offset_address_p): Return false for anything in
23974 V2SImode or V2SFmode.
23976 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
23978 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
23979 except V2SF and V2SI. Rearrange the vector modes, and add comments.
23980 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
23982 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
23983 (rs6000_legitimate_offset_address_p): Ditto.
23984 (rs6000_emit_move): Ditto.
23985 (rs6000_init_builtins): Remove V4HI_type_node.
23987 2017-06-13 Martin Liska <mliska@suse.cz>
23990 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
23991 (gate_asan): Likewise.
23992 * asan.h (asan_no_sanitize_address_p): Remove the function.
23993 (sanitize_flags_p): New function.
23994 * builtins.def: Fix coding style.
23995 * common.opt: Use renamed enum value.
23996 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
23997 * doc/extend.texi: Document no_sanitize attribute.
23998 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
23999 to SANITIZE_UNDEFINED_NONDEFAULT.
24000 * gcc.c (sanitize_spec_function): Use the renamed enum value.
24001 * gimple-fold.c (optimize_atomic_compare_exchange_p):
24002 Use sanitize_flags_p.
24003 * gimplify.c (gimplify_function_tree): Likewise.
24004 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
24005 * opts.c (parse_no_sanitize_attribute): New function.
24006 (common_handle_option): Use renamed enum value.
24007 * opts.h (parse_no_sanitize_attribute): Declare.
24008 * tree.c (sanitize_flags_p): New function.
24009 * tree.h: Declared here.
24010 * tsan.c: Use sanitize_flags_p.
24011 * ubsan.c (ubsan_expand_null_ifn): Likewise.
24012 (instrument_mem_ref): Likewise.
24013 (instrument_bool_enum_load): Likewise.
24014 (do_ubsan_in_current_function): Remove the function.
24015 (pass_ubsan::execute): Use sanitize_flags_p.
24016 * ubsan.h: Remove do_ubsan_in_current_function
24017 * tree-cfg.c (print_no_sanitize_attr_value): New function.
24018 (dump_function_to_file): Use it here.
24020 2017-06-13 Martin Jambor <mjambor@suse.cz>
24022 PR tree-optimization/80803
24023 PR tree-optimization/81063
24024 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
24025 (propagate_subaccesses_across_link): Enqueue subtree whenever
24026 necessary instead of relying on the caller.
24028 2017-06-13 Martin Jambor <mjambor@suse.cz>
24030 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
24031 that have a first_link.
24032 (sort_and_splice_var_accesses): Do not check first_link before
24034 (subtree_mark_written_and_enqueue): Likewise.
24035 (propagate_all_subaccesses): Likewise and do not stop at first
24036 parent with a first_link.
24038 2017-06-13 Martin Jambor <mjambor@suse.cz>
24040 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
24043 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24045 * match.pd: New pattern.
24047 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24049 * tree-vrp.c (is_masked_range_test): New function.
24050 (register_edge_assert_for): Determine ranges for
24053 2017-06-13 Yury Gribov <tetra2005@gmail.com>
24055 PR tree-optimization/67328
24056 * fold-const.c (maskable_range_p): New function.
24057 (build_range_check): Generate bittests if possible.
24059 2017-06-13 Martin Liska <mliska@suse.cz>
24061 * gimple-pretty-print.c (dump_probability): Add new argument.
24062 (dump_edge_probability): Dump both probability and count.
24063 (dump_gimple_label): Likewise.
24064 (dump_gimple_bb_header): Likewise.
24066 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
24069 * config/avr/avr-devices.c: Fix indentation.
24070 * config/avr/gen-avr-mmcu-specs.c: Dito.
24072 2017-06-13 Richard Biener <rguenther@suse.de>
24074 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
24075 instead get vector type from stmt_info.
24076 (vectorizable_reduction): Adjust. Remove dead code.
24078 2017-06-13 Richard Biener <rguenther@suse.de>
24080 PR middle-end/81065
24081 * fold-const.c (extract_muldiv_1): Remove bogus distribution
24082 case of C * (x * C2 + C3).
24083 (fold_addr_of_array_ref_difference): Properly fold index difference.
24085 2017-06-12 David S. Miller <davem@davemloft.net>
24088 * config/sparc/sparc.md (return expander): Emit frame blockage if
24089 function uses alloca.
24091 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
24093 * combine.c (make_field_assignment): Check len rather than the mode
24094 precision when calling force_to_mode.
24096 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
24098 Support multilibs and devices that see flash in RAM address range.
24101 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
24102 (avr_mcu_t) <flash_pm_offset>: New field.
24103 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
24104 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
24105 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
24106 (AVR_TINY_PM_OFFSET): Remove macro.
24107 * config/avr/avr.opt (-mshort-calls): New option.
24108 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24109 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
24110 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
24111 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
24112 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
24113 instead of avr_arch->have_jmp_call.
24114 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
24115 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
24116 avr_arch->flash_pm_offset to define.
24117 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
24118 new field flash_pm_offset. Add entry for avrxmega3.
24119 (avr_texinfo): Add entry for avrxmega3.
24120 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
24121 attiny212, attiny214,
24122 attiny412, attiny414, attiny416, attiny417,
24123 attiny814, attiny816, attiny817,
24124 attiny1614, attiny1616, attiny1617,
24125 attiny3214, attiny3216, attiny3217.
24126 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
24127 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
24128 (avr_print_operand_address) [AVR_TINY]: Same.
24129 (avr_asm_init_sections) <readonly_data_section>: Only patch
24130 callback if avr_arch->flash_pm_offset = 0.
24131 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
24132 for rodata if avr_arch->flash_pm_offset != 0.
24133 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
24134 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
24135 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
24136 (m_options): Append opt_rcall.
24137 (m_dirnames): Append dir_rcall.
24138 * config/avr/t-multilib: Regenerate.
24140 * configure.ac [target=avr]: Check whether avrxmega3 default
24141 linker description file works as needed.
24142 * configure: Regenerate.
24143 * doc/avr-mmcu.texi: Regenerate.
24144 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
24145 <__AVR_ARCH__>: Document avrxmega3 and 103.
24146 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
24147 <__AVR_SHORT_CALLS__>: Document it.
24148 <__AVR_PM_BASE_ADDRESS__>: Document it.
24149 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
24150 (AVR Variable Attributes) <progmem>: Document this is
24151 not needed for avrxmega3.
24152 (AVR Named Address Spaces) <__flash>: Dito.
24154 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
24156 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
24158 2017-06-12 Doug Rupp <rupp@adacore.com>
24160 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
24161 Append vxworks-stdint.h to the tm_file list.
24162 * config/vxworks-stdint.h: New file.
24164 2017-06-12 Martin Liska <mliska@suse.cz>
24166 PR tree-optimization/81041
24167 * tree-profile.c (gimple_gen_ic_func_profiler):
24168 Create an extra BB in profile-generate
24169 (gimple_gen_time_profiler): Likewise.
24171 2017-06-12 Jakub Jelinek <jakub@redhat.com>
24173 PR tree-optimization/81003
24174 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
24175 (update_range_test): Use it instead of force_gimple_operand_gsi.
24177 2017-06-12 Richard Biener <rguenther@suse.de>
24179 PR tree-optimization/81053
24180 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
24181 with backedge value not defined in loop. Simplify def stmt
24184 2017-06-11 Tom de Vries <tom@codesourcery.com>
24187 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
24189 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
24190 nvptx_cannot_force_const_mem.
24192 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24194 * opts.c (finish_options): Move test for flag_split_stack after
24195 it has been initialized.
24197 2017-06-11 Jason Merrill <jason@redhat.com>
24199 * tree.h (id_equal): New.
24200 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
24201 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
24202 instead of strcmp of IDENTIFIER_POINTER.
24204 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24206 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
24207 (mark_all_inlined_calls_cdtor): Fix formating.
24208 (inline_transform): Rescale profile before inlining.
24210 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24212 * cgraph.h (cgraph_edge::clone): Update prototype.
24213 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
24214 (cgraph_node::create_clone): Update.
24215 (cgraph_node::create_version_clone): Update.
24216 * tree-inline.c (copy_bb): Update.
24217 (expand_call_inline): Update.
24219 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24221 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
24222 factored out from ...
24223 (rs6000_emit_prologue): ... here.
24225 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
24227 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
24228 factored out from ...
24229 (rs6000_emit_prologue): ... here.
24231 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24233 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
24235 (handle_missing_profiles): Fix computation of tp_first_run.
24236 (counts_to_freqs): Do not touch freqs when count is 0.
24238 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
24240 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
24243 2017-06-10 Tom de Vries <tom@codesourcery.com>
24245 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24246 attributes): Document signal effective target.
24248 2017-06-10 Tom de Vries <tom@codesourcery.com>
24250 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24251 Document effective target stack_size.
24253 2017-06-09 David Malcolm <dmalcolm@redhat.com>
24255 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
24256 to the edit_context if they can be auto-applied.
24258 2017-06-9 Ian Lance Taylor <iant@golang.org>
24260 * opts.c (finish_options): If -fsplit-stack, disable implicit
24261 -forder-blocks-and-partition.
24262 * doc/invoke.texi (Optimize Options): Document that when using
24263 -fsplit-stack -forder-blocks-and-partition is not implicitly
24266 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24268 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
24269 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
24270 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
24271 * builtins.def (abort, trap, unreachable): Declare cold.
24272 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
24273 * tree-core.h (ECF_COLD): New.
24274 * tree.c (set_call_expr_flags): Handle ECF_COLD.
24275 (build_common_builtin_nodes): Mark unreachable and abort as cold.
24277 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24279 * predict.c (unlikely_executed_stmt_p): Cleanup.
24281 2017-06-09 Richard Biener <rguenther@suse.de>
24283 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
24284 model if the ref is always written to.
24286 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24288 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
24290 2017-06-09 Tamar Christina <tamar.christina@arm.com>
24292 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
24295 2017-06-09 Tom de Vries <tom@codesourcery.com>
24298 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
24299 "target cannot support label values" when encountering LABEL_REF.
24301 2017-06-09 Martin Liska <mliska@suse.cz>
24303 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
24304 (gimple_gen_ic_func_profiler): Emit direct comparison
24305 of __gcov_indirect_call_callee with NULL.
24306 (gimple_gen_time_profiler): Change probability from
24307 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
24309 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24311 * profile.c (edge_gcov_counts): Turn to pointer.
24312 (compute_branch_probabilities, compute_branch_probabilities): Update.
24313 (branch_prob): Do not clear edge_gcov_count.
24314 * profile.h (edge_gcov_counts): Turn to pointer.
24315 (edge_gcov_count): Update.
24317 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
24319 * gimple.h (gimple_check_failed): Mark cold.
24321 2017-06-09 Richard Biener <rguenther@suse.de>
24323 PR tree-optimization/66623
24324 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
24325 refactor check_reduction into two parts, properly computing
24326 whether we have to check reduction validity for outer loop
24329 2017-06-09 Richard Biener <rguenther@suse.de>
24331 PR tree-optimization/79483
24332 * graphite-scop-detection.c (order): New global.
24333 (get_order): Compute bb to order mapping that satisfies code
24334 generation constraints.
24335 (cmp_pbbs): New helper.
24336 (build_scops): Start domwalk at entry block, sort generated
24339 2017-06-09 Richard Biener <rguenther@suse.de>
24341 PR middle-end/81007
24342 * ipa-polymorphic-call.c
24343 (ipa_polymorphic_call_context::restrict_to_inner_class):
24344 Skip FIELD_DECLs with error_mark_node type.
24345 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
24348 2017-06-09 Martin Liska <mliska@suse.cz>
24350 * predict.c (struct branch_predictor): New struct.
24351 (test_prediction_value_range): New test.
24352 (predict_c_tests): New function.
24353 * selftest-run-tests.c (selftest::run_tests): Run the function.
24354 * selftest.h: Declare new tests.
24356 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
24359 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
24360 gen_add3_insn did not fail.
24361 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
24362 r0, construct that number in a temporary reg and add that reg to r0.
24363 If asked to put the result in r0 as well, fail.
24365 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
24367 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
24368 for early expansion of vec_eqv.
24370 2017-06-08 Jakub Jelinek <jakub@redhat.com>
24372 PR middle-end/81005
24373 * ubsan.c (instrument_null): Avoid pointless code temporary.
24374 (pass_ubsan::execute): Instrument aggregate arguments of calls.
24376 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
24380 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
24383 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
24386 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24388 * predict.c (unlikely_executed_edge_p): Move ahead.
24389 (probably_never_executed_edge_p): Use it.
24391 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24393 PR middle-end/79988
24394 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
24395 gimple_call_builtin_p call.
24397 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24399 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
24400 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
24401 rtl_check_failed_type2, rtl_check_failed_code1,
24402 rtl_check_failed_code2, rtl_check_failed_code_mode,
24403 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
24404 rtvec_check_failed_bounds, rtl_check_failed_flag,
24405 _fatal_insn_not_found, _fatal_insn): Likewise.
24406 * tree.h (tree_contains_struct_check_failed,
24407 tree_check_failed, tree_not_check_failed,
24408 tree_class_check_failed, tree_range_check_failed,
24409 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
24410 tree_vec_elt_check_failed, phi_node_elt_check_failed,
24411 tree_operand_check_failed, omp_clause_check_failed,
24412 omp_clause_operand_check_failed, omp_clause_range_check_failed):
24415 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24417 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
24418 flag_branch_probabilities.
24419 * ipa-inline.c (edge_badness): Likewise.
24420 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
24421 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
24422 * predict.c (maybe_hot_frequency_p): Likewise.
24423 (probably_never_executed): Likewise.
24424 * sched-ebb.c (schedule_ebbs): Likewise.
24425 * sched-rgn.c (find_single_block_region): Likewise.
24426 * tracer.c (tail_duplicate): Likewise.
24428 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24430 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
24431 longer requires x_flag_profile_use.
24433 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24435 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
24436 instead of flag_reorder_blocks_and_partition.
24437 * dbxout.c (dbxout_function_end): Likewise.
24438 * dwarf2out.c (gen_subprogram_die): Likewise.
24439 * haifa-sched.c (sched_create_recovery_edges): Likewise.
24440 * hw-doloop.c (reorg_loops): Likewise.
24441 * varasm.c (assemble_start_function,
24442 assemble_end_function): Likewise.
24443 (decide_function_section): Do not check for
24444 flag_reorder_blocks_and_partition.
24446 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
24448 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24450 (chkp_get_hard_register_fake_addr_expr): Ditto.
24451 (chkp_build_addr_expr): Add check for hard reg case.
24452 (chkp_parse_array_and_component_ref): Ditto.
24453 (chkp_find_bounds_1): Ditto.
24454 (chkp_process_stmt): Don't generate bounds store for
24457 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24459 * predict.c (maybe_hot_bb_p): Do not check profile status.
24460 (maybe_hot_edge_p): Likewise.
24461 (probably_never_executed): Check for zero counts even if profile
24463 (unlikely_executed_edge_p): New function.
24464 (unlikely_executed_stmt_p): New function.
24465 (unlikely_executed_bb_p): New function.
24466 (set_even_probabilities): Use unlikely predicates.
24467 (combine_predictions_for_bb): Likewise.
24468 (predict_paths_for_bb): Likewise.
24469 (predict_paths_leading_to_edge): Likewise.
24470 (determine_unlikely_bbs): New function.
24471 (estimate_bb_frequencies): Use it.
24472 (compute_function_frequency): Use zero counts even if profile is
24474 * profile-count.h: Fix typo.
24476 2017-08-08 Julia Koval <julia.koval@intel.com>
24478 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
24479 _mm512_mask_cvtsepi16_storeu_epi8,
24480 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
24481 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
24482 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
24483 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
24484 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
24485 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
24486 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
24487 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
24488 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
24489 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
24490 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
24491 __builtin_ia32_pmovuswb256mem_mask,
24492 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
24493 __builtin_ia32_pmovwb512mem_mask): New builtins.
24495 2017-08-08 Julia Koval <julia.koval@intel.com>
24497 PR target/73350,80862
24498 * config/i386/subst.md (round): Fix round pattern.
24499 * config/i386/i386.c (ix86_erase_embedded_rounding):
24500 Fix erasing rounding for the fixed pattern.
24502 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
24504 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
24506 2017-06-08 Martin Liska <mliska@suse.cz>
24508 PR gcov-profile/80911
24509 * gcov.c (block_info::block_info): New constructor.
24511 2017-06-07 Carl Love <cel@us.ibm.com>
24513 * config/rs6000/rs6000-c: The return type of the following
24514 built-in functions was implemented as int not long long. Fix sign
24515 of return value for the unsigned version of vec_mulo and vec_mule.
24516 vector unsigned long long vec_bperm (vector unsigned long long,
24517 vector unsigned char)
24518 vector signed long long vec_mule (vector signed int,
24520 vector unsigned long long vec_mule (vector unsigned int,
24521 vector unsigned int)
24522 vector signed long long vec_mulo (vector signed int,
24524 vector unsigned long long vec_mulo (vector unsigned int,
24525 vector unsigned int)
24526 * doc/extend.texi: Fix the documentation for the built-in
24529 2017-06-07 Carl Love <cel@us.ibm.com>
24532 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
24535 2017-06-07 Carl Love <cel@us.ibm.com>
24537 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
24538 support, Generate doublehv for signed int/float for BE case only.
24540 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
24542 * doc/invoke.texi (mcx16): Rewrite.
24544 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24546 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
24547 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
24548 *mov<mode>_softfloat, and an anonymous splitter): Use
24549 nonimmediate_operand instead of rs6000_nonimmediate_operand.
24551 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24553 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
24555 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24556 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
24557 (rs6000_debug_reg_global): Adjust.
24558 (rs6000_init_hard_regno_mode_ok): Adjust.
24559 (rs6000_dbx_register_number): Adjust.
24560 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
24561 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24562 Remove SPE_ACC and SPEFSCR.
24563 (REG_ALLOC_ORDER): Ditto.
24564 (FRAME_POINTER_REGNUM): Change to 111.
24565 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
24566 (REG_CLASS_NAMES): Ditto.
24567 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
24568 (REGISTER_NAMES): Ditto.
24569 (ADDITIONAL_REG_NAMES): Ditto.
24570 (rs6000_reg_names): Ditto.
24571 * config/rs6000/rs6000.md: Renumber some register number
24574 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24576 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
24578 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
24579 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
24581 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
24582 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
24583 Delete the SPE high registers.
24584 (REG_ALLOC_ORDER): Ditto.
24585 (enum reg_class): Remove SPE_HIGH_REGS.
24586 (REG_CLASS_NAMES): Ditto.
24587 (REG_CLASS_CONTENTS): Delete the SPE high registers.
24588 (REGISTER_NAMES): Ditto.
24589 (rs6000_reg_names): Ditto.
24590 * doc/tm.texi.in: Remove SPE as example.
24591 * doc/tm.texi: Regenerate.
24593 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24595 * config/rs6000/8540.md (ppc8540_brinc): Delete.
24596 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
24597 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
24598 * config/rs6000/rs6000.md (type): Remove "brinc".
24600 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24602 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
24603 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
24604 * config/rs6000/linuxspe.h: Delete file.
24605 * config/rs6000/rs6000.md: Don't include spe.md.
24606 * config/rs6000/spe.h: Delete file.
24607 * config/rs6000/spe.md: Delete file.
24608 * config/rs6000/t-rs6000: Remove spe.md.
24610 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24612 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
24613 (reg_or_none500mem_operand): Delete.
24614 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
24615 instead of reg_or_none500mem_operand.
24617 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24619 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24620 handling of SPE flags.
24621 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
24623 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24625 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
24627 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
24628 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
24629 paired_divv2sf3): Similar.
24630 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
24631 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
24632 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
24634 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
24635 Rename the paired_* instruction patterns.
24636 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
24638 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
24639 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
24640 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
24641 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
24642 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
24643 PAIRED_VECTOR_MODE.
24644 (struct machine_function): Delete field spe_insn_chain_scanned_p.
24645 (spe_func_has_64bit_regs_p): Delete.
24646 (spe_expand_predicate_builtin): Delete.
24647 (spe_expand_evsel_builtin): Delete.
24648 (TARGET_DWARF_REGISTER_SPAN): Do not define.
24649 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
24650 (invalid_e500_subreg): Delete.
24651 (rs6000_legitimize_address): Always force_reg op2 as well, for
24652 paired single memory accesses.
24653 (rs6000_member_type_forces_blk): Delete.
24654 (rs6000_spe_function_arg): Delete.
24655 (rs6000_expand_unop_builtin): Delete SPE handling.
24656 (rs6000_expand_binop_builtin): Ditto.
24657 (spe_expand_stv_builtin): Delete.
24658 (bdesc_2arg_spe): Delete.
24659 (spe_expand_builtin): Delete.
24660 (spe_expand_predicate_builtin): Delete.
24661 (spe_expand_evsel_builtin): Delete.
24662 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
24663 (spe_init_builtins): Delete.
24664 (spe_func_has_64bit_regs_p): Delete.
24665 (savres_routine_name): Delete "info" parameter. Adjust callers.
24666 (rs6000_emit_stack_reset): Ditto.
24667 (rs6000_dwarf_register_span): Delete.
24668 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
24669 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
24670 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
24672 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
24674 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
24675 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
24676 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
24677 mulv2sf3, divv2sf3): Delete expanders.
24679 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24681 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
24683 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24685 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
24686 * config/rs6000/rs6000.c: Ditto.
24688 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24690 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
24691 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
24692 comparison_operator.
24694 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24696 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
24697 * config/rs6000/rs6000.opt: Ditto.
24698 * config/rs6000/t-rtems: Ditto.
24700 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24702 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
24703 TARGET_E500_SINGLE by 0, simplify.
24704 * config/rs6000/rs6000.c: Ditto.
24705 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
24706 (spe_build_register_parallel): Delete.
24707 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
24708 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
24709 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
24710 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
24711 (E500_CONVERT): Delete.
24712 * config/rs6000/spe.md: Remove many patterns and all define_constants.
24714 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
24716 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
24717 * config/rs6000/dfp.md: Ditto.
24718 (negdd2, *negdd2_fpr): Merge.
24719 (absdd2, *absdd2_fpr): Merge.
24720 (negtd2, *negtd2_fpr): Merge.
24721 (abstd2, *abstd2_fpr): Merge.
24722 * config/rs6000/e500.h: Delete file.
24723 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
24724 TARGET_FPRS by 1 and simplify.
24725 * config/rs6000/rs6000-c.c: Ditto.
24726 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
24727 TARGET_DF_SPE by 0.
24728 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
24730 * config/rs6000/rs6000.md: Ditto.
24731 (floatdidf2, *floatdidf2_fpr): Merge.
24732 (move_from_CR_gt_bit): Delete.
24733 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
24734 (E500_CR_IOR_COMPARE): Delete.
24735 (All patterns that require !TARGET_FPRS): Delete.
24736 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
24738 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24740 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
24742 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24744 * graphds.c (add_edge): Intitialize edge's attached data.
24745 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
24746 pointer parameter. Call pointed function on each edge during
24747 graph traversing. Skip traversing the edge when the function
24749 (graphds_dfs, graphds_scc): Ditto.
24750 (for_each_edge): New parameter. Pass the new parameter to callback
24752 * graphds.h (skip_edge_callback): New function pointer type.
24753 (graphds_dfs, graphds_scc): New function pointer parameter.
24754 (graphds_edge_callback, for_each_edge): New parameter.
24756 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24758 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
24759 out code checking if runtime alias check is possible to below ...
24760 Call the new function.
24761 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
24762 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
24764 2017-06-07 Marek Polacek <polacek@redhat.com>
24767 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
24768 TYPE_OVERFLOW_WRAPS check.
24770 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24772 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
24773 if versioning is required.
24774 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
24775 peeling with the check for versioning.
24777 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24779 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
24780 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
24781 Set true to new parameter if new ssa variable is defined.
24782 (vect_gen_vector_loop_niters): Refactor. Set range information
24783 for the new vector loop bound variable.
24784 (vect_do_peeling): Ditto.
24786 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24788 * tree-affine.c (ssa.h): Include header file.
24789 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
24790 has wrapping overflow behavior.
24792 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24794 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
24796 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24798 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
24799 (tree_to_aff_combination): ... here.
24801 2017-06-07 Bin Cheng <bin.cheng@arm.com>
24803 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
24804 reg_pressure model function.
24805 (ivopts_global_cost_for_size): Delete.
24806 (determine_set_costs, iv_ca_recount_cost): Call new model function
24807 ivopts_estimate_reg_pressure.
24809 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24811 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
24812 expensive than udiv. Remove floating point cases from mod.
24814 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24816 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
24817 Increase idiv cost.
24819 2017-06-07 Tamar Christina <tamar.christina@arm.com>
24821 * config/aarch64/aarch64.md
24822 (copysignsf3): Fix mask generation.
24824 2017-06-07 Jakub Jelinek <jakub@redhat.com>
24826 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
24828 (class dump_manager): Add register_dumps method.
24829 * dumpfile.c: Include langhooks.h.
24830 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
24831 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
24832 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
24833 (dump_manager::dump_register): Start with 512 entries instead of 32.
24834 (dump_manager::register_dumps): New method.
24835 * toplev.c (general_init): Instead of invoking register_dumps
24836 langhook, invoke register_dumps method on the dump manager.
24837 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
24840 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
24842 * doc/md.texi: Clarify the restrictions on a define_insn condition.
24843 Say that # requires an associated define_split to exist, and that
24844 the define_split must be suitable for use after register allocation.
24846 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24848 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
24849 (compute_outgoing_frequencies): Also initialize zero counts.
24850 (find_many_sub_basic_blocks): Do not produce uninitialized profile
24851 around loops; preserve more of profile when nothing changes.
24853 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
24855 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
24857 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
24858 * config/arm/arm-cpu-cdata.h: Regenerate.
24859 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
24860 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
24861 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
24862 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
24863 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
24865 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
24867 * config/arm/t-rmprofile: Likewise.
24868 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
24870 2017-06-06 David S. Miller <davem@davemloft.net>
24873 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
24874 blockage if function uses alloca.
24876 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24878 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
24879 New "uid" fields to hold pretty-print IDs of group and ref.
24880 Memory references are now identified as <group_id>:<ref_id>
24881 instead of using [random] addresses.
24882 (dump_mem_details): Simplify, no functional change.
24883 (dump_mem_ref): Simplify and make output more concise.
24884 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
24885 (find_or_create_group): Initialize group uid.
24886 (record_ref): Initialize ref uid. Improve debug output.
24887 (prune_group_by_reuse, should_issue_prefetch_p,)
24888 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
24889 (mark_nontemporal_store, determine_loop_nest_reuse):
24890 Improve debug output.
24892 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
24894 * dbgcnt.def (prefetch): New debug counter.
24895 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
24896 (schedule_prefetches): Stop issueing prefetches if debug counter
24899 2017-06-06 Tom de Vries <tom@codesourcery.com>
24901 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
24902 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
24904 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24906 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
24907 Use aarch64_reg_or_zero predicate for operand 4.
24908 (aarch64_compare_and_swap<mode> define_insn_and_split):
24909 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
24910 (aarch64_store_exclusive<mode>): Likewise for operand 2.
24912 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24914 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
24915 (arm_compute_save_core_reg_mask): This.
24916 (thumb1_compute_save_reg_mask): Rename into ...
24917 (thumb1_compute_save_core_reg_mask): This.
24918 (arm_compute_save_reg0_reg12_mask): Adapt comment.
24919 (arm_compute_frame_layout): Likewise.
24921 2017-06-06 Richard Biener <rguenther@suse.de>
24923 PR tree-optimization/80974
24924 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
24925 keep or clear leaders SSA info.
24927 2017-06-06 Tom de Vries <tom@codesourcery.com>
24929 * config/nvptx/nvptx.c (split_mode_p): New function.
24930 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
24932 2017-06-06 Tom de Vries <tom@codesourcery.com>
24934 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
24936 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24939 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
24942 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24944 * shrink-wrap.c (handle_simple_exit): Update profile.
24945 (try_shrink_wrapping): Upate profile.
24947 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24949 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
24950 (tree_guess_outgoing_edge_probabilities): New.
24951 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
24952 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
24954 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24956 * ipa-split.c (split_function): Initialize return bb profile.
24958 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
24960 * profile.c (compute_branch_probabilities): Also initialize
24961 EXIT_BLOCK profile.
24963 2017-06-06 Richard Biener <rguenther@suse.de>
24965 PR tree-optimization/80928
24966 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
24967 (vect_analyze_loop_operations): Properly guard analysis for
24969 (vect_transform_loop): Likewise.
24970 (vect_analyze_loop_2): Also reset SLP type on PHIs.
24971 (vect_model_induction_cost): Do not cost for pure SLP.
24972 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
24973 of induction in inner loop vectorization.
24974 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
24975 (vect_get_and_check_slp_defs): Handle vect_induction_def.
24976 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
24978 (vect_analyze_slp_cost_1): Cost induction.
24979 (vect_detect_hybrid_slp_stmts): Handle PHIs.
24980 (vect_get_slp_vect_defs): Likewise.
24981 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
24982 (vect_transform_stmt): Handle SLP reductions.
24983 * tree-vectorizer.h (vectorizable_induction): Adjust.
24985 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24987 * config/rs6000/rs6000.c (make_resolver_func): Update
24988 init_lowered_empty_function call.
24990 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
24992 * doc/invoke.texi: Document the -fprofile-abs-path option.
24993 * common.opt (fprofile-abs-path): New option.
24994 * gcov-io.h (gcov_write_filename): Declare.
24995 * gcov-io.c (gcov_write_filename): New function.
24996 * coverage.c (coverage_begin_function): Use gcov_write_filename.
24997 * profile.c (output_location): Likewise.
24999 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25001 * shring-wrap.c: Revert accidental commit.
25003 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
25005 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
25007 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25009 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
25011 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
25012 profile in callgraph edge.
25013 * profile-count.h (apply_probability): If THIS is 0, then result is 0
25014 (apply_scale): Likewise.
25015 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
25016 Also scale profile when inlining function with zero profile.
25017 (initialize_cfun): Update exit block profile even when it is zero.
25018 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
25019 when profile is read.
25021 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25023 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
25024 (CLONE_*): New constants to define the processors we can generate
25025 code for with the target_clone attribute.
25026 (rs6000_clone_map): New array to identify which clone processors
25027 the current program is running on.
25028 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
25029 target_clone attribute.
25030 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
25031 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
25032 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
25033 (cpu_expand_builtin): Add support for target_clone attribute.
25034 (rs6000_valid_attribute_p): Allow "default" attribute.
25035 (get_decl_name): New debug function to simplify printing the
25036 current function name in debugging statements.
25037 (rs6000_clone_priority): New functions to support the target_clone
25038 attribute, and be able to generate code to switch between ISA 2.05
25039 through ISA 3.0 (power6 through power9).
25040 (rs6000_compare_version_priority): Likewise.
25041 (rs6000_get_function_versions_dispatcher): Likewise.
25042 (make_resolver_func): Likewise.
25043 (add_condition_to_bb): Likewise.
25044 (dispatch_function_versions): Likewise.
25045 (rs6000_generate_version_dispatcher_body): Likewise.
25046 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
25047 (fusion_gpr_load_p): Fix a spacing issue.
25048 * doc/extend.texi (Common Function Attributes): Document that the
25049 PowerPC supports the target_clone attribute.
25051 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25053 * config/arm/arm.h: explain F symbol found in description of ARM
25054 register allocation in its legend.
25056 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25058 * config/mips/frame-header-opt.c: Include profile-count.h.
25059 * config/riscv/riscv.c: Include profile-count.h
25061 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
25063 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
25065 (sm_set_flag_if_changed): Add bbs field.
25066 (execute_sm_if_changed_flag_set): Pass BBS.
25067 (execute_sm): Update.
25069 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25071 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
25074 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25076 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
25077 (peephole2): New peephole2 to emit the above.
25078 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
25080 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25082 * config/aarch64/aarch64.c (define_peephole2 above
25083 *sub_<shift>_<mode>): New peephole.
25085 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
25087 * config/i386/i386.c (make_resolver_func): Update.
25088 * Makefile.in: Add profile-count.h and profile-count.o
25089 * auto-profile.c (afdo_indirect_call): Update to new API.
25090 (afdo_set_bb_count): Update.
25091 (afdo_propagate_edge): Update.
25092 (afdo_propagate_circuit): Update.
25093 (afdo_calculate_branch_prob): Update.
25094 (afdo_annotate_cfg): Update.
25095 * basic-block.h: Include profile-count.h
25096 (struct edge_def): Turn count to profile_count.
25097 (struct basic_block_def): Likewie.
25098 (REG_BR_PROB_BASE): Move to profile-count.h
25099 (RDIV): Move to profile-count.h
25100 * bb-reorder.c (max_entry_count): Turn to profile_count.
25101 (find_traces): Update.
25102 (rotate_loop):Update.
25103 (connect_traces):Update.
25104 (sanitize_hot_paths):Update.
25105 * bt-load.c (migrate_btr_defs): Update.
25106 * cfg.c (RDIV): Remove.
25107 (init_flow): Use alloc_block.
25108 (alloc_block): Uninitialize count.
25109 (unchecked_make_edge): Uninitialize count.
25110 (check_bb_profile): Update.
25111 (dump_edge_info): Update.
25112 (dump_bb_info): Update.
25113 (update_bb_profile_for_threading): Update.
25114 (scale_bbs_frequencies_int): Update.
25115 (scale_bbs_frequencies_gcov_type): Update.
25116 (scale_bbs_frequencies_profile_count): New.
25117 * cfg.h (update_bb_profile_for_threading): Update.
25118 (scale_bbs_frequencies_profile_count): Declare.
25119 * cfgbuild.c (compute_outgoing_frequencies): Update.
25120 (find_many_sub_basic_blocks): Update.
25121 * cfgcleanup.c (try_forward_edges): Update.
25122 (try_crossjump_to_edge): Update.
25123 * cfgexpand.c (expand_gimple_tailcall): Update.
25124 (construct_exit_block): Update.
25125 * cfghooks.c (verify_flow_info): Update.
25126 (dump_bb_for_graph): Update.
25127 (split_edge): Update.
25128 (make_forwarder_block): Update.
25129 (duplicate_block): Update.
25130 (account_profile_record): Update.
25131 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
25132 (get_estimated_loop_iterations): Update.
25133 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
25134 (single_likely_exit): Update.
25135 * cfgloopmanip.c (scale_loop_profile): Update.
25137 (set_zero_probability): Update.
25138 (lv_adjust_loop_entry_edge): Update.
25139 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
25140 (purge_dead_edges): Update.
25141 (rtl_account_profile_record): Update.
25142 * cgraph.c (cgraph_node::create): Uninitialize count.
25143 (symbol_table::create_edge): Uninitialize count.
25144 (cgraph_update_edges_for_call_stmt_node): Update.
25145 (cgraph_edge::dump_edge_flags): Update.
25146 (cgraph_node::dump): Update.
25147 (cgraph_edge::maybe_hot_p): Update.
25148 * cgraph.h: Include profile-count.h
25149 (create_clone), create_edge, create_indirect_edge): Update.
25150 (cgraph_node): Turn count to profile_count.
25151 (cgraph_edge0: Likewise.
25152 (make_speculative, clone): Update.
25153 (create_edge): Update.
25154 (init_lowered_empty_function): Update.
25155 * cgraphclones.c (cgraph_edge::clone): Update.
25156 (duplicate_thunk_for_node): Update.
25157 (cgraph_node::create_clone): Update.
25158 * cgraphunit.c (cgraph_node::analyze): Update.
25159 (cgraph_node::expand_thunk): Update.
25160 * final.c (dump_basic_block_info): Update.
25161 * gimple-streamer-in.c (input_bb): Update.
25162 * gimple-streamer-out.c (output_bb): Update.
25163 * graphite.c (print_global_statistics): Update.
25164 (print_graphite_scop_statistics): Update.
25165 * hsa-brig.c: Include basic-block.h.
25166 * hsa-dump.c: Include basic-block.h.
25167 * hsa-gen.c (T sum_slice): Update.
25168 (convert_switch_statements):Update.
25169 * hsa-regalloc.c: Include basic-block.h.
25170 * ipa-chkp.c (chkp_produce_thunks): Update.
25171 * ipa-cp.c (struct caller_statistics): Update.
25172 (init_caller_stats): Update.
25173 (gather_caller_stats): Update.
25174 (ipcp_cloning_candidate_p): Update.
25175 (good_cloning_opportunity_p): Update.
25176 (get_info_about_necessary_edges): Update.
25177 (dump_profile_updates): Update.
25178 (update_profiling_info): Update.
25179 (update_specialized_profile): Update.
25180 (perhaps_add_new_callers): Update.
25181 (decide_about_value): Update.
25182 (ipa_cp_c_finalize): Update.
25183 * ipa-devirt.c (struct odr_type_warn_count): Update.
25184 (struct decl_warn_count): Update.
25185 (struct final_warning_record): Update.
25186 (possible_polymorphic_call_targets): Update.
25187 (ipa_devirt): Update.
25188 * ipa-fnsummary.c (redirect_to_unreachable): Update.
25189 * ipa-icf.c (sem_function::merge): Update.
25190 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
25191 * ipa-inline.c (compute_uninlined_call_time): Update.
25192 (compute_inlined_call_time): Update.
25193 (want_inline_small_function_p): Update.
25194 (want_inline_self_recursive_call_p): Update.
25195 (edge_badness): Update.
25196 (lookup_recursive_calls): Update.
25197 (recursive_inlining): Update.
25198 (inline_small_functions): Update.
25199 (dump_overall_stats): Update.
25200 (dump_inline_stats): Update.
25201 * ipa-profile.c (ipa_profile_generate_summary): Update.
25202 (ipa_propagate_frequency): Update.
25203 (ipa_profile): Update.
25204 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
25205 * ipa-utils.c (ipa_merge_profiles): Update.
25206 * loop-doloop.c (doloop_modify): Update.
25207 * loop-unroll.c (report_unroll): Update.
25208 (unroll_loop_runtime_iterations): Update.
25209 * lto-cgraph.c (lto_output_edge): Update.
25210 (lto_output_node): Update.
25211 (input_node): Update.
25212 (input_edge): Update.
25213 (merge_profile_summaries): Update.
25214 * lto-streamer-in.c (input_cfg): Update.
25215 * lto-streamer-out.c (output_cfg): Update.
25216 * mcf.c (create_fixup_graph): Update.
25217 (adjust_cfg_counts): Update.
25218 (sum_edge_counts): Update.
25219 * modulo-sched.c (sms_schedule): Update.
25220 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
25221 * predict.c (maybe_hot_count_p): Update.
25222 (probably_never_executed): Update.
25223 (dump_prediction): Update.
25224 (combine_predictions_for_bb): Update.
25225 (propagate_freq): Update.
25226 (handle_missing_profiles): Update.
25227 (counts_to_freqs): Update.
25228 (rebuild_frequencies): Update.
25229 (force_edge_cold): Update.
25230 * predict.h: Include profile-count.h
25231 (maybe_hot_count_p, counts_to_freqs): UPdate.
25232 * print-rtl-function.c: Do not include cfg.h
25233 * print-rtl.c: Include basic-block.h
25234 * profile-count.c: New file.
25235 * profile-count.h: New file.
25236 * profile.c (is_edge_inconsistent): Update.
25237 (correct_negative_edge_counts): Update.
25238 (is_inconsistent): Update.
25239 (set_bb_counts): Update.
25240 (read_profile_edge_counts): Update.
25241 (compute_frequency_overlap): Update.
25242 (compute_branch_probabilities): Update; Initialize and deinitialize
25244 (branch_prob): Update.
25245 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
25246 (edge_gcov_count): New.
25247 (bb_gcov_count): New.
25248 * shrink-wrap.c (try_shrink_wrapping): Update.
25249 * tracer.c (better_p): Update.
25250 * trans-mem.c (expand_transaction): Update.
25251 (ipa_tm_insert_irr_call): Update.
25252 (ipa_tm_insert_gettmclone_call): Update.
25253 * tree-call-cdce.c: Update.
25254 * tree-cfg.c (gimple_duplicate_sese_region): Update.
25255 (gimple_duplicate_sese_tail): Update.
25256 (gimple_account_profile_record): Update.
25257 (execute_fixup_cfg): Update.
25258 * tree-inline.c (copy_bb): Update.
25259 (copy_edges_for_bb): Update.
25260 (initialize_cfun): Update.
25261 (freqs_to_counts): Update.
25262 (copy_cfg_body): Update.
25263 (expand_call_inline): Update.
25264 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
25265 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
25266 (try_unroll_loop_completely): Update.
25267 (try_peel_loop): Update.
25268 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25269 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
25270 * tree-ssa-loop-split.c (connect_loops): Update.
25271 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
25272 * tree-ssa-reassoc.c (branch_fixup): Update.
25273 * tree-ssa-tail-merge.c (replace_block_by): Update.
25274 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
25275 (compute_path_counts): Update.
25276 (update_profile): Update.
25277 (recompute_probabilities): Update.
25278 (update_joiner_offpath_counts): Update.
25279 (estimated_freqs_path): Update.
25280 (freqs_to_counts_path): Update.
25281 (clear_counts_path): Update.
25282 (ssa_fix_duplicate_block_edges): Update.
25283 (duplicate_thread_path): Update.
25284 * tree-switch-conversion.c (case_bit_test_cmp): Update.
25285 (struct switch_conv_info): Update.
25286 * tree-tailcall.c (decrease_profile): Update.
25287 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
25288 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25289 * value-prof.c (check_counter): Update.
25290 (gimple_divmod_fixed_value): Update.
25291 (gimple_mod_pow2): Update.
25292 (gimple_mod_subtract): Update.
25293 (gimple_ic_transform): Update.
25294 (gimple_stringop_fixed_value): Update.
25295 * value-prof.h (gimple_ic): Update.
25297 2017-06-02 Carl Love <cel@us.ibm.com>
25299 * config/rs6000/rs6000-c: Add support for built-in functions
25300 vector double vec_doublee (vector signed int);
25301 vector double vec_doublee (vector unsigned int);
25302 vector double vec_doublee (vector float);
25303 vector double vec_doubleh (vector signed int);
25304 vector double vec_doubleh (vector unsigned int);
25305 vector double vec_doubleh (vector float);
25306 vector double vec_doublel (vector signed int);
25307 vector double vec_doublel (vector unsigned int);
25308 vector double vec_doublel (vector float);
25309 vector double vec_doubleo (vector signed int);
25310 vector double vec_doubleo (vector unsigned int);
25311 vector double vec_doubleo (vector float);.
25312 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
25313 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
25315 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
25316 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
25317 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
25319 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
25320 vec_doublel, vec_doubleh.
25321 * doc/extend.texi: Update the built-in documentation file for the
25322 new built-in functions.
25324 2017-06-02 David Malcolm <dmalcolm@redhat.com>
25327 * ipa-inline-analysis.c (free_growth_caches): Set
25328 edge_removal_hook_holder to NULL after removing it.
25330 2017-06-02 Sudakshina Das <sudi.das@arm.com>
25332 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
25333 comparision with zero.
25335 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25336 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25337 for early expansion of vec_min and vec_max builtins.
25338 (builtin_function_type): Add min/max unsigned variants to those
25339 identified as having unsigned arguments.
25341 2017-06-02 Olivier Hainque <hainque@adacore.com>
25343 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
25345 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25347 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
25348 Use VALL_F16 iterator rather than VALL.
25350 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25352 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
25353 Emit CBNZ inside loop when doing a strong exchange and comparing
25354 against zero. Generate the CC flags after the loop.
25356 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
25358 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
25359 (dl_section_ref): New.
25360 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
25361 On AIX, append an expression to subtract the size of the
25362 section length to dl_section_ref.
25364 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
25366 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
25367 for early expansion of vector absolute builtins.
25369 2017-06-02 Richard Biener <rguenther@suse.de>
25371 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
25372 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
25374 2017-06-02 Richard Biener <rguenther@suse.de>
25376 PR tree-optimization/80948
25377 * tree-tailcall.c (find_tail_calls): Track stmts to move in
25378 stmt order as well.
25380 2017-06-02 Richard Biener <rguenther@suse.de>
25382 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
25384 * tree-vect-stmts.c (process_use): Do not mark backedge defs
25385 for inductions as relevant.
25387 2017-06-02 Richard Biener <rguenther@suse.de>
25389 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
25390 (vectorizable_induction): ... this. Remove dead code.
25392 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
25394 * builtins. (expand_builtin_alloca): Remove second parameter and
25395 infer its value from the first parameter instead.
25396 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
25398 2017-06-02 Jakub Jelinek <jakub@redhat.com>
25400 PR rtl-optimization/80903
25401 * loop-doloop.c (add_test): Unshare sequence.
25403 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
25405 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
25407 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
25409 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
25411 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
25412 xlogue_layout::get_instance, logue_layout::xlogue_layout,
25413 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
25414 (xlogue_layout::get_stub_rtx): Make static.
25415 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
25416 (xlogue_layout::compute_stub_managed_regs): Rename to...
25417 (xlogue_layout::count_stub_managed_regs): ...this.
25418 (xlogue_layout::is_stub_managed_reg): New function.
25419 (xlogue_layout::m_stub_names): Rename to...
25420 (xlogue_layout::s_stub_names): ...this, make static.
25421 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
25422 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
25423 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
25424 xlogue_layout::s_stub_names): Instantiate statics.
25425 (stub_managed_regs): Remove.
25426 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
25427 (disable_call_ms2sysv_xlogues): Rename to...
25428 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
25429 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
25431 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
25432 change after reload_completed.
25433 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
25435 (ix86_expand_prologue): Likewise.
25436 (ix86_expand_epilogue): Likewise.
25437 (ix86_expand_split_stack_prologue): Likewise.
25438 (ix86_compute_frame_layout): Remove frame parameter ...
25439 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
25440 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
25442 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
25443 (ix86_frame): Move from here ...
25444 * config/i386/i386.h (ix86_frame): ... to here.
25445 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
25446 complete ix86_frame data structure instead. Remove some_ld_name.
25448 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25450 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
25451 symbols that hold a DECL_VALUE_EXPR.
25453 2017-06-01 Martin Jambor <mjambor@suse.cz>
25455 PR tree-optimization/80898
25456 * tree-sra.c (process_subtree_disqualification): Removed.
25457 (disqualify_candidate): Do not acll
25458 process_subtree_disqualification.
25459 (subtree_mark_written_and_enqueue): New function.
25460 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
25461 RHS has been disqualified and re-queue LHS if necessary. Apart
25462 from that, ignore disqualified RHS.
25464 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25466 * config/s390/s390.c (s390_emit_epilogue): Disable early return
25467 address fetch for z10 or later.
25469 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25471 * config/arc/arc.md (tst_movb): Add guard when splitting.
25473 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25475 * config/arc/arc.c (arc_can_eliminate): Test against
25476 arc_frame_pointer_needed.
25478 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25480 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
25481 to prevent store reordering.
25482 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
25483 (type): Add block type.
25484 (stack_tie): Define special instruction to be used in
25487 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25489 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
25490 constraint. It is not valid for the pattern.
25491 (noncommutative_binary_comparison): Likewise.
25493 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25495 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
25498 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25500 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
25501 be used by the reg-alloc.
25503 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25505 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
25506 reg-alloc when having mul64 or mul32x16 instructions.
25507 (mulsidi3): Likewise.
25508 (umulsidi3): Likewise.
25509 (mulsi32x16): New pattern.
25510 (mulsi64): Likewise.
25511 (mulsidi64): Likewise.
25512 (umulsidi64): Likewise.
25513 (MUL32x16_REG): Define.
25514 (mul64_600): Use MUL32x16_REG.
25515 (mac64_600): Likewise.
25516 (umul64_600): Likewise.
25517 (umac64_600): Likewise.
25519 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
25521 * config/arc/arc.md (mulsi3_700): Make it commutative.
25523 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
25525 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
25527 (*sign_extendsidi2_insn): Likewise for movstosw.
25529 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
25531 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
25532 the type of the input discriminant value. Convert the
25533 discriminant value of signedness vary.
25535 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
25537 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
25538 Add to -Wall section.
25540 2017-06-01 Richard Biener <rguenther@suse.de>
25542 PR middle-end/66313
25543 * fold-const.c (fold_plusminus_mult_expr): If the factored
25544 factor may be zero use a wrapping type for the inner operation.
25545 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
25546 and handle moved defs.
25547 (process_assignment): Properly guard the unary op case. Return a
25548 tri-state indicating that moving the stmt before the call may allow
25549 to continue. Pass through to_move.
25550 (find_tail_calls): Handle moving unrelated defs before
25553 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
25556 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
25557 splitter result in the canonical way.
25559 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
25561 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
25562 also for 32bit target. Update insn attributes.
25563 (zero-extendsidi2 splitter): Allow all registers for operand 1.
25565 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
25567 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
25568 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
25569 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
25570 (_mm_maskz_min_ss): New intrinsics.
25572 2017-05-31 Martin Liska <mliska@suse.cz>
25574 * tree-vect-loop.c (vect_create_epilog_for_reduction):
25575 Change comment style to one we normally use.
25576 (vectorizable_reduction): Likewise.
25577 (vectorizable_induction): Likewise.
25578 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
25579 (vectorizable_call): Likewise.
25580 (vectorizable_simd_clone_call): Likewise.
25581 (vectorizable_conversion): Likewise.
25582 (vectorizable_assignment): Likewise.
25583 (vectorizable_shift): Likewise.
25584 (vectorizable_operation): Likewise.
25585 (vectorizable_store): Likewise.
25586 (vectorizable_load): Likewise.
25587 * tree-vectorizer.h: Likewise.
25589 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
25591 * passes.c (emergency_dump_function): New.
25592 * tree-pass.h (emergency_dump_function): Declare.
25593 * plugin.c (plugins_internal_error_function): Remove.
25594 * plugin.h (plugins_internal_error_function): Remove declaration.
25595 * toplev.c (internal_error_function): New static function. Use it...
25596 (general_init): ...here.
25598 2017-05-31 Graham Markall <graham.markall@embecosm.com>
25600 * config/arc/arc.c (arc_print_operand): Handle constant operands.
25601 (arc_rtx_costs): Add costs for new patterns.
25602 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
25603 * config/arc/predicates.md: Add _1_2_3_operand predicate.
25605 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
25607 * tree-ssa-strlen.c (get_next_strinfo): New function.
25608 (get_stridx_plus_constant): Use it.
25609 (zero_length_string): Likewise.
25610 (adjust_related_strinfos): Likewise.
25611 (adjust_last_stmt): Likewise.
25613 2017-05-31 Richard Biener <rguenther@suse.de>
25616 * config/i386/i386.c (ix86_expand_builtin): Remove assert
25617 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
25619 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
25621 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
25622 loop_vinfo argument and use of dependence distance vectors.
25623 Check instead whether the two references differ only in their
25624 initial value and assume that they have the same alignment if the
25625 difference is a multiple of the vector alignment.
25626 (vect_analyze_data_refs_alignment): Update call accordingly.
25628 2017-05-31 Martin Liska <mliska@suse.cz>
25631 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
25633 2017-05-31 Bin Cheng <bin.cheng@arm.com>
25635 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
25636 (create_intersect_range_checks): Move from ...
25637 * tree-data-ref.c (create_intersect_range_checks_index)
25638 (create_intersect_range_checks): ... to here.
25639 (create_runtime_alias_checks): New function factored from ...
25640 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
25641 here. Call above function.
25642 * tree-data-ref.h (create_runtime_alias_checks): New function.
25644 2017-05-31 Bin Cheng <bin.cheng@arm.com>
25646 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
25647 segment length for dr_b and compute it in wide_int.
25649 2017-05-31 Richard Biener <rguenther@suse.de>
25651 PR tree-optimization/80906
25652 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
25653 and pass through iv_map.
25654 (copy_bb_and_scalar_dependences): Adjust.
25655 (translate_pending_phi_nodes): Likewise.
25656 (copy_loop_close_phi_args): Handle code-generating IVs instead
25659 2017-05-30 David Malcolm <dmalcolm@redhat.com>
25661 * diagnostic-color.c (color_dict): Add "type-diff".
25662 (parse_gcc_colors): Update comment.
25663 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
25664 -fdiagnostics-show-template-tree and -fno-elide-type.
25665 (GCC_COLORS): Add type-diff to example.
25667 (-fdiagnostics-show-template-tree): New.
25668 (-fno-elide-type): New.
25669 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
25670 the pp_format_decoder callback. Call any m_format_postprocessor's
25672 (pretty_printer::pretty_printer): Initialize
25673 m_format_postprocessor.
25674 (pretty_printer::~pretty_printer): Delete any
25675 m_format_postprocessor.
25676 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
25677 (class format_postprocessor): New class.
25678 (struct pretty_printer::format_decoder): Document the new parameters.
25679 (struct pretty_printer::m_format_postprocessor): New field.
25680 * tree-diagnostic.c (default_tree_printer): Update for new
25681 bool and const char ** params.
25682 * tree-diagnostic.h (default_tree_printer): Likewise.
25684 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
25686 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
25687 (lwa_operand): Delete rs6000_gen_cell_microcode test.
25688 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
25689 rs6000_gen_cell_microcode code.
25690 (rs6000_final_prescan_insn): Delete.
25691 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
25692 "warn-cell-microcode" entries.
25693 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
25694 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
25695 throughout. Change cc_reg_not_micro_cr0_operand to
25696 cc_reg_not_cr0_operand throughout.
25697 (*extendhi<mode>2_noload): Delete.
25698 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
25699 (mwarn-cell-microcode): Delete.
25700 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
25701 -mgen-cell-microcode and -mwarn-cell-microcode.
25703 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
25706 * config/i386/constraints.md (Yd): New constraint.
25708 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
25709 and (?Yd, r) alternatives. Update insn attributes.
25710 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
25711 and (?*Yd, r) alternatives. Update insn attributes.
25712 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
25714 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
25716 * gimplify.c (gimplify_modify_expr): Don't create a
25717 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
25720 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
25722 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
25724 2017-05-30 Richard Biener <rguenther@suse.de>
25726 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
25727 and reduc_def fields.
25728 (STMT_VINFO_REDUC_TYPE): New define.
25729 (STMT_VINFO_REDUC_DEF): Likewise.
25730 (vect_force_simple_reduction): Adjust prototype.
25731 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
25732 (vect_is_simple_reduction): Remove check_reduction argument.
25733 (vect_force_simple_reduction): Adjust and set
25734 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25735 (vectorizable_reduction): Do not re-do reduction analysis
25736 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
25737 * tree-parloops.c (gather_scalar_reductions): Adjust.
25739 2017-05-30 Richard Biener <rguenther@suse.de>
25741 PR middle-end/80901
25742 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
25745 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
25747 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
25748 Introduce unknown_misalignment parameter and remove vf.
25749 (vect_peeling_hash_get_lowest_cost):
25750 Pass unknown_misalignment parameter.
25751 (vect_enhance_data_refs_alignment):
25752 Fix unsupportable data ref treatment.
25754 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25756 * tree-vect-data-refs.c (vect_get_data_access_cost):
25757 Workaround for SLP handling.
25758 (vect_enhance_data_refs_alignment):
25759 Compute costs for doing no peeling at all, compare to the best
25760 peeling costs so far and avoid peeling if cheaper.
25762 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25764 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
25765 Return peeling info and set costs to zero for unlimited cost
25767 (vect_enhance_data_refs_alignment): Also inspect all datarefs
25768 with unknown misalignment. Compute and costs for unknown
25769 misalignment, compare them to the costs for known misalignment
25770 and choose the cheapest for peeling.
25772 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25774 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
25775 (vect_get_peeling_costs_all_drs): Create function.
25776 (vect_peeling_hash_get_lowest_cost):
25777 Use vect_get_peeling_costs_all_drs.
25778 (vect_peeling_supportable): Create function.
25779 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
25781 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
25783 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
25784 DR_HAS_NEGATIVE_STEP.
25785 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
25786 (vect_enhance_data_refs_alignment): Use.
25787 (vect_duplicate_ssa_name_ptr_info): Use.
25788 * tree-vectorizer.h (dr_misalignment): Use.
25789 (known_alignment_for_access_p): Use.
25791 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25794 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
25796 (has_section_name): New function.
25798 2017-05-30 Martin Liska <mliska@suse.cz>
25801 * auto-profile.c (get_function_decl_from_block): Fix
25804 2017-05-30 Richard Biener <rguenther@suse.de>
25806 PR middle-end/80876
25807 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
25809 2017-05-30 Martin Liska <mliska@suse.cz>
25811 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
25812 * dumpfile.h (struct dump_file_info): Remove ctors.
25814 2017-05-30 Martin Liska <mliska@suse.cz>
25816 * predict.def: Fix GNU coding style.
25818 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25820 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25821 Mark 'to' argument with ATTRIBUTE_UNUSED.
25823 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
25825 * config/xtensa/xtensa.c (xtensa_emit_call): Use
25826 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
25827 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
25830 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
25832 * doc/install.texi (Options specification): Restore entry of
25833 --enable-sjlj-exceptions.
25835 2017-05-27 Michael Eager <eager@eagercon.com>
25838 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
25840 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
25842 * config/microblaze/microblaze.h
25843 (FIXED_REGISTERS): Update in macro.
25844 (CALL_USED_REGISTERS): Update in macro.
25846 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
25848 * doc/install.texi: Add links to macOS binary distributions.
25850 2017-05-27 Jakub Jelinek <jakub@redhat.com>
25854 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25856 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25858 2017-05-26 Martin Liska <mliska@suse.cz>
25860 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
25862 2017-05-26 Martin Liska <mliska@suse.cz>
25864 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
25865 always leading ';; '.
25866 (dump_bb_info): Likewise.
25867 (brief_dump_cfg): Likewise.
25868 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
25869 * dumpfile.c: Remove usage of TDF_VERBOSE.
25870 * dumpfile.h (enum dump_kind): Likewise.
25871 (dump_gimple_bb_header): Do not use TDF_COMMENT.
25872 * print-tree.c (debug_verbose): Remove.
25873 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
25874 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
25875 * tree-diagnostic.c (default_tree_printer): Replace
25876 TDF_DIAGNOSTIC with TDF_SLIM.
25878 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25880 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
25881 in parameter loop, rather than loop_vinfo.
25882 (create_intersect_range_checks): Ditto.
25883 (vect_create_cond_for_alias_checks): Update call to above functions.
25885 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25887 PR tree-optimization/80815
25888 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
25889 for merging runtime alias checks. Handle negative DR_STEPs.
25891 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25893 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
25895 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
25896 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
25897 out code pruning runtime alias checks.
25898 * tree-data-ref.c (prune_runtime_alias_test_list): New function
25899 factored out from above.
25900 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25902 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
25904 (prune_runtime_alias_test_list): New decalaration.
25906 2017-05-26 Bin Cheng <bin.cheng@arm.com>
25908 * tree-vect-data-refs.c (compare_tree): Rename and move ...
25909 * tree-data-ref.c (data_ref_compare_tree): ... to here.
25910 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
25911 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
25912 (operator==, comp_dr_with_seg_len_pair): Ditto.
25913 (vect_prune_runtime_alias_test_list): Ditto.
25915 2017-05-26 Martin Liska <mliska@suse.cz>
25918 * params.def: Bound partial-inlining-entry-probability param.
25920 2017-05-26 Marek Polacek <polacek@redhat.com>
25923 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
25926 2017-05-26 Richard Biener <rguenther@suse.de>
25928 PR tree-optimization/80842
25929 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
25932 2017-05-26 Richard Biener <rguenther@suse.de>
25934 PR tree-optimization/80844
25935 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
25937 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
25939 * doc/md.texi (Machine Constraints): Update x86 family
25940 machine constraints section to match 'config/i386/constraints.md'.
25942 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
25944 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
25946 2017-05-25 Nathan Sidwell <nathan@acm.org>
25948 * doc/invoke.texi (--enable-languages): Update documentation.
25950 2017-05-25 Martin Liska <mliska@suse.cz>
25952 * dumpfile.c: Add TDF_FOLDING.
25953 * dumpfile.h (enum dump_kind): Likewise.
25954 * genmatch.c (dt_simplify::gen_1): Use it.
25956 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25958 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
25960 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25962 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
25963 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
25965 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
25967 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
25968 * match.pd (X == C): Rewrite it here.
25969 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
25970 with_certain_nonzero_bits2): New predicates.
25971 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
25973 2017-05-24 Nathan Sidwell <nathan@acm.org>
25975 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
25978 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
25981 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
25983 * config/powerpcspe: New port. Files are copied from the rs6000
25984 port, with "rs6000" in filenames replaced by "powerpcspe".
25986 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
25988 PR rtl-optimization/80754
25989 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
25991 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
25993 * config/sparc/sparc.md (length): Return the correct value for -mflat
25994 sibcalls to match output_sibcall.
25996 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
26000 * config/rs6000/rs6000.c (struct machine_function): Add new field
26002 (rs6000_get_separate_components): Init that field, use it.
26003 (rs6000_components_for_bb): Use the field.
26005 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
26007 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
26009 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
26011 PR middle-end/80823
26012 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
26014 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26017 * config/s390/s390.c (s390_check_qrst_address): Check incoming
26018 address against address_operand predicate.
26019 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
26021 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
26023 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
26024 parameters passed indirectly.
26026 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
26028 * config/i386/i386.md (*movdi_internal): Remove SSE4
26029 alternative 18 (?r, *v). Update insn attributes.
26030 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
26031 Update insn attributes.
26032 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
26033 Update insn attributes.
26034 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
26035 alternative 1 (r, v). Remove isa attribute.
26036 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
26037 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
26038 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
26040 2017-05-23 Tom de Vries <tom@codesourcery.com>
26042 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
26045 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26047 * cgraphunit.c (symbol_table::process_new_functions): Update.
26048 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
26049 (inline_generate_summary): Rename to ...
26050 (ipa_fn_summary_generate): ... this one.
26051 (inline_read_summary): Rename to ...
26052 (ipa_fn_summary_read): ... this one.
26053 (inline_write_summary): Rename to ...
26054 (ipa_fn_summary_write): ... this one.
26055 (inline_free_summary): Rename to ...
26056 (ipa_free_fn_summary): ... this one.
26057 (pass_data_local_fn_summary, pass_local_fn_summary,
26058 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
26059 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
26060 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
26061 make_pass_ipa_fn_summary): New.
26062 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
26063 inline_write_summary, inline_free_summary): Remove.
26064 (ipa_free_fn_summary) : New.
26065 * ipa-inline.c (ipa_inline): Update.
26066 (pass_ipa_inline): Do not generate summaries.
26067 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
26069 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
26070 and add pass_ipa_fn_summary.
26071 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
26073 (make_pass_inline_parameters): Remove.
26075 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
26077 * omp-low.c (struct omp_context): Remove "default_kind" member.
26080 * omp-offload.c (execute_oacc_device_lower): Remove the
26081 parallelism dimensions function attributes for unparallelized
26082 OpenACC kernels constructs.
26084 2017-05-23 Martin Liska <mliska@suse.cz>
26086 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
26088 (cgraph_edge::make_speculative): Likewise.
26089 (cgraph_edge::resolve_speculation): Likewise.
26090 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26091 (cgraph_node::dump): Likewise.
26092 * cgraph.h: Likewise.
26093 * cgraphunit.c (analyze_functions): Likewise.
26094 (symbol_table::compile): Likewise.
26095 * ipa-cp.c (print_all_lattices): Likewise.
26096 (determine_versionability): Likewise.
26097 (initialize_node_lattices): Likewise.
26098 (ipcp_verify_propagated_values): Likewise.
26099 (estimate_local_effects): Likewise.
26100 (update_profiling_info): Likewise.
26101 (create_specialized_node): Likewise.
26102 (perhaps_add_new_callers): Likewise.
26103 (decide_about_value): Likewise.
26104 (decide_whether_version_node): Likewise.
26105 (identify_dead_nodes): Likewise.
26106 (ipcp_store_bits_results): Likewise.
26107 * ipa-devirt.c (dump_targets): Likewise.
26108 (ipa_devirt): Likewise.
26109 * ipa-icf.c (sem_item::dump): Likewise.
26110 (sem_function::equals): Likewise.
26111 (sem_variable::equals): Likewise.
26112 (sem_item_optimizer::read_section): Likewise.
26113 (sem_item_optimizer::execute): Likewise.
26114 (congruence_class::dump): Likewise.
26115 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
26116 (dump_inline_summary): Likewise.
26117 (estimate_node_size_and_time): Likewise.
26118 (inline_analyze_function): Likewise.
26119 * ipa-inline-transform.c (inline_call): Likewise.
26120 * ipa-inline.c (report_inline_failed_reason): Likewise.
26121 (want_early_inline_function_p): Likewise.
26122 (edge_badness): Likewise.
26123 (update_edge_key): Likewise.
26124 (inline_small_functions): Likewise.
26125 * ipa-profile.c (ipa_profile): Likewise.
26126 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
26127 (ipa_make_edge_direct_to_target): Likewise.
26128 (remove_described_reference): Likewise.
26129 (ipa_impossible_devirt_target): Likewise.
26130 (propagate_controlled_uses): Likewise.
26131 (ipa_print_node_params): Likewise.
26132 (ipcp_transform_function): Likewise.
26133 * ipa-pure-const.c (pure_const_read_summary): Likewise.
26134 (propagate_pure_const): Likewise.
26135 * ipa-reference.c (generate_summary): Likewise.
26136 (read_write_all_from_decl): Likewise.
26137 (propagate): Likewise.
26138 (ipa_reference_read_optimization_summary): Likewise.
26139 * ipa-utils.c (ipa_merge_profiles): Likewise.
26140 * ipa.c (walk_polymorphic_call_targets): Likewise.
26141 (symbol_table::remove_unreachable_nodes): Likewise.
26142 (ipa_single_use): Likewise.
26143 * passes.c (execute_todo): Likewise.
26144 * predict.c (drop_profile): Likewise.
26145 * symtab.c (symtab_node::get_dump_name): New function.
26146 (symtab_node::dump_name): Likewise.
26147 (symtab_node::dump_asm_name): Likewise.
26148 (symtab_node::dump_references): Likewise.
26149 (symtab_node::dump_referring): Likewise.
26150 (symtab_node::dump_base): Likewise.
26151 (symtab_node::debug_symtab): Likewise.
26152 * tree-sra.c (convert_callers_for_node): Likewise.
26153 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
26154 * value-prof.c (init_node_map): Likewise.
26156 2017-05-23 Martin Liska <mliska@suse.cz>
26158 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
26159 and symtab_node::debug_symtab to symbol_table::debug.
26160 * cgraphunit.c (analyze_functions): Use the renamed function.
26161 (symbol_table::compile): Likewise.
26162 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
26163 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
26164 * passes.c (execute_todo): Likewise.
26165 * symtab.c (symbol_table::dump): New function.
26166 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
26168 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26170 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
26171 that nonconst implies exec.
26173 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26175 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26176 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26177 (inline_edge_summary_vec): Turn into ...
26178 (ipa_call_summaries): ... this one.
26179 (redirect_to_unreachable, edge_set_predicate,
26180 evaluate_properties_for_edge, inline_summary_alloc,
26181 reset_ipa_call_summary, reset_inline_summary,
26182 inline_summary_t::duplicate): Update.
26183 (inline_edge_duplication_hook): Turn to ...
26184 (ipa_call_summary_t::duplicate): ... this one.
26185 (inline_edge_removal_hook): Turn to ...
26186 (ipa_call_summary_t::remove): ... this one.
26187 (dump_inline_edge_summary): Turn to ...
26188 (dump_ipa_call_summary): ... this one.
26189 (estimate_function_body_sizes): Update.
26190 (inline_update_callee_summaries): Update.
26191 (remap_edge_change_prob): Update.
26192 (remap_edge_summaries): Update.
26193 (inline_merge_summary): Update.
26194 (do_estimate_edge_time): Update.
26195 (inline_generate_summary): Update.
26196 (inline_read_section): Update.
26197 (inline_read_summary): Update.
26198 (inline_free_summary): Update.
26199 * ipa-inline.c (can_inline_edge_p): Update.
26200 (compute_inlined_call_time): Update.
26201 (want_inline_small_function_p): Update.
26202 (edge_badness): Update.
26203 (early_inliner): Update.
26204 * ipa-inline.h (inline_edge_summary): Turn to ...
26205 (ipa_call_summary): ... this one.
26206 (ipa_call_summary_t): New class.
26207 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26208 (ipa_call_summaries): New.
26209 (inline_edge_summary): Remove.
26210 (estimate_edge_growth): Update.
26211 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26212 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26213 * ipa-split.c (execute_split_functions): Update.
26214 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26216 2017-05-23 Tom de Vries <tom@codesourcery.com>
26218 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26219 attributes): Document rdrand effective target.
26221 2017-05-23 Tom de Vries <tom@codesourcery.com>
26223 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
26224 attributes): Sort alphabetically.
26226 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
26228 * config/avr/genmultilib.awk: Use gsub instead of gensub.
26230 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
26233 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
26234 V2DF/V2DI splat into two separate patterns, one that handles
26235 registers, and the other that only handles memory. Drop support
26236 for splatting from a GPR on ISA 2.07 and then splitting the
26237 splat into direct move and splat.
26238 (vsx_splat_<mode>_reg): Likewise.
26239 (vsx_splat_<mode>_mem): Likewise.
26241 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
26243 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
26245 2017-05-22 Jakub Jelinek <jakub@redhat.com>
26247 PR middle-end/80809
26248 * omp-low.c (finish_taskreg_remap): New function.
26249 (finish_taskreg_scan): If unit size of ctx->record_type
26250 is non-constant, unshare the size expression and replace
26251 decls in it with possible outer var refs.
26253 PR middle-end/80809
26254 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
26255 GOVD_SHARED rather than GOVD_PRIVATE with it.
26256 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
26257 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
26259 PR middle-end/80853
26260 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
26261 as last argument to build_outer_var_ref for pointer bases of array
26262 section reductions.
26264 2017-05-19 Martin Sebor <msebor@redhat.com>
26266 * print-tree.c (print_node): Print DECL_READ_P flag.
26268 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26270 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
26271 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
26272 * cgraph.c: Likewise.
26273 * cgraphunit.c: Likewise.
26274 * gengtype.c: Likewise.
26275 * ipa-cp.c: Likewise.
26276 * ipa-devirt.c: Likewise.
26277 * ipa-icf.c: Likewise.
26278 * ipa-predicate.c: Likewise.
26279 * ipa-profile.c: Likewise.
26280 * ipa-prop.c: Likewise.
26281 * ipa-split.c: Likewise.
26283 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
26284 edge_predicate_pool, dump_inline_hints,
26285 inline_summary::account_size_time, redirect_to_unreachable,
26286 edge_set_predicate, set_hint_predicate,
26287 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
26288 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
26289 inline_summary_t::remove, remap_hint_predicate_after_duplication,
26290 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
26291 ipa_call_summary_t::remove, initialize_growth_caches,
26292 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
26293 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
26294 mark_modified, unmodified_parm_1, unmodified_parm,
26295 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
26296 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
26297 compute_bb_predicates, will_be_nonconstant_expr_predicate,
26298 will_be_nonconstant_predicate, record_modified_bb_info,
26299 get_minimal_bb, record_modified, param_change_prob,
26300 phi_result_unknown_predicate, predicate_for_phi_result,
26301 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
26302 estimate_function_body_sizes, compute_inline_parameters,
26303 compute_inline_parameters_for_curren, pass_data_inline_parameters,
26304 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
26305 inline_update_callee_summaries, remap_edge_change_prob,
26306 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
26307 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
26308 inline_analyze_function, inline_summary_t::insert,
26309 inline_generate_summary, read_ipa_call_summary, inline_read_section,
26310 inline_read_summary, write_ipa_call_summary, inline_write_summary,
26311 inline_free_summary): Move to ipa-fnsummary.h
26312 (predicate_t): Remove.
26313 * ipa-fnsummary.c: New file.
26314 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
26315 (enum inline_hints_vals, inline_hints, agg_position_info,
26316 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
26317 inline_summaries, ipa_call_summary, ipa_call_summary_t,
26318 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
26319 dump_inline_summary, dump_inline_hints, inline_generate_summary,
26320 inline_read_summary, inline_write_summary, inline_free_summary,
26321 inline_analyze_function, initialize_inline_failed,
26322 inline_merge_summary, inline_update_overall_summary,
26323 compute_inline_parameters): Move to ipa-fnsummary.h
26324 * ipa-fnsummary.h: New file.
26325 * ipa-inline-transform.h: Include ipa-inline.h.
26326 * ipa-inline.c: LIkewise.
26328 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26330 * ipa-inline.c (edge_badness): Use inlined_time instead of
26331 inline_summaries->get.
26333 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26335 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
26337 2017-05-22 Nathan Sidwell <nathan@acm.org>
26339 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
26340 (fdump-lang): Document 'raw' option.
26341 * dumpfile.h (TDI_tu): Delete.
26342 * dumpfile.c (dump_files): Remove translation-unit.
26343 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
26345 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
26347 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
26348 command option from $(AWK) call.
26349 * config/avr/genmultilib.awk: Simplify and rewrite so that it
26350 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
26351 [FORMAT]: Remove handling of variable.
26352 * config/avr/t-multilib: Regenerate.
26354 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26356 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
26358 (dump_inline_summary): Do not print self_time.
26359 (estimate_function_body_sizes): Do not set self_time.
26360 (compute_inline_parameters): Likewise.
26361 (inline_read_section, inline_write_summary): Do not stream self_time.
26362 * ipa-inline.h (inline_summary): Drop self_time.
26364 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
26366 * ipa-inline-analysis.c (account_size_time): Rename to ...
26367 (inline_summary::account_size_time): ... this one.
26368 (reset_ipa_call_summary): Turn to ...
26369 (ipa_call_summary::reset): ... this one.
26370 (reset_inline_summary): Turn to ...
26371 (inline_summary::reset): ... this one.
26372 (inline_summary_t::remove): Update.
26373 (inline_summary_t::duplicate): Update.
26374 (ipa_call_summary_t::remove): Update.
26375 (dump_inline_summary): Update.
26376 (estimate_function_body_sizes): Update.
26377 (compute_inline_parameters): Update.
26378 (estimate_node_size_and_time): Update.
26379 (inline_merge_summary): Update.
26380 (inline_update_overall_summary): Update.
26381 (inline_read_section): Update.
26382 (inline_write_summary): Update.
26383 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
26384 add account_size_time and reset member functions.
26385 (ipa_call_summary): Add reset function.
26386 * ipa-predicate.h (predicate::operator &): Constify.
26388 2017-05-22 Richard Biener <rguenther@suse.de>
26390 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
26392 2017-05-19 Jason Merrill <jason@redhat.com>
26394 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
26396 2017-05-19 Marek Polacek <polacek@redhat.com>
26399 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
26400 TYPE_OVERFLOW_WRAPS checks.
26402 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
26404 * tree-core.h (enum omp_clause_default_kind): Add
26405 "OMP_CLAUSE_DEFAULT_PRESENT".
26406 * tree-pretty-print.c (dump_omp_clause): Handle it.
26407 * gimplify.c (enum gimplify_omp_var_data): Add
26408 "GOVD_MAP_FORCE_PRESENT".
26409 (gimplify_adjust_omp_clauses_1): Map it to
26410 "GOMP_MAP_FORCE_PRESENT".
26411 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
26413 * gimplify.c (oacc_default_clause): Clarify.
26415 2017-05-19 Nathan Sidwell <nathan@acm.org>
26417 LANG_HOOK_REGISTER_DUMPS
26418 * toplev.c (general_init): Call register dump lang hook.
26419 * doc/invoke.texi: Document -fdump-lang option family.
26420 * dumpfile.c (dump_files): Remove class dump here.
26421 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
26422 * dumpfile.h (tree_dump_index): Remove TDI_class.
26423 * langhooks-def.h (lhd_register_dumps): Declare.
26424 (LANG_HOOKS_REGISTER_DUMPS): Define.
26425 (LANG_HOOKS_INITIALIZER): Add it.
26426 * langhooks.c (lhd_register_dumps): Define.
26427 * langhooks.h (struct lang_hooks): Add register_dumps.
26429 2017-05-19 Nathan Sidwell <nathan@acm.org>
26431 * context.h (context::set_passes): New.
26432 * context.c (context::context): Do not create pass manager.
26433 * toplev.c (general_init): Create pass manager here.
26435 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
26437 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
26438 use this splitter if two add or or instructions would also work for
26439 the constant we want to generate.
26441 2017-05-19 Richard Biener <rguenther@suse.de>
26444 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
26445 predicate evaluation.
26447 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26449 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
26451 * ipa-inline.c (want_inline_small_function_p): Do not cast to
26454 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
26456 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
26457 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
26458 (inline_edge_summary_vec): Turn into ...
26459 (ipa_call_summaries): ... this one.
26460 (redirect_to_unreachable, edge_set_predicate,
26461 evaluate_properties_for_edge, inline_summary_alloc,
26462 reset_ipa_call_summary, reset_inline_summary,
26463 inline_summary_t::duplicate): Update.
26464 (inline_edge_duplication_hook): Turn to ...
26465 (ipa_call_summary_t::duplicate): ... this one.
26466 (inline_edge_removal_hook): Turn to ...
26467 (ipa_call_summary_t::remove): ... this one.
26468 (dump_inline_edge_summary): Turn to ...
26469 (dump_ipa_call_summary): ... this one.
26470 (estimate_function_body_sizes): Update.
26471 (inline_update_callee_summaries): Update.
26472 (remap_edge_change_prob): Update.
26473 (remap_edge_summaries): Update.
26474 (inline_merge_summary): Update.
26475 (do_estimate_edge_time): Update.
26476 (inline_generate_summary): Update.
26477 (inline_read_section): Update.
26478 (inline_read_summary): Update.
26479 (inline_free_summary): Update.
26480 * ipa-inline.c (can_inline_edge_p): Update.
26481 (compute_inlined_call_time): Update.
26482 (want_inline_small_function_p): Update.
26483 (edge_badness): Update.
26484 (early_inliner): Update.
26485 * ipa-inline.h (inline_edge_summary): Turn to ...
26486 (ipa_call_summary): ... this one.
26487 (ipa_call_summary_t): New class.
26488 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
26489 (ipa_call_summaries): New.
26490 (inline_edge_summary): Remove.
26491 (estimate_edge_growth): Update.
26492 * ipa-profile.c (ipa_propagate_frequency_1): Update.
26493 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26494 * ipa-split.c (execute_split_functions): Update.
26495 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
26497 2017-05-19 Richard Biener <rguenther@suse.de>
26499 PR middle-end/80764
26500 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
26502 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
26504 * config/rs6000/rs6000.c (struct machine_function): Add field
26505 fpr_is_wrapped_separately.
26506 (rs6000_get_separate_components): Use 64 components. Handle the
26507 new FPR components.
26508 (rs6000_components_for_bb): Handle the FPR components.
26509 (rs6000_emit_prologue_components): Handle the FPR components.
26510 (rs6000_emit_epilogue_components): Handle the FPR components.
26511 (rs6000_set_handled_components): Handle the FPR components.
26512 (rs6000_emit_prologue): Don't output prologue code for those FPRs
26513 that are already separately shrink-wrapped.
26514 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
26515 that are already separately shrink-wrapped.
26517 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26520 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
26523 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
26524 (define_peephole2 for Altivec d-form load): Add peepholes to catch
26525 cases where the register allocator uses a move and an offsettable
26526 memory operation to/from a FPR register on ISA 2.06/2.07.
26527 (define_peephole2 for Altivec d-form store): Likewise.
26529 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
26532 * config/i386/mmx.md (*mov<mode>_internal): Enable
26533 alternatives 11, 12, 13 and 14 also for 32bit targets.
26534 Remove alternatives 15, 16, 17 and 18.
26535 * config/i386/sse.md (vec_concatv2di): Change
26536 alternative (!x, *y) to (x, ?!*Yn).
26538 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
26540 * dumpfile.h (enum dump_kind): Remove stray comma.
26542 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26544 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
26545 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
26546 predicate::num_conditions
26547 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
26548 (CHANGED): turn into predicate::changed.
26549 (agg_position_info): Move to ipa-predicate.h
26550 (add_condition, predicate::add_clause, predicate::operator &=,
26551 predicate::or_with, predicate::evaluate, predicate::probability,
26552 dump_condition, dump_clause, predicate::dump,
26553 predicate::remap_after_duplication, predicate::remap_after_inlining,
26554 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
26555 (evaluate_conditions_for_known_args): Update.
26556 (set_cond_stmt_execution_predicate): Update.
26557 * ipa-inline.h: Include ipa-predicate.h
26558 (condition, inline_param_summary, conditions, agg_position_info,
26559 predicate): Move to ipa-predicate.h
26560 * ipa-predicate.c: New file.
26561 * ipa-predicate.h: New file.
26563 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
26565 * final.c (leaf_function_p): Check we are not in a sequence.
26567 2017-05-18 Martin Liska <mliska@suse.cz>
26569 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
26570 * dumpfile.c (dump_register): Use new enum dump_kind.
26571 (get_dump_file_name): Likewise.
26572 (dump_enable_all): Likewise.
26573 (dump_switch_p_1): Likewise.
26574 (enable_rtl_dump_file): Remove usage of TDF_RTL.
26575 * dumpfile.h (enum dump_kind): New enum type.
26576 (struct dump_file_info): Create constructor and
26577 format fields and comments.
26578 * passes.c (pass_manager::register_one_dump_file):
26580 * statistics.c (statistics_early_init): Likewise.
26581 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
26582 TDF_TREE with TDF_SLIM.
26583 (gather_memory_references_ref): Likewise.
26585 2017-05-18 Martin Liska <mliska@suse.cz>
26587 * vec.h (struct vnull): Use it.
26589 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
26591 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
26592 (true_predicate, false_predicate, true_predicate_p,
26593 false_predicate_p): Remove.
26594 (single_cond_predicate, not_inlined_predicate): Turn to member function
26596 (add_condition): Update.
26597 (add_clause): Turn to...
26598 (predicate::add_clause): ... this one; update; allow passing NULL
26600 (and_predicates): Turn to ...
26601 (predicate::operator &=): ... this one.
26602 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
26603 (or_predicates): Turn to ...
26604 (predicate::or_with): ... this one.
26605 (evaluate_predicate): Turn to ...
26606 (predicate::evaluate): ... this one.
26607 (predicate_probability): Turn to ...
26608 (predicate::probability): ... this one.
26609 (dump_condition): Update.
26610 (dump_predicate): Turn to ...
26611 (predicate::dump): ... this one.
26612 (account_size_time): Update.
26613 (edge_set_predicate): Update.
26614 (set_hint_predicate): UPdate.
26615 (evaluate_conditions_for_known_args): Update.
26616 (evaluate_properties_for_edge): Update.
26617 (remap_predicate_after_duplication): Turn to...
26618 (predicate::remap_after_duplication): ... this one.
26619 (remap_hint_predicate_after_duplication): Update.
26620 (inline_summary_t::duplicate): UPdate.
26621 (dump_inline_edge_summary): Update.
26622 (dump_inline_summary): Update.
26623 (set_cond_stmt_execution_predicate): Update.
26624 (set_switch_stmt_execution_predicate): Update.
26625 (compute_bb_predicates): Update.
26626 (will_be_nonconstant_expr_predicate): Update.
26627 (will_be_nonconstant_predicate): Update.
26628 (phi_result_unknown_predicate): Update.
26629 (predicate_for_phi_result): Update.
26630 (array_index_predicate): Update.
26631 (estimate_function_body_sizes): Update.
26632 (estimate_node_size_and_time): Update.
26633 (estimate_ipcp_clone_size_and_time): Update.
26634 (remap_predicate): Rename to ...
26635 (predicate::remap_after_inlining): ... this one.
26636 (remap_hint_predicate): Update.
26637 (inline_merge_summary): Update.
26638 (inline_update_overall_summary): Update.
26639 (estimate_size_after_inlining): Update.
26640 (read_predicate): Rename to ...
26641 (predicate::stream_in): ... this one.
26642 (read_inline_edge_summary): Update.
26643 (write_predicate): Rename to ...
26644 (predicate::stream_out): ... this one.
26645 (write_inline_edge_summary): Update.
26646 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
26647 (clause_t): Turn to uint32_t
26648 (predicate): Turn to class; implement constructor and operators
26650 (size_time_entry): Update.
26651 (inline_summary): Update.
26652 (inline_edge_summary): Update.
26654 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
26656 * fold-const.c (fold_binary_loc): Move transformation...
26657 * match.pd (C - X CMP X): ... here.
26659 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
26661 * config/sparc/sparc.c (sparc_option_override): Set function
26662 alignment for -mcpu=niagara7 to 64 to match the I$ line.
26663 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
26665 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
26667 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
26669 2017-05-18 Marek Polacek <polacek@redhat.com>
26672 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
26673 (pass_ubsan::execute): Call gimple_assign_single_p instead of
26674 gimple_assign_load_p.
26676 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
26678 PR middle-end/80692
26679 * real.c (do_compare): Give decimal_do_compare preference over
26680 comparing just the signs.
26682 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
26684 * doc/md.texi (Canonicalization of Instructions): Describe the
26685 canonical form of instructions that inherently set a condition
26688 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
26690 PR middle-end/80775
26691 * tree-cfg.c: Move deletion of unreachable case statements to after
26692 the merging of consecutive case labels.
26694 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
26696 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
26697 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
26698 restoring of callee-saved registers.
26700 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
26702 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
26703 * config/visium/visium.c (single_set_and_flags): Likewise.
26704 * config/visium/visium.md (Substitutions): Likewise.
26706 2017-05-17 Martin Liska <mliska@suse.cz>
26708 * cfg.c: Introduce dump_flags_t type and
26709 use it instead of int type.
26711 * cfghooks.c: Likewise.
26712 * cfghooks.h (struct cfg_hooks): Likewise.
26713 * cfgrtl.c: Likewise.
26714 * cfgrtl.h: Likewise.
26715 * cgraph.c (cgraph_node::get_body): Likewise.
26716 * coretypes.h: Likewise.
26717 * domwalk.c: Likewise.
26718 * domwalk.h: Likewise.
26719 * dumpfile.c (struct dump_option_value_info): Likewise.
26720 (dump_enable_all): Likewise.
26721 (dump_switch_p_1): Likewise.
26722 (opt_info_switch_p): Likewise.
26723 * dumpfile.h (enum tree_dump_index): Likewise.
26724 (struct dump_file_info): Likewise.
26725 * genemit.c: Likewise.
26726 * generic-match-head.c: Likewise.
26727 * gengtype.c (open_base_files): Likewise.
26728 * gimple-pretty-print.c: Likewise.
26729 * gimple-pretty-print.h: Likewise.
26730 * graph.c (print_graph_cfg): Likewise.
26731 * graphite-scop-detection.c (dot_all_sese): Likewise.
26732 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
26733 * loop-unroll.c (report_unroll): Likewise.
26734 * passes.c (pass_manager::register_one_dump_file): Likewise.
26735 * print-tree.c: Likewise.
26736 * statistics.c: Likewise.
26737 * tree-cfg.c: Likewise.
26738 * tree-cfg.h: Likewise.
26739 * tree-dfa.c: Likewise.
26740 * tree-dfa.h: Likewise.
26741 * tree-dump.c (dump_function): Likewise.
26742 * tree-dump.h (struct dump_info): Likewise.
26743 * tree-pretty-print.c: Likewise.
26744 * tree-pretty-print.h: Likewise.
26745 * tree-ssa-live.c: Likewise.
26746 * tree-ssa-live.h: Likewise.
26747 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
26748 * tree-vect-loop.c: Likewise.
26749 * tree-vect-slp.c: Likewise.
26751 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
26752 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26754 PR tree-optimization/80457
26755 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
26756 of all arguments to a statement as scalar_to_vec operations.
26757 (vectorizable_call): Adjust call to vect_model_simple_cost for
26759 (vectorizable_conversion): Likewise.
26760 (vectorizable_assignment): Likewise.
26761 (vectorizable_shift): Likewise.
26762 (vectorizable_operation): Likewise.
26763 (vectorizable_comparison): Likewise.
26764 (vect_is_simple_cond): Record the def types for operands.
26765 (vectorizable_condition): Likewise, call vect_model_simple_cost.
26766 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
26767 for statement argument count.
26769 2017-05-16 Carl Love <cel@us.ibm.com>
26771 * config/rs6000/rs6000-c: Add support for built-in functions
26772 vector unsigned long long vec_bperm (vector unsigned long long,
26773 vector unsigned char)
26774 vector signed long long vec_mule (vector signed int,
26776 vector unsigned long long vec_mule (vector unsigned int,
26777 vector unsigned int)
26778 vector signed long long vec_mulo (vector signed int,
26780 vector unsigned long long vec_mulo (vector unsigned int,
26781 vector unsigned int)
26782 vector signed char vec_sldw (vector signed char,
26783 vector signed char,
26785 vector unsigned char vec_sldw (vector unsigned char,
26786 vector unsigned char,
26788 vector signed short vec_sldw (vector signed short,
26789 vector signed short,
26791 vector unsigned short vec_sldw (vector unsigned short,
26792 vector unsigned short,
26794 vector signed int vec_sldw (vector signed int,
26797 vector unsigned int vec_sldw (vector unsigned int,
26798 vector unsigned int,
26800 vector signed long long vec_sldw (vector signed long long,
26801 vector signed long long,
26803 vector unsigned long long vec_sldw (vector unsigned long long,
26804 vector unsigned long long,
26806 * config/rs6000/rs6000-c: Add support for built-in functions
26807 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
26808 * config/rs6000/altivec.h: Add defintion for vec_sldw.
26809 * doc/extend.texi: Update the built-in documentation for the
26810 new built-in functions.
26812 2017-05-16 Marek Polacek <polacek@redhat.com>
26816 * tree.c (save_expr): Don't fold the expression.
26818 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
26820 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
26821 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
26822 and (?*y,m). Update insn attributes.
26824 2017-05-16 Martin Liska <mliska@suse.cz>
26826 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
26827 flags argument of print_gimple_stmt, print_gimple_expr,
26828 print_generic_stmt and print_generic_expr.
26829 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
26830 * coretypes.h: Likewise.
26831 * except.c (dump_eh_tree): Likewise.
26832 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
26833 * gimple-pretty-print.h: Likewise.
26834 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
26835 (backprop::push_to_worklist): Likewise.
26836 (backprop::pop_from_worklist): Likewise.
26837 (backprop::process_use): Likewise.
26838 (backprop::intersect_uses): Likewise.
26839 (note_replacement): Likewise.
26840 * gimple-ssa-store-merging.c
26841 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
26842 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
26843 (pass_store_merging::execute): Likewise.
26844 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
26845 (ssa_base_cand_dump_callback): Likewise.
26846 (dump_incr_vec): Likewise.
26847 (replace_refs): Likewise.
26848 (replace_mult_candidate): Likewise.
26849 (create_add_on_incoming_edge): Likewise.
26850 (create_phi_basis): Likewise.
26851 (insert_initializers): Likewise.
26852 (all_phi_incrs_profitable): Likewise.
26853 (introduce_cast_before_cand): Likewise.
26854 (replace_one_candidate): Likewise.
26855 * gimplify.c (gimplify_expr): Likewise.
26856 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
26857 (set_rename): Likewise.
26858 (rename_uses): Likewise.
26859 (copy_loop_phi_nodes): Likewise.
26860 (add_close_phis_to_merge_points): Likewise.
26861 (copy_loop_close_phi_args): Likewise.
26862 (copy_cond_phi_args): Likewise.
26863 (graphite_copy_stmts_from_block): Likewise.
26864 (translate_pending_phi_nodes): Likewise.
26865 * graphite-poly.c (print_pdr): Likewise.
26866 (dump_gbb_cases): Likewise.
26867 (dump_gbb_conditions): Likewise.
26868 (print_scop_params): Likewise.
26869 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
26870 (build_cross_bb_scalars_use): Likewise.
26871 (gather_bbs::before_dom_children): Likewise.
26872 * hsa-dump.c (dump_hsa_immed): Likewise.
26873 * ipa-cp.c (print_ipcp_constant_value): Likewise.
26874 (get_replacement_map): Likewise.
26875 * ipa-inline-analysis.c (dump_condition): Likewise.
26876 (estimate_function_body_sizes): Likewise.
26877 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
26878 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
26879 * ipa-prop.c (ipa_dump_param): Likewise.
26880 (ipa_print_node_jump_functions_for_edge): Likewise.
26881 (ipa_modify_call_arguments): Likewise.
26882 (ipa_modify_expr): Likewise.
26883 (ipa_dump_param_adjustments): Likewise.
26884 (ipa_dump_agg_replacement_values): Likewise.
26885 (ipcp_modif_dom_walker::before_dom_children): Likewise.
26886 * ipa-pure-const.c (check_stmt): Likewise.
26887 (pass_nothrow::execute): Likewise.
26888 * ipa-split.c (execute_split_functions): Likewise.
26889 * omp-offload.c (dump_oacc_loop_part): Likewise.
26890 (dump_oacc_loop): Likewise.
26891 * trans-mem.c (tm_log_emit): Likewise.
26892 (tm_memopt_accumulate_memops): Likewise.
26893 (dump_tm_memopt_set): Likewise.
26894 (dump_tm_memopt_transform): Likewise.
26895 * tree-cfg.c (gimple_verify_flow_info): Likewise.
26896 (print_loop): Likewise.
26897 * tree-chkp-opt.c (chkp_print_addr): Likewise.
26898 (chkp_gather_checks_info): Likewise.
26899 (chkp_get_check_result): Likewise.
26900 (chkp_remove_check_if_pass): Likewise.
26901 (chkp_use_outer_bounds_if_possible): Likewise.
26902 (chkp_reduce_bounds_lifetime): Likewise.
26903 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
26904 (chkp_mark_completed_bounds): Likewise.
26905 (chkp_register_incomplete_bounds): Likewise.
26906 (chkp_mark_invalid_bounds): Likewise.
26907 (chkp_maybe_copy_and_register_bounds): Likewise.
26908 (chkp_build_returned_bound): Likewise.
26909 (chkp_get_bound_for_parm): Likewise.
26910 (chkp_build_bndldx): Likewise.
26911 (chkp_get_bounds_by_definition): Likewise.
26912 (chkp_generate_extern_var_bounds): Likewise.
26913 (chkp_get_bounds_for_decl_addr): Likewise.
26914 * tree-chrec.c (chrec_apply): Likewise.
26915 * tree-data-ref.c (dump_data_reference): Likewise.
26916 (dump_subscript): Likewise.
26917 (dump_data_dependence_relation): Likewise.
26918 (analyze_overlapping_iterations): Likewise.
26919 * tree-inline.c (expand_call_inline): Likewise.
26920 (tree_function_versioning): Likewise.
26921 * tree-into-ssa.c (dump_defs_stack): Likewise.
26922 (dump_currdefs): Likewise.
26923 (dump_names_replaced_by): Likewise.
26924 (dump_update_ssa): Likewise.
26925 (update_ssa): Likewise.
26926 * tree-object-size.c (pass_object_sizes::execute): Likewise.
26927 * tree-parloops.c (build_new_reduction): Likewise.
26928 (try_create_reduction_list): Likewise.
26929 (ref_conflicts_with_region): Likewise.
26930 (oacc_entry_exit_ok_1): Likewise.
26931 (oacc_entry_exit_single_gang): Likewise.
26932 * tree-pretty-print.h: Likewise.
26933 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
26934 (get_scalar_evolution): Likewise.
26935 (add_to_evolution): Likewise.
26936 (get_loop_exit_condition): Likewise.
26937 (analyze_evolution_in_loop): Likewise.
26938 (analyze_initial_condition): Likewise.
26939 (analyze_scalar_evolution): Likewise.
26940 (instantiate_scev): Likewise.
26941 (number_of_latch_executions): Likewise.
26942 (gather_chrec_stats): Likewise.
26943 (final_value_replacement_loop): Likewise.
26944 (scev_const_prop): Likewise.
26945 * tree-sra.c (dump_access): Likewise.
26946 (disqualify_candidate): Likewise.
26947 (create_access): Likewise.
26948 (reject): Likewise.
26949 (maybe_add_sra_candidate): Likewise.
26950 (create_access_replacement): Likewise.
26951 (analyze_access_subtree): Likewise.
26952 (analyze_all_variable_accesses): Likewise.
26953 (sra_modify_assign): Likewise.
26954 (initialize_constant_pool_replacements): Likewise.
26955 (find_param_candidates): Likewise.
26956 (decide_one_param_reduction): Likewise.
26957 (replace_removed_params_ssa_names): Likewise.
26958 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
26959 * tree-ssa-copy.c (dump_copy_of): Likewise.
26960 (copy_prop_visit_cond_stmt): Likewise.
26961 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
26962 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
26963 (record_equivalences_from_stmt): Likewise.
26964 * tree-ssa-dse.c (compute_trims): Likewise.
26965 (delete_dead_call): Likewise.
26966 (delete_dead_assignment): Likewise.
26967 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
26968 (forward_propagate_into_cond): Likewise.
26969 (pass_forwprop::execute): Likewise.
26970 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
26971 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
26973 (move_computations_worker): Likewise.
26974 (execute_sm): Likewise.
26975 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
26976 (remove_exits_and_undefined_stmts): Likewise.
26977 (remove_redundant_iv_tests): Likewise.
26978 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
26979 (adjust_iv_update_pos): Likewise.
26980 * tree-ssa-math-opts.c (bswap_replace): Likewise.
26981 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
26982 (value_replacement): Likewise.
26983 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
26984 * tree-ssa-pre.c (print_pre_expr): Likewise.
26985 (get_representative_for): Likewise.
26986 (create_expression_by_pieces): Likewise.
26987 (insert_into_preds_of_block): Likewise.
26988 (eliminate_insert): Likewise.
26989 (eliminate_dom_walker::before_dom_children): Likewise.
26990 (eliminate): Likewise.
26991 (remove_dead_inserted_code): Likewise.
26992 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
26993 * tree-ssa-reassoc.c (get_rank): Likewise.
26994 (eliminate_duplicate_pair): Likewise.
26995 (eliminate_plus_minus_pair): Likewise.
26996 (eliminate_not_pairs): Likewise.
26997 (undistribute_ops_list): Likewise.
26998 (eliminate_redundant_comparison): Likewise.
26999 (update_range_test): Likewise.
27000 (optimize_range_tests_var_bound): Likewise.
27001 (optimize_vec_cond_expr): Likewise.
27002 (rewrite_expr_tree): Likewise.
27003 (rewrite_expr_tree_parallel): Likewise.
27004 (linearize_expr): Likewise.
27005 (break_up_subtract): Likewise.
27006 (linearize_expr_tree): Likewise.
27007 (attempt_builtin_powi): Likewise.
27008 (attempt_builtin_copysign): Likewise.
27009 (transform_stmt_to_copy): Likewise.
27010 (transform_stmt_to_multiply): Likewise.
27011 (dump_ops_vector): Likewise.
27012 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
27013 (print_scc): Likewise.
27014 (set_ssa_val_to): Likewise.
27015 (visit_reference_op_store): Likewise.
27016 (visit_use): Likewise.
27017 (sccvn_dom_walker::before_dom_children): Likewise.
27018 (run_scc_vn): Likewise.
27019 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
27021 (expr_hash_elt::print): Likewise.
27022 (const_and_copies::pop_to_marker): Likewise.
27023 (const_and_copies::record_const_or_copy_raw): Likewise.
27024 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
27025 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
27026 (dump_predicates): Likewise.
27027 (find_uninit_use): Likewise.
27028 (warn_uninitialized_phi): Likewise.
27029 (pass_late_warn_uninitialized::execute): Likewise.
27030 * tree-ssa.c (verify_vssa): Likewise.
27031 (verify_ssa): Likewise.
27032 (maybe_optimize_var): Likewise.
27033 * tree-vrp.c (dump_value_range): Likewise.
27034 (dump_all_value_ranges): Likewise.
27035 (dump_asserts_for): Likewise.
27036 (register_edge_assert_for_2): Likewise.
27037 (vrp_visit_cond_stmt): Likewise.
27038 (vrp_visit_switch_stmt): Likewise.
27039 (vrp_visit_stmt): Likewise.
27040 (vrp_visit_phi_node): Likewise.
27041 (simplify_cond_using_ranges_1): Likewise.
27042 (fold_predicate_in): Likewise.
27043 (evrp_dom_walker::before_dom_children): Likewise.
27044 (evrp_dom_walker::push_value_range): Likewise.
27045 (evrp_dom_walker::pop_value_range): Likewise.
27046 (execute_early_vrp): Likewise.
27048 2017-05-16 Richard Biener <rguenther@suse.de>
27050 * dwarf2out.c (loc_list_from_tree_1): Do not create
27051 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
27053 2017-05-16 Richard Biener <rguenther@suse.de>
27055 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
27057 (note_variable_value_in_expr): If we resolved the decl ref
27058 do not push to the stack.
27060 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
27062 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
27063 operations in fast-math mode.
27064 (vaddq_f16): Likewise.
27065 (vmul_f16): Likewise.
27066 (vmulq_f16): Likewise.
27067 (vsub_f16): Likewise.
27068 (vsubq_f16): Likewise.
27069 * config/arm/neon.md (add<mode>3): New.
27071 (fma:<VH:mode>3): New. Also remove outdated comment.
27074 2017-05-16 Martin Liska <mliska@suse.cz>
27078 * ipa-devirt.c (warn_types_mismatch): Fix typo.
27079 (odr_types_equivalent_p): Likewise.
27081 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
27083 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
27085 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
27088 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
27089 non-interunit SSE move alternatives with '?'.
27090 (zero-extendsidi peephole2): New peephole to skip intermediate
27091 general register in SSE zero-extend sequence.
27093 2017-05-15 Jeff Law <law@redhat.com>
27095 * reorg.c (relax_delay_slots): Create a new variable to hold
27096 the temporary target rather than clobbering TARGET_LABEL.
27098 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
27099 missing argument to extract_bit_field call.
27100 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
27102 2017-05-15 Martin Liska <mliska@suse.cz>
27105 * gcc.c (process_command): Do not allow empty argument of -o option.
27107 2017-05-15 Renlin Li <renlin.li@arm.com>
27109 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
27110 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
27111 * config/aarch64/constraints.md (Usf): Add long call check.
27112 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
27113 (call_value): Likewise.
27114 (sibcall): Likewise.
27115 (sibcall_value): Likewise.
27117 (call_value_insn): New.
27118 (sibcall_insn): Update rtx pattern.
27119 (sibcall_value_insn): Likewise.
27120 (call_internal): Remove.
27121 (call_value_internal): Likewise.
27122 (sibcall_internal): Likewise.
27123 (sibcall_value_internal): Likewise.
27124 (call_reg): Likewise.
27125 (call_symbol): Likewise.
27126 (call_value_reg): Likewise.
27127 (call_value_symbol): Likewise.
27129 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
27132 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
27134 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
27136 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
27137 compatible with CCGOCmode and with CCZmode.
27139 2017-05-14 Martin Sebor <msebor@redhat.com>
27141 PR middle-end/77671
27142 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
27143 (gimple_fold_builtin_snprintf): Same.
27144 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
27145 (gimple_fold_builtin_snprintf): Same.
27146 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
27147 of character types.
27148 (is_call_safe): New function.
27149 (try_substitute_return_value): Call it.
27150 (try_simplify_call): New function.
27151 (pass_sprintf_length::handle_gimple_call): Call it.
27153 2017-05-14 Martin Sebor <msebor@redhat.com>
27155 PR middle-end/80669
27156 * builtins.c (expand_builtin_stpncpy): Simplify.
27158 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
27160 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
27161 * config/i386/i386.h
27162 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
27163 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
27164 (struct machine_function): Add new members call_ms2sysv,
27165 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
27166 (struct machine_frame_state): New fields sp_realigned and
27167 sp_realigned_offset.
27168 * config/i386/i386.c
27169 (enum xlogue_stub): New enum.
27170 (enum xlogue_stub_sets): New enum.
27171 (class xlogue_layout): New class.
27172 (struct ix86_frame): New fields stack_realign_allocate_offset,
27173 stack_realign_offset and outlined_save_offset. Modify comments to
27174 detail stack layout when using out-of-line stubs.
27175 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
27176 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
27177 -mcall-ms2sysv-xlogues.
27178 (stub_managed_regs): New static variable.
27179 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
27180 registers managed by out-of-line stub.
27181 (disable_call_ms2sysv_xlogues): New function.
27182 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
27183 m->call_ms2sysv when appropriate and compute frame layout for
27185 (sp_valid_at, fp_valid_at): New inline functions.
27186 (choose_basereg): New function.
27187 (choose_baseaddr): Add align parameter, use choose_basereg and modify
27189 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
27190 Use align parameter of choose_baseaddr to generated aligned SSE movs
27192 (pro_epilogue_adjust_stack): Modify to track
27193 machine_frame_state::sp_realigned.
27194 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
27195 (ix86_nsaved_sseregs): Likewise.
27196 (ix86_emit_save_regs): Likewise.
27197 (ix86_emit_save_regs_using_mov): Likewise.
27198 (ix86_emit_save_sse_regs_using_mov): Likewise.
27199 (get_scratch_register_on_entry): Likewise.
27200 (gen_frame_set): New function.
27201 (gen_frame_load): Likewise.
27202 (gen_frame_store): Likewise.
27203 (emit_outlined_ms2sysv_save): Likewise.
27204 (emit_outlined_ms2sysv_restore): Likewise.
27205 (ix86_expand_prologue): Modify stack re-alignment code and call
27206 emit_outlined_ms2sysv_save when appropriate.
27207 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
27208 parameter rtx_insn *insn, which allows the function to be used to only
27209 generate the notes.
27210 (ix86_expand_epilogue): Modify validity checks of frame and stack
27211 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
27212 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
27213 * config/i386/predicates.md
27214 (save_multiple): New predicate.
27215 (restore_multiple): Likewise.
27216 * config/i386/sse.md
27217 (save_multiple<mode>): New pattern.
27218 (save_multiple_realign<mode>): Likewise.
27219 (restore_multiple<mode>): Likewise.
27220 (restore_multiple_and_return<mode>): Likewise.
27221 (restore_multiple_leave_return<mode>): Likewise.
27222 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
27224 2017-05-14 Julia Koval <julia.koval@intel.com>
27226 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
27227 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
27228 (__builtin_ia32_xsetbv): New builtins.
27229 * config/i386/i386.c (ix86_expand_special_args_builtin):
27231 (ix86_expand_builtin): Special expand for new intrinsics.
27232 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
27233 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
27234 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
27236 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27238 * cfganal.c (inverted_post_order_compute): Change argument type
27240 * cfganal.h (inverted_post_order_compute): Adjust prototype.
27241 * df-core.c (rest_of_handle_df_initialize): Adjust.
27242 (rest_of_handle_df_finish): Likewise.
27243 (df_analyze_1): Likewise.
27244 (df_analyze): Likewise.
27245 (loop_inverted_post_order_compute): Change argument to be a vec *.
27246 (df_analyze_loop): Adjust.
27247 (df_get_n_blocks): Likewise.
27248 (df_get_postorder): Likewise.
27249 * df.h (struct df_d): Change field to be a vec.
27250 * lcm.c (compute_laterin): Adjust.
27251 (compute_available): Likewise.
27252 * lra-lives.c (lra_create_live_ranges_1): Likewise.
27253 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
27254 * tree-ssa-pre.c (compute_antic): Likewise.
27256 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27258 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
27259 (depth_first_search::depth_first_search): Change structure init
27260 function to this constructor.
27261 (depth_first_search::add_bb): Rename function to this member.
27262 (depth_first_search::execute): Likewise.
27263 (flow_dfs_compute_reverse_finish): Adjust.
27265 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27267 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
27268 (longest_simple_path): Likewise.
27269 * shrink-wrap.c (spread_components): Likewise.
27270 (disqualify_problematic_components): Likewise.
27271 (emit_common_heads_for_components): Likewise.
27272 (emit_common_tails_for_components): Likewise.
27273 (insert_prologue_epilogue_for_components): Likewise.
27275 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27277 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
27280 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27282 * df-core.c (df_set_blocks): Start using auto_bitmap.
27283 (df_compact_blocks): Likewise.
27284 * df-problems.c (df_rd_confluence_n): Likewise.
27285 * df-scan.c (df_insn_rescan_all): Likewise.
27286 (df_process_deferred_rescans): Likewise.
27287 (df_update_entry_block_defs): Likewise.
27288 (df_update_exit_block_uses): Likewise.
27289 (df_entry_block_bitmap_verify): Likewise.
27290 (df_exit_block_bitmap_verify): Likewise.
27291 (df_scan_verify): Likewise.
27292 * lra-constraints.c (lra_constraints): Likewise.
27293 (undo_optional_reloads): Likewise.
27294 (lra_undo_inheritance): Likewise.
27295 * lra-remat.c (calculate_gen_cands): Likewise.
27296 (do_remat): Likewise.
27297 * lra-spills.c (assign_spill_hard_regs): Likewise.
27298 (spill_pseudos): Likewise.
27299 * tree-ssa-pre.c (bitmap_set_and): Likewise.
27300 (bitmap_set_subtract_values): Likewise.
27302 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27304 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
27305 management with auto_bitmap.
27306 (fix_inter_tick): Likewise.
27307 (fix_recovery_deps): Likewise.
27308 * ira.c (add_store_equivs): Likewise.
27309 (find_moveable_pseudos): Likewise.
27310 (split_live_ranges_for_shrink_wrap): Likewise.
27311 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
27312 (rtx_reuse_manager::seen_def_p): Likewise.
27313 (rtx_reuse_manager::set_seen_def): Likewise.
27314 * print-rtl.h (class rtx_reuse_manager): Likewise.
27316 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27318 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
27320 (migrate_btr_def): Likewise.
27321 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
27322 * df-core.c (loop_post_order_compute): Likewise.
27323 (loop_inverted_post_order_compute): Likewise.
27324 * hsa-common.h: Likewise.
27325 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
27326 * init-regs.c (initialize_uninitialized_regs): Likewise.
27327 * ipa-inline.c (resolve_noninline_speculation): Likewise.
27328 (inline_small_functions): Likewise.
27329 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
27330 * ira.c (combine_and_move_insns): Likewise.
27331 (build_insn_chain): Likewise.
27332 * loop-invariant.c (find_invariants): Likewise.
27333 * lower-subreg.c (propagate_pseudo_copies): Likewise.
27334 * predict.c (tree_predict_by_opcode): Likewise.
27335 (predict_paths_leading_to): Likewise.
27336 (predict_paths_leading_to_edge): Likewise.
27337 (estimate_loops_at_level): Likewise.
27338 (estimate_loops): Likewise.
27339 * shrink-wrap.c (try_shrink_wrapping): Likewise.
27340 (spread_components): Likewise.
27341 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
27342 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
27343 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
27344 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
27345 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
27346 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
27347 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
27348 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
27349 (mark_threaded_blocks): Likewise.
27350 (thread_through_all_blocks): Likewise.
27351 * tree-ssa.c (verify_ssa): Likewise.
27352 (execute_update_addresses_taken): Likewise.
27353 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
27355 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27357 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
27359 (post_order_compute): Likewise.
27360 (inverted_post_order_compute): Likewise.
27361 (pre_and_rev_post_order_compute_fn): Likewise.
27363 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27365 * genrecog.c (int_set::int_set): Explicitly construct our
27366 auto_vec base class.
27367 * vec.h (auto_vec::auto_vec): New constructor.
27369 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27371 * bitmap.h (class auto_bitmap): New constructor taking
27372 bitmap_obstack * argument.
27374 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
27376 * bitmap.h (class auto_bitmap): Change type of m_bits to
27377 bitmap_head, and adjust ctor / dtor and member operators.
27379 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
27381 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
27382 when returned register mode doesn't match original mode.
27384 2017-05-12 Jeff Law <law@redhat.com>
27385 Jakub Jelinek <jakub@redhat.com>
27387 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
27388 we look for cc setter after the compare-elim changes.
27389 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
27390 within the vector to match what compare-elim now expects.
27391 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
27392 (xorsi3_flags, one_cmplsi2_flags): Likewise.
27394 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
27395 after the compare-elim changes.
27396 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
27397 the vector to match what compare-elim now expects.
27398 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
27399 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
27400 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
27401 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
27402 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
27404 * config/visium/visium.c (single_set_and_flags): Fix where
27405 we look for cc setter after the compare-elim changes.
27406 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
27407 with the vector to match what compare-elim now expects.
27408 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
27409 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
27410 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
27411 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
27412 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
27413 (neg<mode>2_insn_set_overflow): Likewise.
27415 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
27417 PR middle-end/79794
27418 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
27419 maybe_expand_insn call, set ops[0].target. If still set after call,
27420 set alt_rtl. Add extra arg to recursive calls.
27421 (extract_bit_field): Add alt_rtl argument. Pass to
27423 * expmed.h (extract_bit_field): Fix prototype.
27424 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
27425 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
27426 to extract_bit_field_calls.
27427 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
27428 Pass alt_rtl to extract_bit_field calls.
27429 * calls.c (store_unaligned_arguments_into_psuedos)
27430 load_register_parameters): Pass extra NULL to extract_bit_field calls.
27431 * optabs.c (maybe_legitimize_operand): Clear op->target when call
27433 * optabs.h (struct expand_operand): Add target bitfield.
27435 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27437 * compare-elim.c (try_eliminate_compare): Canonicalize
27438 operation with embedded compare to
27439 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
27440 (set (reg) (operation)].
27442 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
27444 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
27447 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
27448 cost of adding a carry flag for ADC instruction.
27449 [case MINUS]: Ignore the cost of subtracting a carry flag
27450 for SBB instruction.
27452 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
27454 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
27456 * config/rs6000/bmiintrin.h: New file.
27457 * config/rs6000/bmi2intrin.h: New file.
27458 * config/rs6000/x86intrin.h: New file.
27460 2017-05-12 Jeff Law <law@redhat.com>
27462 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
27465 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
27467 PR middle-end/80707
27468 * tree-cfg.c: Remove cfg edges of unreachable case statements.
27470 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27472 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27473 early expansion of vector divide builtins.
27474 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
27475 builtins identified as having unsigned arguments.
27477 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27479 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
27480 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
27481 expansion of vector logical operations (and, andc, or, xor,
27484 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27486 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
27487 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
27489 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
27491 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
27492 early GIMPLE expansion of vector multiplies.
27494 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27496 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
27497 TARGET_HAVE_MOVT conditional.
27498 (movt splitter): Likewise.
27500 2017-05-12 Richard Biener <rguenther@suse.de>
27502 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
27503 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27504 Fold all stmts not inplace.
27506 2017-05-12 Richard Biener <rguenther@suse.de>
27508 PR tree-optimization/80713
27509 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
27510 inserted_exprs bit for not removed stmts.
27512 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
27514 PR middle-end/69921
27515 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
27516 parallelized" attribute for parallelized OpenACC kernels.
27517 * omp-offload.c (execute_oacc_device_lower): Use it.
27519 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
27520 Set "oacc kernels" attribute.
27521 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
27522 parameter. Adjust all users.
27523 (oacc_fn_attrib_kernels_p): Remove function.
27524 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
27525 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
27526 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
27527 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
27528 assert "oacc kernels" attribute is set.
27530 2017-05-11 Carl Love <cel@us.ibm.com>
27532 * config/rs6000/rs6000-c: Add support for built-in functions
27533 vector unsigned char vec_popcnt (vector signed char)
27534 vector unsigned char vec_popcnt (vector unsigned char)
27535 vector unsigned short vec_popcnt (vector signed short)
27536 vector unsigned short vec_popcnt (vector unsigned short)
27537 vector unsigned int vec_popcnt (vector signed int)
27538 vector unsigned int vec_popcnt (vector unsigned int)
27539 vector unsigned long long vec_popcnt (vector signed long long)
27540 vector unsigned long long vec_popcnt (vector unsigned long long)
27541 vector signed long long vec_slo (vector signed long long,
27542 vector signed char)
27543 vector signed long long vec_slo (vector signed long long,
27544 vector unsigned char)
27545 vector unsigned long long vec_slo (vector unsigned long long,
27546 vector signed char)
27547 vector unsigned long long vec_slo (vector unsigned long long,
27548 vector unsigned char)
27549 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
27550 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
27551 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
27552 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
27553 * doc/extend.texi: Update the built-in documentation file for the
27554 new built-in functions.
27556 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
27558 * attribs.h (sorted_attr_string): Move machine independent
27559 functions for target clone support from the i386 port to common
27560 code. Rename ix86_function_versions to common_function_versions.
27561 Rename make_name to make_unique_name.
27562 (common_function_versions): Likewise.
27563 (make_unique_name): Likewise.
27564 (make_dispatcher_decl): Likewise.
27565 (is_function_default_version): Likewise.
27566 * attribs.c (attr_strcmp): Likewise.
27567 (sorted_attr_string): Likewise.
27568 (common_function_versions): Likewise.
27569 (make_unique_name): Likewise.
27570 (make_dispatcher_decl): Likewise.
27571 (is_function_default_version): Likewise.
27572 * config/i386/i386.c (attr_strcmp): Likewise.
27573 (sorted_attr_string): Likewise.
27574 (ix86_function_versions): Likewise.
27575 (make_name): Likewise.
27576 (make_dispatcher_decl): Likewise.
27577 (is_function_default_version): Likewise.
27578 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
27580 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27583 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
27584 Account for direct move costs for vec_construct of integer
27587 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
27590 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
27591 (UNSPEC_STX_ATOMIC): Ditto.
27592 (loaddi_via_sse): New insn.
27593 (storedi_via_sse): Ditto.
27594 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
27595 Update corresponding peephole2 patterns.
27596 (atomic_storedi_fpu): Ditto.
27598 2017-05-11 Julia Koval <julia.koval@intel.com>
27600 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
27601 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
27603 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
27604 (__builtin_ia32_rsqrt14ss_mask): New builtins.
27605 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
27607 2017-05-11 Nathan Sidwell <nathan@acm.org>
27609 * graphite-poly.c: Include dumpfile.h.
27611 * dumpfle.h (dump_function): Declare here ...
27612 * tree-dump.h (dump_function): ... not here.
27613 * dumpfile.c: #include tree-cfg.h.
27614 (dump_function): Move here from ...
27615 * tree-dump.c (dump_function): ... here.
27616 * gimplify.c: #include splay-tree.h, not tree-dump.h.
27617 * graphite-poly.c: Don't include tree-dump.h.
27618 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
27619 * print-tree.c: Likewise.
27620 * stor-layout.c: Likewise.
27621 * tree-nested.c: Likewise.
27623 * dumpfile.c (dump_start): Use TDF_FLAGS.
27624 (dump_enable_all): Fix TDF_KIND check thinko.
27626 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
27628 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27629 array entries to represent two legal parameterizations of the
27630 overloaded __builtin_cmpb function, as represented by the
27631 P6_OV_BUILTIN_CMPB constant.
27632 (altivec_resolve_overloaded_builtin): Add special case handling
27633 for the __builtin_cmpb function, as represented by the
27634 P6_OV_BUILTIN_CMPB constant.
27635 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
27636 (BU_P6_64BIT_2): New macro.
27637 (BU_P6_OVERLOAD_2): New macro
27638 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
27639 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
27640 (CMPB): Add overload support to represent both 32-bit and 64-bit
27641 compare-bytes function.
27642 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
27643 support for TARGET_CMPB.
27644 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
27645 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27646 documentation of the __builtin_cmpb overloaded built-in function.
27648 2017-05-11 Richard Biener <rguenther@suse.de>
27650 PR tree-optimization/80705
27651 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
27652 bases are not vectorizable.
27654 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27656 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
27657 when counting register pressure.
27659 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27661 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
27662 (iv_ca_more_deps): Renamed to ...
27663 (iv_ca_compare_deps): ... this.
27664 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
27666 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27668 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
27670 (determine_group_iv_costs): ... here.
27671 (find_inv_vars_cb): Record inv var if it's not recorded before.
27673 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27675 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
27676 (get_shiftadd_cost): Ditto.
27678 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27680 * tree-ssa-address.c: Include header file.
27681 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
27683 (add_to_parts): Refactor.
27684 (addr_to_parts): New parameter. Update use of move_hint_to_base.
27685 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
27688 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27690 PR tree-optimization/53090
27691 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
27693 (extract_cond_operands): Detect condition with IV on both sides
27694 and return COMP_IV_EXPR_2.
27695 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
27696 (rewrite_use_compare): Simplify by removing call to function
27697 extract_cond_operands.
27699 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27701 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
27702 (extract_cond_operands): Detect condition comparing against non-
27703 invariant bound and return appropriate enum value.
27704 (find_interesting_uses_cond): Update use of extract_cond_operands.
27705 Handle its return value accordingly.
27706 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
27708 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27710 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
27711 nonlinear iv_use computation in loop invariant sensitive way.
27713 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27715 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
27716 (find_iv_candidates): Call relate_compare_use_with_all_cands.
27718 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27720 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
27721 (dump_cand): Support iv_cand.inv_exprs.
27722 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
27724 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
27727 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27729 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
27731 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
27732 as local function. Include necessary header files.
27733 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
27735 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27737 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
27739 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27741 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
27742 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
27743 RSHIFT_EXPR and BIT_NOT_EXPR.
27745 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27747 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
27748 (adjust_setup_cost): New parameter supporting round up adjustment.
27749 (struct address_cost_data): Delete.
27750 (force_expr_to_var_cost): Don't bound cost with spill_cost.
27751 (split_address_cost, ptr_difference_cost): Delete.
27752 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
27753 (struct ainc_cost_data): New struct.
27754 (get_address_cost_ainc): New function.
27755 (get_address_cost, get_computation_cost): Reimplement.
27756 (determine_group_iv_cost_address): Record inv_expr for all uses of
27758 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
27759 (iv_ca_has_deps): Reimplemented to ...
27760 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
27762 (iv_ca_extend): Call iv_ca_more_deps.
27764 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27766 * tree-ssa-address.c (struct mem_address): Move to header file.
27767 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
27768 * tree-ssa-address.h (struct mem_address): Move from C file.
27769 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
27771 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27773 * tree-affine.h (aff_combination_type): New interface.
27774 (aff_combination_zero_p): Remove static.
27775 (aff_combination_const_p): New interface.
27776 (aff_combination_singleton_var_p): New interfaces.
27778 2017-05-11 Richard Biener <rguenther@suse.de>
27780 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
27781 Skip unreachable blocks and destinations.
27782 (eliminate): Move stmt removal and fixup ...
27783 (fini_eliminate): ... here. Skip inserted exprs.
27784 (pass_pre::execute): Move fini_pre after fini_eliminate.
27785 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
27786 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
27787 PRE to get rid of dead code that has invalid SSA form and
27788 split critical edges again.
27790 2017-05-11 Bin Cheng <bin.cheng@arm.com>
27792 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
27794 2017-05-11 Richard Biener <rguenther@suse.de>
27796 * passes.c (execute_function_todo): Verify loops if they are
27797 said to be up-to-date.
27798 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
27799 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
27801 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
27804 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
27805 handle calling assemble_external ourself.
27808 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
27809 modes with zero size. Enhance comment.
27811 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27813 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
27814 built-ins for vec_xl and vec_xst with short and char pointer
27817 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27819 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
27820 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
27821 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
27822 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
27823 (_mm_maskz_min_round_ss): New intrinsics.
27824 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
27825 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
27826 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
27827 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
27828 (__builtin_ia32_minss_mask_round): New builtins.
27829 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27830 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27831 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
27833 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
27834 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
27836 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
27839 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
27841 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
27842 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
27843 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
27844 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
27845 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
27846 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
27847 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
27848 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27849 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
27850 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
27851 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
27852 (__builtin_ia32_mulss_mask_round): New builtins.
27853 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
27854 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
27855 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
27857 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
27858 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
27860 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
27863 2017-05-10 Julia Koval <julia.koval@intel.com>
27865 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
27866 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
27867 (_mm256_setr_m128i): New intrinsics.
27869 2017-05-10 Julia Koval <julia.koval@intel.com>
27871 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
27872 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
27873 (_mm_maskz_rcp14_ss): New intrinsics.
27874 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
27875 (__builtin_ia32_rcp14ss_mask): New builtins.
27876 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
27878 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
27880 PR tree-optimization/51513
27881 * tree-cfg.c (gimple_seq_unreachable_p): New function.
27882 (assert_unreachable_fallthru_edge_p): Use it.
27883 (group_case_labels_stmt): Likewise.
27884 * tree-cfg.h: Prototype it.
27885 * stmt.c: Include cfghooks.h and tree-cfg.h.
27886 (emit_case_dispatch_table) <gap_label>: New local variable.
27887 Use it to fill dispatch table gaps.
27888 Test for default_label before updating probabilities.
27889 (expand_case) <default_label>: Remove unneeded initialization.
27890 Test for unreachable default case statement and remove its edge.
27891 Set default_label accordingly.
27892 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
27894 2017-05-10 Carl Love <cel@us.ibm.com>
27896 * config/rs6000/rs6000-c: Add support for built-in functions
27897 vector signed char vec_neg (vector signed char)
27898 vector signed short int vec_neg (vector short int)
27899 vector signed int vec_neg (vector signed int)
27900 vector signed long long vec_neg (vector signed long long)
27901 vector float vec_neg (vector float)
27902 vector double vec_neg (vector double)
27903 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
27905 * config/rs6000/altivec.h: Add define for vec_neg
27906 * doc/extend.texi: Update the built-in documentation for the
27907 new built-in functions.
27909 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27911 PR tree-optimization/77644
27912 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
27914 2017-05-10 Nathan Sidwell <nathan@acm.org>
27916 * dumpfile.h (TDI_lang_all): New.
27917 (TDF_KIND): New. Renumber others
27918 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
27920 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
27922 (get_dump_file_name): Adjust suffix generation.
27923 (dump_enable_all): Use TDF_KIND.
27924 * doc/invoke.texi (-fdump-lang-all): Document.
27926 * dumpfile.h: Tabify.
27928 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
27931 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
27932 Move member access before delete.
27934 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
27936 * tree-inline.c (expand_call_inline): Split block at stmt
27939 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
27942 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
27943 are now unused after splitting mov{sf,sd}_hardfloat.
27944 (f32_lr2): Likewise.
27945 (f32_lm): Likewise.
27946 (f32_lm2): Likewise.
27947 (f32_li): Likewise.
27948 (f32_li2): Likewise.
27949 (f32_lv): Likewise.
27950 (f32_sr): Likewise.
27951 (f32_sr2): Likewise.
27952 (f32_sm): Likewise.
27953 (f32_sm2): Likewise.
27954 (f32_si): Likewise.
27955 (f32_si2): Likewise.
27956 (f32_sv): Likewise.
27957 (f32_dm): Likewise.
27958 (f32_vsx): Likewise.
27959 (f32_av): Likewise.
27960 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
27961 For movsf, order stores so the VSX stores occur before the GPR
27962 store which encourages the register allocator to use a traditional
27963 FPR instead of a GPR. For movsd, order the stores so that the GPR
27964 store comes before the VSX stores to allow the power6 to work.
27965 This is due to the power6 not having a 32-bit integer store
27966 instruction from a FPR.
27967 (movsf_hardfloat): Likewise.
27968 (movsd_hardfloat): Likewise.
27970 2017-05-09 Martin Sebor <msebor@redhat.com>
27972 PR translation/80280
27973 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
27976 PR translation/80280
27977 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
27978 data member added in r247778.
27979 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
27981 2017-05-09 Nathan Sidwell <nathan@acm.org>
27983 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
27985 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
27988 2017-05-09 Marek Polacek <polacek@redhat.com>
27990 * doc/invoke.texi: Fix typo.
27992 2017-05-09 Richard Biener <rguenther@suse.de>
27994 * tree-vrp.c (vrp_val_is_max): Adjust comment.
27995 (vrp_val_is_min): Likewise.
27996 (set_value_range_to_value): Likewise.
27997 (set_value_range_to_nonnegative): Likewise.
27998 (gimple_assign_nonzero_p): Likewise.
27999 (gimple_stmt_nonzero_p): Likewise.
28000 (vrp_int_const_binop): Likewise. Remove unreachable case.
28001 (adjust_range_with_scev): Adjust comments.
28002 (compare_range_with_value): Likewise.
28003 (extract_range_from_phi_node): Likewise.
28004 (test_for_singularity): Likewise.
28006 2017-05-09 Richard Biener <rguenther@suse.de>
28008 * tree-vrp.c (get_single_symbol): Add assert that we don't
28009 get overflowed constants as invariant part.
28010 (compare_values_warnv): Add comment before the TREE_NO_WARNING
28011 checks. Use wi::cmp instead of recursing for integer constants.
28012 (compare_values): Just ignore whether we assumed undefined
28013 overflow instead of failing the compare.
28014 (extract_range_for_var_from_comparison_expr): Add comment before the
28015 TREE_NO_WARNING sets.
28016 (test_for_singularity): Likewise.
28017 (extract_range_from_comparison): Do not disable optimization
28018 when we assumed undefined overflow.
28019 (extract_range_basic): Remove init of unused var.
28021 2017-05-09 Richard Biener <rguenther@suse.de>
28023 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
28024 (extract_range_from_multiplicative_op_1): Adjust.
28025 (extract_range_from_binary_expr_1): Use int_const_binop.
28027 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
28030 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
28031 rs6000_store_data_bypass_p in seven define_bypass directives and
28032 in several comments.
28033 * config/rs6000/rs6000-protos.h: Add prototype for
28034 rs6000_store_data_bypass_p function.
28035 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
28036 function implements slightly different (rs6000-specific) semantics
28037 than store_data_bypass_p, returning false rather than aborting
28038 with assertion error when arguments do not satisfy the
28039 requirements of store data bypass.
28040 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
28041 rs6000_store_data_bypass_p.
28043 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
28045 * config/xtensa/xtensa-protos.h
28046 (xtensa_initial_elimination_offset): New declaration.
28047 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28048 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
28049 macro definition, add case for FRAME_POINTER_REGNUM when
28050 FRAME_GROWS_DOWNWARD.
28051 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
28052 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
28053 xtensa_initial_elimination_offset.
28055 2017-05-08 Nathan Sidwell <nathan@acm.org>
28057 * doc/invoke.texi: Alphabetize -fdump options.
28059 2017-05-08 Martin Sebor <msebor@redhat.com>
28061 PR translation/80280
28062 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
28064 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
28066 * target.def (compute_frame_layout): New optional target hook.
28067 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
28068 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
28069 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
28071 * reload1.c (verify_initial_elim_offsets): Likewise.
28072 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
28073 (use_simple_return_p): Call arm_compute_frame_layout if needed.
28074 (arm_get_frame_offsets): Split up into this ...
28075 (arm_compute_frame_layout): ... and this function.
28077 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
28079 * config/aarch64/constraints.md (Usa): New constraint.
28080 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
28082 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
28084 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
28085 with_multilib_list after it has been checked.
28087 2017-05-08 Richard Biener <rguenther@suse.de>
28089 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
28090 (bitmap_set_subtract_values): Likewise.
28092 2017-05-08 Richard Biener <rguenther@suse.de>
28094 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
28095 (gimple_assign_nonzero): ... this and remove strict_overflow_p
28097 (gimple_stmt_nonzero_warnv_p): Rename to ...
28098 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
28100 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
28101 (extract_range_basic): Adjust, do not disable propagation on
28102 strict overflow sensitive simplification.
28103 (vrp_visit_cond_stmt): Likewise.
28105 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
28107 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
28108 body size unconditionally.
28110 2017-05-07 Jeff Law <law@redhat.com>
28113 2017-05-06 Jeff Law <law@redhat.com>
28114 PR tree-optimization/78496
28115 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28118 PR tree-optimization/78496
28119 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28120 (simplify_stmt_using_ranges): Call it.
28121 (vrp_dom_walker::before_dom_children): Extract equivalences
28122 from an ASSERT_EXPR with an equality comparison against a
28125 2017-05-06 Jeff Law <law@redhat.com>
28127 PR tree-optimization/78496
28128 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
28131 PR tree-optimization/78496
28132 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
28133 (simplify_stmt_using_ranges): Call it.
28134 (vrp_dom_walker::before_dom_children): Extract equivalences
28135 from an ASSERT_EXPR with an equality comparison against a
28138 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28140 * lra-constraints.c (lra_copy_reg_equiv): New function.
28141 (split_reg): Use it to copy equivalence information from the
28142 original register to the spill register.
28144 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
28146 PR rtl-optimization/75964
28147 * simplify-rtx.c (simplify_const_relational_operation): Remove
28148 invalid handling of comparisons of integer ABS.
28150 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
28152 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
28153 initialize to zero.
28154 (init_regs): Remove declaration.
28155 (function_arg_advance_32): Initialize error_p as boolean variable.
28157 2017-05-05 Nathan Sidwell <nathan@acm.org>
28159 * store-motion.c (remove_reachable_equiv_notes): Reformat long
28160 lines. Use for (;;).
28162 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28164 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
28165 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
28166 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
28167 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
28168 VF=2 that require versioning.
28170 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28172 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
28175 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28177 * diagnostic.h (diagnostic_override_option_index): Convert from
28178 macro to inline function.
28180 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28182 * diagnostic.c (last_module_changed_p): New function.
28183 (set_last_module): New function.
28184 (diagnostic_report_current_module): Convert macro usage to
28185 the above functions.
28186 * diagnostic.h (diagnostic_context::last_module): Strengthen
28187 from const line_map * to const line_map_ordinary *.
28188 (diagnostic_last_module_changed): Delete macro.
28189 (diagnostic_set_last_module): Delete macro.
28191 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28193 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
28194 with diagnostic_report_diagnostic.
28195 (diagnostic_n_impl_richloc): Likewise.
28196 * diagnostic.h (report_diagnostic): Delete macro.
28197 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
28198 with diagnostic_report_diagnostic.
28199 * substring-locations.c (format_warning_va): Likewise.
28201 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28203 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
28204 save/restor of format_spec. Move option-printing code to...
28205 (print_option_information): ...this new function, and
28206 reimplement by simply printing to the pretty_printer,
28207 rather than appending to the format string.
28209 2017-05-05 David Malcolm <dmalcolm@redhat.com>
28211 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
28212 handling logic into...
28213 (update_effective_level_from_pragmas): ...this new function.
28215 2017-05-04 Andrew Waterman <andrew@sifive.com>
28217 * config/riscv/riscv.opt (mstrict-align): New option.
28218 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
28219 (SLOW_UNALIGNED_ACCESS): Define.
28220 (riscv_slow_unaligned_access): Declare.
28221 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
28223 (riscv_slow_unaligned_access): New variable.
28224 (rocket_tune_info): Set slow_unaligned_access to true.
28225 (optimize_size_tune_info): Set slow_unaligned_access to false.
28226 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
28227 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
28228 (riscv_option_override): Set riscv_slow_unaligned_access.
28229 * doc/invoke.texi: Add -mstrict-align to RISC-V.
28231 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
28233 * config/riscv/riscv.md: Unify indentation.
28235 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28240 * config/rs6000/rs6000.md (u code attribute): Add FIX and
28242 (extendsi<mode>2): Add support for doing sign extension via
28243 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
28244 don't have ISA 3.0 instructions.
28245 (extendsi<mode>2 splitter): Likewise.
28246 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
28247 generate the normal insns since SImode can now go in vector
28248 registers. Disallow the special UNSPECs needed for previous
28249 machines to hide SImode being used. Add new insns
28250 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
28251 (fix_trunc<mode>si2_stfiwx): Likewise.
28252 (fix_trunc<mode>si2_internal): Likewise.
28253 (fixuns_trunc<mode>si2): Likewise.
28254 (fixuns_trunc<mode>si2_stfiwx): Likewise.
28255 (fctiw<u>z_<mode>_smallint): Likewise.
28256 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
28257 of floating point to 32-bit integer from doing a direct move to
28258 the GPR registers to do a store.
28259 (fctiwz_<mode>): Break long line.
28261 2017-05-05 Bin Cheng <bin.cheng@arm.com>
28263 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
28264 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
28265 (addr_list, addr_offset_valid_p): New.
28266 (split_address_groups): Check offset validity with above function.
28267 (gt-tree-ssa-loop-ivopts.h): Include header file.
28269 2017-05-05 Nathan Sidwell <nathan@acm.org>
28271 * config.gcc (arm*-*-*): Add missing 'fi'.
28273 2017-05-05 Steve Ellcey <sellcey@cavium.com>
28275 * doc/invoke.texi (-fopt-info): Explicitly say order of options
28276 included in -fopt-info does not matter.
28277 * doc/optinfo.texi (-fopt-info): Fix description of default
28278 behavour. Explicitly say order of options included in -fopt-info
28281 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28283 * config.gcc: Allow combinations of aprofile and rmprofile values for
28284 --with-multilib-list.
28285 * config/arm/t-multilib: New file.
28286 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
28287 variables. Remove setting of ISA and floating-point ABI in
28288 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
28289 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
28290 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
28291 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
28293 * config/arm/t-rmprofile: Likewise except for the matches changes.
28294 * doc/install.texi (--with-multilib-list): Document the combination of
28295 aprofile and rmprofile values and warn about pitfalls in doing that.
28297 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28299 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
28300 (movdi_aarch64): Likewise.
28302 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28304 PR tree-optimization/80632
28305 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
28307 (build_arrays): Initialize it for virtual phis.
28308 (fix_phi_nodes): Use it for virtual phis.
28310 PR tree-optimization/80558
28311 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
28312 [x, y] op z into [x op, y op z] for op & or | if conditions
28315 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
28316 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28319 * config/arm/arm.md (use_literal_pool): Remove.
28320 (64-bit immediate split): No longer takes cost into consideration
28321 if arm_disable_literal_pool is enabled.
28322 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
28323 used when arm_disable_literal_pool is enabled.
28324 (arm_max_const_double_inline_cost): Remove use of
28325 arm_disable_literal_pool.
28326 (push_minipool_fix): Add assert.
28327 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
28328 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
28329 (no_literal_pool_sf_immediate): New.
28331 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28333 PR tree-optimization/80613
28334 * tree-ssa-dce.c (propagate_necessity): Remove cases for
28335 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
28337 2017-05-05 Richard Biener <rguenther@suse.de>
28339 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
28341 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
28343 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
28344 of this flag from insn conditions due to removal from r247495.
28346 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
28348 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
28350 (arm_early_store_addr_dep_ptr): Likewise.
28351 * config/arm/aarch-common-protos.h
28352 (arm_early_load_addr_dep_ptr): Add prototype.
28353 (arm_early_store_addr_dep_ptr): Likewise.
28354 * config/arm/cortex-a53.md: Add new bypasses.
28356 2017-05-05 Jakub Jelinek <jakub@redhat.com>
28358 * tree.c (next_type_uid): Change type to unsigned.
28359 (type_hash_canon): Decrement back next_type_uid if
28360 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
28361 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
28364 2017-05-04 Martin Sebor <msebor@redhat.com>
28366 * builtins.c: Fix a trivial typo in a comment.
28368 PR middle-end/79234
28369 * builtins.c (check_sizes): Adjust to handle reading past the end.
28370 Avoid printing excessive upper bound of ranges. Use %E to print
28371 tree nodes instead of converting them to %wu.
28372 (expand_builtin_memchr): New function.
28373 (compute_dest_size): Rename...
28374 (compute_objsize): ...to this.
28375 (expand_builtin_memcpy): Adjust.
28376 (expand_builtin_mempcpy): Adjust.
28377 (expand_builtin_strcat): Adjust.
28378 (expand_builtin_strcpy): Adjust.
28379 (check_strncat_sizes): Adjust.
28380 (expand_builtin_strncat): Adjust.
28381 (expand_builtin_strncpy): Adjust and simplify.
28382 (expand_builtin_memset): Adjust.
28383 (expand_builtin_bzero): Adjust.
28384 (expand_builtin_memcmp): Adjust.
28385 (expand_builtin): Handle memcmp.
28386 (maybe_emit_chk_warning): Check strncat just once.
28388 2017-05-04 Martin Sebor <msebor@redhat.com>
28390 PR preprocessor/79214
28391 PR middle-end/79222
28392 PR middle-end/79223
28393 * builtins.c (check_sizes): Add inlining context and issue
28394 warnings even when -Wno-system-headers is set.
28395 (check_strncat_sizes): Same.
28396 (expand_builtin_strncat): Same.
28397 (expand_builtin_memmove): New function.
28398 (expand_builtin_stpncpy): Same.
28399 (expand_builtin): Handle memmove and stpncpy.
28401 2017-05-04 Bin Cheng <bin.cheng@arm.com>
28403 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
28404 which is not used any more.
28406 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28408 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
28410 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28412 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
28413 (cortexa53_tunings): Likewise.
28414 (cortexa57_tunings): Likewise.
28415 (cortexa72_tunings): Likewise.
28416 (cortexa73_tunings): Likewise.
28418 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28420 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
28421 Set loop alignment to 8.
28423 2017-05-04 Martin Sebor <msebor@redhat.com>
28425 PR translation/80280
28426 * builtins.c (expand_builtin_object_size): Add missing quoting to
28427 %D and like directives.
28428 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
28429 (hsa_type_for_tree_type): Same.
28430 (verify_function_arguments): Same.
28431 * symtab.c (symbol_table::change_decl_assembler_name): Same.
28432 * varasm.c (get_section): Same.
28435 2017-05-04 Martin Sebor <msebor@redhat.com>
28437 PR translation/80280
28438 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
28440 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
28442 * config/aarch64/aarch64.c (generic_addrcost_table):
28443 Change HI/TI mode setting.
28445 2017-05-04 Martin Jambor <mjambor@suse.cz>
28447 PR tree-optimization/80622
28448 * tree-sra.c (comes_initialized_p): New function.
28449 (build_accesses_from_assign): Only set write lazily when
28450 comes_initialized_p is false.
28451 (analyze_access_subtree): Use comes_initialized_p.
28452 (propagate_subaccesses_across_link): Assert !comes_initialized_p
28453 instead of testing for PARM_DECL.
28455 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28457 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
28458 constraint on operand 0 to allow more general addressing modes.
28459 Adjust output template.
28460 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
28462 * config/aarch64/aarch64-protos.h
28463 (aarch64_address_valid_for_prefetch_p): Declare prototype.
28464 * config/aarch64/constraints.md (Dp): New address constraint.
28465 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
28468 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
28470 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
28471 update use of estimate_ipcp_clone_size_and_time.
28472 (estimate_local_effects): Update use of
28473 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
28474 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
28475 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
28476 Return nonspecialized time.
28478 2017-05-04 Richard Biener <rguenther@suse.de>
28480 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
28481 for the last VUSE which def dominates the PHI. Directly call
28483 (get_continuation_for_phi_1): Remove.
28485 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
28487 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
28488 to explain the use of truncating division. Cap the number of
28489 iterations to the maximum given by nb_iterations_upper_bound,
28492 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28494 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
28495 * configure: Regenerate.
28496 * config.in: Regenerate.
28497 * config/i386/driver-mingw32.c: new file.
28498 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
28499 * config.host: Link driver-mingw32.o on MinGW host.
28500 * doc/install.texi: Document new --enable-mingw-wildcard configure
28503 2017-05-04 Marek Polacek <polacek@redhat.com>
28505 PR tree-optimization/80612
28506 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
28508 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28509 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
28511 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
28512 (movt splitter): Likewise.
28513 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
28514 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
28515 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
28516 block for Thumb-1 with MOVT.
28517 (thumb2_legitimate_address_p): Move code block ...
28518 (can_avoid_literal_pool_for_label_p): ... into this new function.
28519 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
28521 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
28522 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
28523 "M-profile targets with the MOVT instruction".
28525 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
28527 * config/arm/arm-builtins.c (arm_init_builtins): Rename
28528 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
28529 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
28531 2017-05-04 Martin Liska <mliska@suse.cz>
28533 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
28534 variable cond_code.
28536 2017-05-04 Richard Biener <rguenther@suse.de>
28538 * tree.c (array_at_struct_end_p): Handle arrays at struct
28539 end with flexarrays more conservatively. Refactor and treat
28540 arrays of arrays or aggregates more strict. Fix
28541 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
28542 * tree.c (array_at_struct_end_p): Adjust prototype.
28543 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
28544 * gimple-fold.c (get_range_strlen): Likewise.
28545 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
28547 2017-05-04 Richard Biener <rguenther@suse.de>
28549 PR tree-optimization/31130
28550 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
28552 (supports_overflow_infinity): Likewise.
28553 (is_negative_overflow_infinity): Likewise.
28554 (is_positive_overflow_infinity): Likewise.
28555 (is_overflow_infinity): Likewise.
28556 (stmt_overflow_infinity): Likewise.
28557 (overflow_infinity_range_p): Likewise.
28558 (usable_range_p): Remove as always returning true.
28559 (make_overflow_infinity): Remove.
28560 (negative_overflow_infinity): Likewise.
28561 (positive_overflow_infinity): Likewise.
28562 (avoid_overflow_infinity): Likewise.
28563 (set_value_range): Adjust accordingly.
28564 (set_value_range_to_nonnegative): Likewise, remove now unused
28565 overflow_infinity arg.
28566 (vrp_operand_equal_p): Adjust.
28567 (update_value_range): Likewise.
28568 (range_int_cst_singleton_p): Likewise.
28569 (operand_less_p): Likewise.
28570 (compare_values_warnv): Likewise.
28571 (extract_range_for_var_from_comparison_expr): Likewise.
28572 (vrp_int_const_binop): Likewise.
28573 (zero_nonzero_bits_from_vr): Likewise.
28574 (extract_range_from_multiplicative_op_1): Likewise.
28575 (extract_range_from_binary_expr_1): Likewise.
28576 (extract_range_from_unary_expr): Likewise.
28577 (extract_range_from_comparison): Likewise.
28578 (extract_range_basic): Likewise.
28579 (adjust_range_with_scev): Likewise.
28580 (compare_ranges): Likewise.
28581 (compare_range_with_value): Likewise.
28582 (dump_value_range): Likewise.
28583 (test_for_singularity): Likewise, remove strict_overflow_p parameter
28585 (simplify_cond_using_ranges): Adjust.
28587 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28589 * brig-builtins.def: Added a builtin for class_f64.
28590 * builtin-types.def: Added a builtin type needed by class_f64.
28592 2017-05-03 Jason Merrill <jason@redhat.com>
28594 * timevar.def: Add TV_CONSTEXPR.
28596 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28598 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
28600 2017-05-03 Martin Jambor <mjambor@suse.cz>
28602 * ipa-prop.c (ipa_update_after_lto_read): Removed.
28603 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
28604 * ipa-cp.c (ipcp_propagate_stage): Do not call
28605 ipa_update_after_lto_read.
28606 * ipa-inline.c (ipa_inline): Likewise.
28608 2017-05-03 Martin Jambor <mjambor@suse.cz>
28610 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
28611 tag. Added a default constructor and a destructor.
28612 (ipa_edge_args_sum_t): New class;
28613 (ipa_edge_args_sum): Declare.
28614 (ipa_edge_args_vector): Remove declaration.
28615 (IPA_EDGE_REF): Use ipa_edge_args_sum.
28616 (ipa_free_edge_args_substructures): Remove declaration.
28617 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
28618 (ipa_edge_args_info_available_for_edge_p): Likewise.
28619 * ipa-prop.c (ipa_edge_args_vector): Removed.
28620 (edge_removal_hook_holder): Likewise.
28621 (edge_duplication_hook_holder): Likewise.
28622 (ipa_edge_args_sum): New variable.
28623 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
28624 ipa_edge_args_vector.
28625 (ipa_free_edge_args_substructures): Likewise.
28626 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
28627 ipa_edge_args_vector.
28628 (ipa_edge_removal_hook): Turned into method
28629 ipa_edge_args_sum_t::remove.
28630 (ipa_edge_duplication_hook): Turned into method
28631 ipa_edge_args_sum_t::duplicate.
28632 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
28633 registering edge hooks.
28634 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
28635 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
28636 ipa_edge_args_sum instead of ipa_edge_args_vector.
28637 * ipa-profile.c (ipa_profile): Likewise.
28639 2017-05-03 Martin Jambor <mjambor@suse.cz>
28641 * symbol-summary.h (function_summary): New method exists.
28642 (function_summary::symtab_removal): Deallocate through release.
28643 (call_summary): New class.
28644 (gt_ggc_mx): New overload.
28645 (gt_pch_nx): Likewise.
28646 (gt_pch_nx): Likewise.
28648 2017-05-03 Jeff Law <law@redhat.com>
28650 PR tree-optimization/78496
28651 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
28652 from simplify_cond_using_ranges. Split off code to walk
28653 backwards through casts into ...
28654 (simplify_cond_using_ranges_2): New function.
28655 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
28656 (execute_vrp): After identifying jump threads, call
28657 simplify_cond_using_ranges_2.
28659 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
28662 * ipa-inline.h (inline_summary): Add ctor.
28663 (create_ggc): Do not use ggc_cleared_alloc.
28665 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
28666 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28668 * gcc.c (handle_braces): Support escaping in switch matching
28670 * doc/invoke.texi (Spec Files): Document it.
28671 Remove superfluous @code markup in items.
28673 2017-05-03 David Malcolm <dmalcolm@redhat.com>
28675 * diagnostic-show-locus.c (struct column_range): New struct.
28676 (get_affected_columns): New function.
28677 (get_printed_columns): New function.
28678 (struct correction): New struct.
28679 (correction::ensure_capacity): New function.
28680 (correction::ensure_terminated): New function.
28681 (struct line_corrections): New struct.
28682 (line_corrections::~line_corrections): New dtor.
28683 (line_corrections::add_hint): New function.
28684 (layout::print_trailing_fixits): Reimplement in terms of the new
28686 (selftest::test_overlapped_fixit_printing): New function.
28687 (selftest::diagnostic_show_locus_c_tests): Call it.
28689 2017-05-03 Nathan Sidwell <nathan@acm.org>
28691 Canonicalize canonical type hashing
28692 * tree.h (type_hash_canon_hash): Declare.
28693 * tree.c (type_hash_list, attribute_hash_list): Move into
28694 type_hash_canon_hash.
28695 (build_type_attribute_qual_variant): Break out hash code calc into
28696 type_hash_canon_hash.
28697 (type_hash_canon_hash): New. Generic type hash computation.
28698 (build_range_type_1, build_array_type_1, build_function_type,
28699 build_method_type_directly, build_offset_type, build_complex_type,
28700 make_vector_type): Call it.
28702 2017-05-03 Richard Biener <rguenther@suse.de>
28704 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
28705 When all DRs have unknown misaligned do not always peel
28706 when there is a store but apply the same costing model as if
28707 there were only loads.
28709 2017-05-03 Richard Biener <rguenther@suse.de>
28712 PR tree-optimization/80492
28713 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
28714 compare_base_decls returning dont-know properly.
28716 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
28718 * config/arm/iterators.md (CCSI): New mode iterator.
28719 (arch): New mode attribute.
28720 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
28721 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
28722 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
28723 code iterator for success result mode.
28724 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
28725 the corresponding new insn generators.
28727 2017-05-03 Bin Cheng <bin.cheng@arm.com>
28730 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28731 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28733 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
28735 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
28736 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
28737 (DDR_A): Wrap DDR argument in brackets.
28738 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
28739 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
28740 (DDR_REVERSED_P): Likewise.
28742 2017-05-03 Jakub Jelinek <jakub@redhat.com>
28744 PR tree-optimization/79472
28745 * tree-switch-conversion.c (struct switch_conv_info): Add
28746 contiguous_range and default_case_nonstandard fields.
28747 (collect_switch_conv_info): Compute contiguous_range and
28748 default_case_nonstandard fields, don't clear final_bb if
28749 contiguous_range and only the default case doesn't have the required
28751 (check_all_empty_except_final): Set default_case_nonstandard instead
28752 of failing if contiguous_range and the default case doesn't have empty
28754 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
28755 and only the default case doesn't have the required constants. Skip
28757 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
28758 if default_case_nonstandard.
28759 (build_constructors): Build constant 1 just once. Assert that default
28760 values aren't inserted in between cases if contiguous_range. Skip
28762 (build_arrays): Skip virtual phis.
28763 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
28764 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
28765 Handle virtual phis.
28766 (gen_inbound_check): Handle default_case_nonstandard case.
28767 (process_switch): Adjust check_final_bb caller. Call
28768 gather_default_values with the first non-default case instead of
28769 default case if default_case_nonstandard.
28771 2017-05-02 Nathan Sidwell <nathan@acm.org>
28773 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
28774 check. Fix formatting.
28776 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
28778 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
28779 errors when comparing specialized and unspecialized times.
28781 2017-05-02 David Malcolm <dmalcolm@redhat.com>
28783 * diagnostic-show-locus.c
28784 (layout::should_print_annotation_line_p): Make private.
28785 (layout::print_annotation_line): Make private.
28786 (layout::annotation_line_showed_range_p): Make private.
28787 (layout::show_ruler): Make private.
28788 (layout::print_source_line): Make private. Pass in line and
28789 line_width, rather than calling location_get_source_line. Drop
28791 (layout::print_leading_fixits): New method.
28792 (layout::print_any_fixits): Rename to...
28793 (layout::print_trailing_fixits): ...this, and make private.
28794 Don't print newline fixits.
28795 (diagnostic_show_locus): Move logic for printing one row into...
28796 (layout::print_line): ...this new function. Move the
28797 location_get_source_line call and error-handling from
28798 print_source_line to here. Call print_leading_fixits, and rename
28799 print_any_fixits to print_trailing_fixits.
28800 (selftest::test_fixit_insert_containing_newline): Update now that
28801 newlines are partially supported.
28802 (selftest::test_fixit_insert_containing_newline_2): New test.
28803 (selftest::test_fixit_replace_containing_newline): Update comments.
28804 (selftest::diagnostic_show_locus_c_tests): Call the new test.
28805 * edit-context.c (class added_line): New class.
28806 (class edited_line): Describe newline handling in comment.
28807 (edited_line::actually_edited_p): New method.
28808 (edited_line::print_content): Delete redundant decl.
28809 (edited_line::m_predecessors): New field.
28810 (edited_file::print_content): Call edited_line::print_content.
28811 (edited_file::print_diff): Update to support newlines.
28812 (edited_file::print_diff_hunk): Likewise.
28813 (edited_file::print_run_of_changed_lines): New function.
28814 (edited_file::print_diff_line): Convert to...
28815 (print_diff_line): ...this.
28816 (edited_file::get_effective_line_count): New function.
28817 (edited_line::edited_line): Initialize new field m_predecessors.
28818 (edited_line::~edited_line): Clean up m_predecessors.
28819 (edited_line::apply_fixit): Handle newlines.
28820 (edited_line::get_effective_line_count): New function.
28821 (edited_line::print_content): New function.
28822 (edited_line::print_diff_lines): New function.
28823 (selftest::test_applying_fixits_insert_containing_newline): New
28825 (selftest::test_applying_fixits_replace_containing_newline): New
28827 (selftest::insert_line): New function.
28828 (selftest::test_applying_fixits_multiple_lines): Add example of
28830 (selftest::edit_context_c_tests): Call the new tests.
28832 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28834 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
28835 parameter cand. Update dump information.
28836 (get_computation_cost): Update uses.
28838 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28840 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
28841 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
28842 (get_computation_at, rewrite_use_address): Update use of
28843 get_computation_aff.
28845 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28847 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
28848 (get_computation): Delete.
28849 (get_computation_cost): Implement like get_computation_cost_at.
28850 Use get_computation_at.
28851 (get_computation_cost_at): Delete.
28852 (rewrite_use_nonlinear_expr): Use get_computation_at.
28853 (rewrite_use_compare, remove_unused_ivs): Ditto.
28855 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28857 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
28859 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28861 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
28862 (ivopts_global_cost_for_size): Rename parameter and update uses.
28863 (iv_ca_recount_cost): Update uses.
28864 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
28865 candidates seperately in n_invs and n_cands.
28866 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
28868 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28870 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
28871 (find_inv_vars_cb): New.
28872 (find_depends): Renamed to ...
28873 (find_inv_vars): ... this.
28874 (add_candidate_1, force_var_cost): Call find_inv_vars.
28875 (split_address_cost, determine_group_iv_cost_cond): Ditto.
28877 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28879 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
28880 inv_vars. Add inv_exprs.
28881 (struct iv_cand): Rename depends_on to inv_vars.
28882 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
28883 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
28884 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
28885 (dump_cand): Dump inv_vars.
28886 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
28887 (record_invariant, find_depends, add_candidate_1): Ditto.
28888 (set_group_iv_cost, force_var_cost): Ditto.
28889 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
28890 (get_computation_cost_at, get_computation_cost): Ditto.
28891 (determine_group_iv_cost_generic): Ditto.
28892 (determine_group_iv_cost_address): Ditto.
28893 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
28894 (determine_group_iv_costs): Ditto.
28895 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
28896 (iv_ca_set_remove_invariants): Renamed to ...
28897 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
28898 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
28899 (iv_ca_set_add_invariants): Renamed to ...
28900 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
28901 (iv_ca_set_cp): Use iv_ca_set_add_invs.
28902 (iv_ca_has_deps): Support inv_vars and inv_exprs.
28903 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
28904 (create_new_ivs): Remove useless dump.
28906 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28908 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
28910 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
28911 (iv_ca_set_no_cp, create_new_iv): Ditto.
28913 2017-05-02 Bin Cheng <bin.cheng@arm.com>
28915 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
28917 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
28919 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
28920 function tree_check2.
28922 2017-05-02 Martin Liska <mliska@suse.cz>
28924 * doc/gcov.texi: Add missing preposition.
28925 * gcov.c (function_info::function_info): Properly fill up
28926 all member variables.
28928 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28930 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
28932 2017-05-02 Tamar Christina <tamar.christina@arm.com>
28934 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
28936 2017-05-02 Martin Liska <mliska@suse.cz>
28939 * lto-streamer-in.c (lto_read_tree_1): Remove
28940 LTO_STREAMER_DEBUG.
28941 * lto-streamer.c (struct tree_hash_entry): Likewise.
28942 (struct tree_entry_hasher): Likewise.
28943 (tree_entry_hasher::hash): Likewise.
28944 (tree_entry_hasher::equal): Likewise.
28945 (lto_streamer_init): Likewise.
28946 (lto_orig_address_map): Likewise.
28947 (lto_orig_address_get): Likewise.
28948 (lto_orig_address_remove): Likewise.
28949 * lto-streamer.h: Likewise.
28950 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
28951 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
28953 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
28955 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
28956 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
28957 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
28958 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
28959 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
28960 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
28961 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
28962 (mm_maskz_sub_ss): New intrinsics.
28963 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28964 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
28965 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
28966 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
28967 (__builtin_ia32_subss_mask_round): New builtins.
28968 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
28969 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
28970 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
28972 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
28973 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
28975 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
28978 2017-05-02 Martin Jambor <mjambor@suse.cz>
28980 PR tree-optimization/78687
28981 * tree-sra.c (access): New field parent.
28982 (process_subtree_disqualification): New function.
28983 (disqualify_candidate): Call it.
28984 (build_accesses_from_assign): Reset write flag if creating an
28986 (build_access_subtree): Fill in parent field and also prpagate
28987 down grp_write flag.
28988 (create_artificial_child_access): New parameter set_grp_write, set
28989 grp_write to its value.
28990 (propagate_subaccesses_across_link): Also propagate grp_write flag
28992 (propagate_all_subaccesses): Push the closest parent back to work
28993 queue if add_access_to_work_queue returned true.
28995 2017-05-02 Richard Biener <rguenther@suse.de>
28997 * common.opt (fstrict-overflow): Alias negative to fwrapv.
28998 * doc/invoke.texi (fstrict-overflow): Remove all traces of
28999 -fstrict-overflow documentation.
29000 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
29001 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
29002 flag_strict_overflow.
29003 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
29004 * lto-opts.c (lto_write_options): Do not stream it.
29005 * lto-wrapper.c (merge_and_complain): Do not handle it.
29006 * opts.c (default_options_table): Do not set -fstrict-overflow.
29007 (finish_options): Likewise do not clear it when sanitizing.
29008 * simplify-rtx.c (simplify_const_relational_operation): Do not
29009 test flag_strict_overflow.
29011 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29013 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
29014 using enabled attribute.
29015 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
29016 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
29017 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
29018 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
29019 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
29020 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
29021 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
29022 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
29023 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
29024 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
29026 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
29028 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
29030 2017-05-02 Richard Biener <rguenther@suse.de>
29032 PR tree-optimization/80591
29034 2017-04-10 Richard Biener <rguenther@suse.de>
29036 * tree-ssa-structalias.c (find_func_aliases): Properly handle
29039 2017-05-02 Richard Biener <rguenther@suse.de>
29041 PR tree-optimization/80549
29042 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
29043 (cleanup_tree_cfg_noloop): Create forwarders to known loop
29044 headers if they do not have a preheader.
29046 2017-05-02 Martin Liska <mliska@suse.cz>
29049 * common.opt: Fix typo.
29050 * doc/invoke.texi: Likewise.
29052 2017-05-01 Jan Beulich <jbeulich@suse.com>
29054 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
29055 swapping, add (x,x,m,x,n) alternative.
29057 2017-05-01 Nathan Sidwell <nathan@acm.org>
29059 * calls.c (combine_pending_stack_adjustment_and_call): Remove
29060 unnecessary unadjusted_alignment check.
29062 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
29065 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
29067 * gimplify.c (gimplify_cilk_detach): New function.
29068 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
29069 * tree-core.h: Document EXPR_CILK_SPAWN.
29070 * tree.h (EXPR_CILK_SPAWN): Define.
29072 2017-05-01 David Malcolm <dmalcolm@redhat.com>
29074 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
29075 to use new fixit_hint representation, using the "replace" logic.
29076 (get_line_span_for_fixit_hint): Likewise.
29077 (layout::print_any_fixits): Likewise.
29078 (selftest::test_one_liner_many_fixits): Rename to...
29079 (selftest::test_one_liner_many_fixits_1): ...this, and update
29080 comment and expected output to reflect that the multiple fix-it
29081 hints are now consolidated into one insertion.
29082 (selftest::test_one_liner_many_fixits_2): New test.
29083 (selftest::test_diagnostic_show_locus_one_liner): Update for
29085 (selftest::test_fixit_consolidation): Update for fix-it API
29087 * diagnostic.c (print_parseable_fixits): Likewise.
29088 * edit-context.c (edited_line::m_line_events): Convert from
29089 auto_vec <line_event *> to auto_vec <line_event>.
29090 (class line_event): Convert from abstract base class to a concrete
29091 class, taking over the role of replace_event.
29092 (class insert_event): Delete.
29093 (class replace_event): Rename to class line_event. Convert to
29095 (edit_context::add_fixits): Reimplement.
29096 (edit_context::apply_insert): Delete.
29097 (edit_context::apply_replace): Rename to...
29098 (edit_context::apply_fixit): ...this. Convert to half-open range.
29099 (edited_file::apply_insert): Delete.
29100 (edited_file::apply_replace): Rename to...
29101 (edited_file::apply_fixit): ...this.
29102 (edited_line::~edited_line): Drop deletion of events.
29103 (edited_line::apply_insert): Delete.
29104 (edited_line::apply_replace): Rename to...
29105 (edited_line::apply_fixit): ...this. Convert to half-open range.
29106 Update for change to type of m_line_events.
29107 * edit-context.h (edit_context::apply_insert): Delete.
29108 (edit_context::apply_replace): Rename to...
29109 (edit_context::apply_fixit): ...this.
29111 2017-05-01 Martin Sebor <msebor@redhat.com>
29113 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
29116 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
29119 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
29120 __get_cpuid_max returns 0.
29121 (__get_cpuid_count): Ditto.
29123 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
29125 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
29126 replacement expression is another instance of one of its arguments.
29128 2017-05-01 Jakub Jelinek <jakub@redhat.com>
29131 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
29132 check for stack push/pop autoinc.
29133 * config/i386/i386.c (ix86_agi_dependent): Return false
29134 if the only reason why modified_in_p returned true is that
29135 addr is SP based and set_insn is a push or pop.
29137 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29139 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
29142 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
29145 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
29146 (account_size_time): Use two predicates - exec_pred and
29148 (evaluate_conditions_for_known_args): Compute both clause and
29150 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
29151 (inline_summary_t::duplicate): Update.
29152 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
29154 (compute_inline_parameters): Likewise.
29155 (estimate_edge_size_and_time): Update caluclation of time.
29156 (estimate_node_size_and_time): Compute both time and nonspecialized
29158 (estimate_ipcp_clone_size_and_time): Update.
29159 (inline_merge_summary): Update.
29160 (do_estimate_edge_time): Update.
29161 (do_estimate_edge_size): Update.
29162 (do_estimate_edge_hints): Update.
29163 (inline_read_section, inline_write_summary): Stream both new predicates.
29164 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
29166 (compute_inlined_call_time): Cleanup.
29167 (big_speedup_p): Update.
29168 (edge_badness): Update.
29169 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
29170 (size_time_entry): Replace predicate by exec_predicate and
29171 nonconst_predicate.
29172 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
29173 (estimate_edge_time): Return also nonspec_time.
29174 (reset_edge_growth_cache): Update.
29176 2017-04-29 Jakub Jelinek <jakub@redhat.com>
29178 PR rtl-optimization/80491
29179 * ifcvt.c (noce_process_if_block): When looking for x setter
29180 with missing else_bb, don't check only the insn right before
29181 cond_earliest, but look for the last insn that x is modified in
29182 within the same bb.
29184 PR rtl-optimization/80491
29185 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
29187 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
29189 PR tree-optimization/80487
29190 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
29192 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29194 PR tree-optimization/79697
29195 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
29196 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
29197 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
29199 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
29200 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
29202 2017-04-28 Martin Sebor <msebor@redhat.com>
29204 PR tree-optimization/80523
29205 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
29206 (init_target_to_host_charmap, target_to_host, target_strtol10): New
29208 (maybe_warn, format_directive, parse_directive): Use new functions.
29209 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
29211 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
29213 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
29215 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
29217 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
29218 target_header_dir): Set correctly.
29219 * configure: Regenerated.
29220 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29221 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29222 instead of SYSTEM_HEADER_DIR.
29224 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
29226 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
29227 (estimate_local_effects): Likewise.
29228 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
29229 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
29230 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
29231 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
29232 do_estimate_edge_time, estimate_edge_time): Likewise.
29233 * ipa-inline-analysis.c (estimate_node_size_and_time,
29234 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
29235 (estimate_time_after_inlining): Remove.
29237 2017-04-28 Martin Liska <mliska@suse.cz>
29239 * doc/gcov.texi: Enhance documentation of gcov.
29241 2017-04-28 Martin Liska <mliska@suse.cz>
29243 * doc/gcov.texi: Sort options in alphabetic order.
29244 * doc/gcov-dump.texi: Likewise.
29245 * doc/gcov-tool.texi: Likewise.
29246 * gcov.c (print_usage): Likewise.
29247 * gcov-dump.c (print_usage): Likewise.
29248 * gcov-tool.c (print_merge_usage_message): Likewise.
29249 (print_rewrite_usage_message): Likewise.
29250 (print_overlap_usage_message): Likewise.
29252 2017-04-28 Martin Liska <mliska@suse.cz>
29254 PR gcov-profile/53915
29255 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
29257 2017-04-28 Martin Liska <mliska@suse.cz>
29259 PR gcov-profile/79891
29260 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
29261 is marked by compiler as living on a line.
29262 (get_cycles_count): Remove usage of the union.
29263 (output_intermediate_file): Likewise.
29264 (find_source): Fix GNU coding style.
29265 (accumulate_line_counts): Remove old non-all block mode.
29266 (output_lines): Remove usage of the union.
29267 * profile.c (output_location): Include all BBs, even if
29268 belonging to a same line (and file) as a previous BB.
29270 2017-04-28 Martin Liska <mliska@suse.cz>
29272 * gcov.c (process_args): Handle new argument 'w'.
29273 (read_graph_file): Assign ID to BBs.
29274 (output_branch_count): Display BB # if verbose flag is set.
29275 (output_lines): Likewise for arcs.
29276 (print_usage): Add '--verbose' option help.
29277 * doc/gcov.texi: Document --verbose (-w) option.
29279 2017-04-28 Martin Liska <mliska@suse.cz>
29281 * gcov.c (struct block_location_info): New struct.
29282 (process_file): Fill up the new structure.
29283 (read_graph_file): Replace usage of encoding by the newly added
29285 (add_line_counts): Likewise.
29286 (accumulate_line_counts): Remove usage of the union.
29287 (function_info::function_info): New function.
29288 (function_info::~function_info): Likewise.
29289 (process_file): Call delete instead of release_function.
29290 (release_function): Release the function.
29291 (release_structures): Call delete instead of release_function.
29292 (solve_flow_graph): Replace usage of num_blocks.
29293 (find_exception_blocks): Likewise.
29294 (output_lines): Fix GNU coding style.
29296 2017-04-28 Martin Liska <mliska@suse.cz>
29299 * coverage.c (coverage_remove_note_file): New function.
29300 * coverage.h: Declare the function.
29301 * toplev.c (finalize): Clean if an error has been seen.
29303 2017-04-28 Martin Liska <mliska@suse.cz>
29305 PR gcov-profile/80031
29306 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
29307 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
29308 * gcov.c (read_graph_file): Read just number of blocks.
29309 * profile.c (branch_prob): Do not stream 0 flags per a basic
29312 2017-04-28 Martin Liska <mliska@suse.cz>
29314 * gcov-dump.c (tag_*): Add new argument to declarations.
29315 (dump_gcov_file): Likewise.
29316 (tag_blocks): Add and use new argument depth.
29317 (tag_arcs): Likewise.
29318 (tag_lines): Likewise.
29319 (tag_counters): Likewise.
29320 (tag_summary): Likewise.
29321 (dump_working_sets): Use depth to do a proper indentation.
29323 2017-04-28 Jakub Jelinek <jakub@redhat.com>
29326 * cgraph.h (symtab_node::debug_symtab): No longer inline.
29327 * symtab.c (symtab_node::debug_symtab): Move definition here.
29329 2017-04-28 Richard Biener <rguenther@suse.de>
29331 * lto-streamer.h (LTO_major_version): Bump to 7.
29333 2017-04-28 Richard Biener <rguenther@suse.de>
29335 * tree-vrp.c (assert_info): New struct.
29336 (add_assert_info): New helper.
29337 (register_edge_assert_for_2): Refactor to add asserts to a vector
29339 (register_edge_assert_for_1): Likewise.
29340 (register_edge_assert_for): Likewise.
29341 (finish_register_edge_assert_for): New helper actually registering
29342 asserts where live on edge.
29343 (find_conditional_asserts): Adjust.
29344 (find_switch_asserts): Likewise.
29345 (evrp_dom_walker::try_find_new_range): Generalize.
29346 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
29348 2017-04-27 Marek Polacek <polacek@redhat.com>
29351 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
29352 arg10 and arg11 to itype.
29354 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
29356 * doc/extend.texi (Object Size Checking): Improve grammar.
29358 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
29361 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
29362 that the logic for permitting reciprocal estimates matches that
29365 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29368 * tree.c (type_cache_hasher::equal): Only compare
29369 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
29370 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
29371 non-aggregate element types.
29372 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
29373 about the flag on ARRAY_TYPEs in the comment, formatting fix.
29375 2017-04-27 Richard Biener <rguenther@suse.de>
29377 PR middle-end/80533
29378 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
29379 stripping ARRAY_REFs from MEM_EXPR make sure we're not
29380 keeping a reference to a trailing array.
29382 2017-04-27 Richard Biener <rguenther@suse.de>
29384 PR middle-end/80539
29385 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
29386 being in loop-closed SSA form conservatively.
29387 (chrec_fold_multiply_poly_poly): Likewise.
29389 2017-04-27 Tamar Christina <tamar.christina@arm.com>
29391 PR middle-end/79665
29392 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
29393 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
29395 2017-04-27 Jakub Jelinek <jakub@redhat.com>
29398 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
29399 (aarch64_function_arg_alignment): Return unsigned int again, but still
29400 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
29401 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29402 Don't emit -Wpsabi note.
29403 (aarch64_function_arg_boundary): Likewise.
29404 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29407 2017-04-26 Nathan Sidwell <nathan@acm.org>
29409 * tree.h (crc32_unsigned_n): Declare.
29410 (crc32_unsigned, crc32_unsigned): Make inline.
29411 * tree.c (crc32_unsigned_bits): Replace with ...
29412 (crc32_unsigned_n): ... this.
29413 (crc32_unsigned, crc32_byte): Remove.
29414 (crc32_string): Remove unnecessary braces.
29416 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29418 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
29419 * ipa-inline-analysis.c (MAX_TIME): Remove.
29420 (account_size_time): Use sreal for time.
29421 (dump_inline_summary): Update.
29422 (estimate_function_body_sizes): Update.
29423 (estimate_edge_size_and_time): Update.
29424 (estimate_calls_size_and_time): Update.
29425 (estimate_node_size_and_time): Update.
29426 (inline_merge_summary): Update.
29427 (inline_update_overall_summary): Update.
29428 (estimate_time_after_inlining): Update.
29429 (inline_read_section): Update.
29430 (inline_write_summary): Update.
29431 * ipa-inline.c (compute_uninlined_call_time): Update.
29432 (compute_inlined_call_time): Update.
29433 (recursive_inlining): Update.
29434 (inline_small_functions): Update.
29435 (dump_overall_stats): Update.
29436 * ipa-inline.h: Include sreal.h.
29437 (size_time_entry): Turn time to sreal.
29438 (inline_summary): Turn self_time nad time to sreal.
29440 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
29442 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
29444 (sreal::stream_out, sreal::stream_in): New.
29445 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
29447 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29449 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
29452 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
29455 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
29457 (dimode_scalar_chain::compute_convert_gain): Ditto.
29458 (dimode_scalar_chain::make_vector_copies): Ditto.
29459 (dimode_scalar_chain::convert_reg): Ditto.
29460 (dimode_scalar_chain::convert_insn): Ditto.
29461 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
29462 (VI248_AVX512BW_1): New mode iterator.
29463 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
29464 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
29467 2017-04-25 Martin Sebor <msebor@redhat.com>
29469 PR tree-optimization/80497
29470 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
29471 constants are representable in HOST_WIDE_INT.
29472 (parse_directive): Ditto.
29474 2017-04-25 Martin Sebor <msebor@redhat.com>
29477 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
29478 (new_zero_array): Adjust signature.
29479 (dom_info::dom_init): Used unsigned rather that size_t.
29480 (dom_info::dom_info): Same.
29482 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29483 Jakub Jelinek <jakub@redhat.com>
29486 * config/arm/arm.c: Include gimple.h.
29487 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29488 returns negative, increment ncrn only if it returned positive.
29489 (arm_needs_doubleword_align): Return int instead of bool,
29490 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
29491 members, but if there is any such non-FIELD_DECL
29492 > PARM_BOUNDARY aligned decl, return -1 instead of false.
29493 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
29494 returns negative, increment nregs only if it returned positive.
29495 (arm_setup_incoming_varargs): Likewise.
29496 (arm_function_arg_boundary): Emit -Wpsabi note if
29497 arm_needs_doubleword_align returns negative, return
29498 DOUBLEWORD_ALIGNMENT only if it returned positive.
29500 2017-04-25 Marek Polacek <polacek@redhat.com>
29503 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
29504 first argument to type.
29506 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
29509 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
29510 type checks to test for compatibility instead of equality.
29512 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
29513 Jakub Jelinek <jakub@redhat.com>
29516 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
29518 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
29519 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
29520 the alignment computation, but return their maximum in warn_alignment.
29521 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
29522 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
29524 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
29525 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
29528 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29530 * config/arc/simdext.md (dmpyh): Fix typo.
29532 2017-04-25 Richard Biener <rguenther@suse.de>
29534 PR tree-optimization/80492
29535 * alias.c (compare_base_decls): Handle registers with asm
29536 specification conservatively.
29537 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
29538 compare_base_decls returning dont-know properly.
29540 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29542 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
29543 (legitimate_offset_address_p): New function.
29544 (arc_legitimate_address_p): Use above function.
29546 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29548 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
29550 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29552 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
29553 ACCH registers whenever they are available.
29555 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29557 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
29558 double regs fix when not used.
29560 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29562 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
29564 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
29565 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
29567 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29569 * config/arc/arc.c (arc_output_addsi): Check for h-register class
29570 when emitting short ADD instructions.
29572 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
29574 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
29576 (cmpsi_cc_c_insn): Likewise.
29577 (cbranchsi4_scratch): Compute proper instruction length using
29578 compact_hreg_operand.
29579 * config/arc/predicates.md (compact_hreg_operand): New predicate.
29581 2017-04-25 Richard Biener <rguenther@suse.de>
29583 PR middle-end/80509
29584 * passes.c (pass_manager::pass_manager): Initialize
29585 m_name_to_pass_map.
29587 2017-04-25 Richard Biener <rguenther@suse.de>
29589 PR tree-optimization/79201
29590 * tree-ssa-sink.c (statement_sink_location): Handle calls.
29592 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29595 * config/s390/vector.md: Split MEM->GPR vector moves for
29596 non-s_operand addresses.
29598 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29601 * config/s390/predicates.md (reload_const_wide_int_operand): New
29603 * config/s390/s390.md ("movti"): Remove d/P alternative.
29604 ("movti_bigconst"): New pattern definition.
29606 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29609 * s390-protos.h (s390_expand_cs_hqi): Removed.
29610 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
29611 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
29612 modes as well as CCZ1mode and CCZmode.
29613 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
29614 signature of s390_emit_compare_and_swap.
29615 (s390_expand_cs_hqi): Likewise, make static.
29616 (s390_expand_cs_tdsi): Generate an explicit compare before trying
29617 compare-and-swap, in some cases.
29618 (s390_expand_cs): Wrapper function.
29619 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
29621 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
29622 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
29623 patterns for small and large integers. Forbid symref memory operands.
29624 Move expander to s390.c. Require cc register.
29625 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
29626 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
29627 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
29628 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
29629 symref memory operands. Remove CC mode and call s390_match_ccmode
29631 ("atomic_exchange<mode>"): Allow and implement all integer modes.
29633 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29635 * config/s390/s390.md (define_peephole2): New peephole to help
29636 combining the load-and-test pattern with volatile memory.
29638 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
29640 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
29641 with CCZmode for TARGET_Z196.
29643 2017-04-25 Jakub Jelinek <jakub@redhat.com>
29645 PR rtl-optimization/80501
29646 * combine.c (make_compound_operation_int): Set subreg_code to SET
29647 even for AND with mask of the sign bit of mode.
29649 PR rtl-optimization/80500
29650 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
29651 sum's initial value.
29653 2017-04-25 Julian Brown <julian@codesourcery.com>
29654 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29656 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
29658 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
29660 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
29662 2017-04-25 Julian Brown <julian@codesourcery.com>
29663 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29665 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
29666 (thunderx2t99_sha): New Reservation.
29668 2017-04-25 Julian Brown <julian@codesourcery.com>
29669 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29671 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
29672 type for 1-element load.
29674 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
29676 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
29678 2017-04-24 Martin Jambor <mjambor@suse.cz>
29680 PR tree-optimization/80293
29681 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
29682 char arrays not totally scalarizable if it is false.
29683 (analyze_all_variable_accesses): Pass correct value in the new
29684 parameter. Add a statistics counter.
29686 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
29688 PR middle-end/79931
29689 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
29691 2017-04-24 Richard Biener <rguenther@suse.de>
29693 PR tree-optimization/80494
29694 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
29695 out for complex types.
29697 2017-04-24 Richard Biener <rguenther@suse.de>
29699 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
29700 * tree-ssa-sccvn.c (print_scc): Print SCC size.
29701 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
29702 (DFS): Adjust and never fail.
29703 (sccvn_dom_walker::fail): Remove.
29704 (sccvn_dom_walker::before_dom_children): Adjust.
29705 (run_scc_vn): Likewise and never fail.
29706 * tree-ssa-pre.c (pass_pre::execute): Adjust.
29707 (pass_fre::execute): Likewise.
29709 2017-04-24 Richard Biener <rguenther@suse.de>
29711 PR tree-optimization/79725
29712 * tree-ssa-sink.c (statement_sink_location): Return whether
29713 failure reason was zero uses. Move that check later.
29714 (sink_code_in_bb): Deal with zero uses by removing the stmt
29717 2017-04-24 Richard Biener <rguenther@suse.de>
29720 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
29721 pointer-based references.
29723 2017-04-24 Richard Biener <rguenther@suse.de>
29726 * pass_manager.h (pass_manager::operator new): Remove.
29727 (pass_manager::operator delete): Likewise.
29728 * passes.c (pass_manager::operator new): Remove.
29729 (pass_manager::operator delete): Likewise.
29730 (pass_manager::pass_manager): Zero individual pass members.
29732 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
29735 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
29736 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
29737 Check "XEXP (src, 1)" operand here.
29738 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
29739 Check "XEXP (src, 1)" operand here.
29740 (dimode_scalar_chain::make_vector_copies): Detect count register
29741 of a shift instruction. Zero extend count register from QImode
29742 to DImode to satisfy vector shift pattern count operand predicate.
29743 Substitute vector shift count operand with a DImode copy.
29744 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
29747 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
29749 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
29750 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29751 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
29752 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
29753 (UNSPEC_NOREX_MEM): Remove definition.
29755 2017-04-21 Richard Biener <rguenther@suse.de>
29757 PR tree-optimization/79547
29758 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
29759 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
29760 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
29761 without any constraints.
29763 2017-04-21 Richard Biener <rguenther@suse.de>
29765 PR tree-optimization/78847
29766 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
29768 2017-04-21 Richard Biener <rguenther@suse.de>
29770 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
29771 (build_distinct_type_copy): Likewise.
29772 (build_variant_type_copy): Likewise.
29773 * tree.c (build_qualified_type): Pass down mem-stat info.
29774 (build_distinct_type_copy): Likewise.
29775 (build_variant_type_copy): Likewise.
29777 2017-04-21 Richard Biener <rguenther@suse.de>
29779 PR tree-optimization/80237
29780 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
29782 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
29783 for a simplified result.
29785 2016-04-21 Richard Biener <rguenther@suse.de>
29787 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
29788 sth as strict as a simple_iv but a chrec without symbols and an
29789 operand defined in the loop we are peeling (and not some subloop).
29790 (propagate_constants_for_unrolling): Propagate all constants.
29792 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29795 * config/i386/i386.c (print_reg): Remove assert for disalowed
29796 regno values, call output_operand_lossage instead.
29798 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
29801 * config/i386/constraints.md (Yc): New register constraint.
29802 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
29803 Use Yc constraint for alternative 2 of operand 0. Remove
29804 preferred_for_speed attribute.
29806 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29808 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
29809 lastprivate clauses in SIMT case.
29811 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
29813 * doc/invoke.texi (-Wextra-semi): Document new warning option.
29815 2017-04-20 Richard Biener <rguenther@suse.de>
29817 PR tree-optimization/57796
29818 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
29819 as N scalar stores.
29820 (vect_model_load_cost): Cost gathers as N scalar loads.
29822 2017-04-20 Richard Biener <rguenther@suse.de>
29824 * ggc-page.c (ggc_allocated_p): Rename to ...
29825 (safe_lookup_page_table_entry): ... this and return the lookup
29827 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
29829 2017-04-20 Richard Biener <rguenther@suse.de>
29831 PR tree-optimization/80453
29832 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
29833 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
29834 from the conditions.
29835 (vn_phi_eq): Pass them down.
29836 (vn_phi_lookup): Record them.
29837 (vn_phi_insert): Likewise.
29839 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
29841 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
29842 uninitialized variable warning to avoid buffer overrun.
29844 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
29847 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
29848 is suppressed for '{ 0 }' in C.
29850 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29852 * BASE-VER: Set to 8.0.0.
29854 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
29856 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
29857 priority .init_array and .fini_array section with SECTION_NOTYPE
29860 2017-04-20 Jakub Jelinek <jakub@redhat.com>
29862 PR middle-end/80423
29863 * tree.h (build_array_type): Add typeless_storage default argument.
29864 * tree.c (type_cache_hasher::equal): Also compare
29865 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
29866 (build_array_type): Add typeless_storage argument, set
29867 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
29869 (build_nonshared_array_type): Adjust build_array_type_1 caller.
29870 (build_array_type): Likewise. Add typeless_storage argument.
29872 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29873 Jakub Jelinek <jakub@redhat.com>
29875 PR tree-optimization/80426
29876 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
29877 operation on symbolic operands, also compute the overflow for the
29878 invariant part when the operation degenerates into a negation.
29880 2017-04-19 Jakub Jelinek <jakub@redhat.com>
29883 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
29884 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
29887 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
29889 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
29892 * config/avr/avr.c (tree.h): Include it.
29893 (cgraph.h): Include it.
29894 (avr_encode_section_info): Don't warn for uninitialized progmem
29895 variable if it's just an alias.
29897 2017-04-19 Richard Biener <rguenther@suse.de>
29900 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
29901 when needed by AutoPGO.
29903 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
29906 * doc/lto.texi: Remove an extra 'that'.
29908 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
29910 PR rtl-optimization/80429
29911 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
29912 are only used in debug insns.
29914 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
29915 Vladimir Makarov <vmakarov@redhat.com>
29917 * config/sparc/predicates.md (input_operand): Add comment. Return
29918 true for any memory operand when LRA is in progress.
29919 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
29921 2017-04-18 Jeff Law <law@redhat.com>
29924 * mips.md ({return,simple_return}_internal): Do not overwrite
29927 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29929 PR tree-optimization/80443
29930 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
29931 instead of adding 1, subtract -1 and similarly instead of subtracting
29934 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
29936 PR rtl-optimization/80357
29937 * haifa-sched.c (tmp_bitmap): New variable.
29938 (model_recompute): Handle duplicate use records.
29939 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
29940 (free_global_sched_pressure_data): Free it.
29942 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
29945 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29946 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29947 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29948 instead of SYSTEM_HEADER_DIR.
29950 2017-04-18 Jeff Law <law@redhat.com>
29952 PR middle-end/80422
29953 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
29954 predecessors after walking up the insn chain.
29956 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29959 * dwarf2out.c (modified_type_die): Try harder not to emit internal
29960 sizetype type into debug info.
29962 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29965 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
29966 unneeded test for TARGET_UPPER_REGS_SF.
29967 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
29969 2017-04-18 Jakub Jelinek <jakub@redhat.com>
29972 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
29973 instead of gsi_after_labels.
29975 2017-04-18 Jeff Law <law@redhat.com>
29977 * regcprop.c (maybe_mode_change): Avoid creating copies of the
29981 2017-04-13 Jeff Law <law@redhat.com>
29982 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
29983 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
29985 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
29988 * config/avr/avr.c (intl.h): Include it.
29989 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
29991 2017-04-18 Martin Liska <mliska@suse.cz>
29993 PR gcov-profile/78783
29994 * gcov-tool.c (gcov_output_files): Validate that destination
29995 file is either removed by the tool or by a user.
29997 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
29998 Guy Benyei <guybe@mellanox.com>
30000 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
30001 block, and do not negate it, the stored id is already negative.
30003 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
30005 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
30007 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
30010 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
30011 masks of options that should be turned off if the VSX vector
30012 options are turned off.
30013 (OTHER_P8_VECTOR_MASKS): Likewise.
30014 (OTHER_VSX_VECTOR_MASKS): Likewise.
30015 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
30016 rs6000_disable_incompatible_switches to validate no type switches
30018 (rs6000_incompatible_switch): New function to disallow turning on
30019 other vector options if -mno-vsx, -mno-power8-vector, or
30020 -mno-power9-vector are specified.
30022 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30024 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
30026 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30028 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
30029 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
30030 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
30031 (ARG_POINTER_CFA_OFFSET): Likewise.
30033 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
30035 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
30036 conditions to take advantage of various optimizations.
30038 2017-04-13 Jeff Law <law@redhat.com>
30040 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
30041 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
30042 (zero_extendsidi2_dext): Likewise.
30044 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30047 * fold-const.c (fold_ternary_loc): Revert
30048 use op0 instead of fold_convert_loc (loc, type, arg0) part of
30051 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
30053 PR rtl-optimization/80343
30054 * lra-remat.c (update_scratch_ops): Assign original hard reg to
30055 new scratch pseudo.
30057 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
30060 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
30061 to ubsan_encode_value.
30063 2017-04-13 Jeff Law <law@redhat.com>
30065 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
30066 appearing in DEBUG_INSNs.
30068 2017-04-13 Martin Liska <mliska@suse.cz>
30070 PR gcov-profile/80413
30071 * gcov-io.c (gcov_write_string): Copy to buffer just when
30072 allocated size is greater than zero.
30074 2017-04-13 Jakub Jelinek <jakub@redhat.com>
30077 * dwarf2out.c (decls_for_scope): Ignore declarations of
30078 current_function_decl in BLOCK_NONLOCALIZED_VARS.
30080 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
30083 * ipa-visibility.c (non_local_p): Fix typos.
30084 (localize_node): When localizing symbol in same comdat group,
30085 dissolve the group only when we know external symbols are going
30087 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
30089 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30091 PR tree-optimization/79390
30092 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
30093 order does not result in usable sequence, retry with reversed operand
30099 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
30100 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
30101 op0 instead of fold_convert_loc (loc, type, arg0).
30103 2017-04-12 Jeff Law <law@redhat.com>
30105 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
30106 has a delay slot in the generated code.
30108 * config/cris/cris.md (cris_preferred_reload_class): Return
30109 GENNONACR_REGS rather than GENERAL_REGS.
30111 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30114 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
30115 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
30116 signedness of the result type.
30118 2017-04-12 Richard Biener <rguenther@suse.de>
30119 Jeff Law <law@redhat.com>
30121 PR tree-optimization/80359
30122 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
30123 trim stores to TARGET_MEM_REFs.
30125 2017-04-12 Richard Biener <rguenther@suse.de>
30127 PR tree-optimization/79390
30128 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
30129 threading case even more.
30131 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
30134 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
30135 for quad_address_p for TImode, instead of just not indexed_address.
30137 2017-04-12 Richard Biener <rguenther@suse.de>
30138 Bernd Edlinger <bernd.edlinger@hotmail.de>
30140 PR middle-end/79671
30141 * alias.c (component_uses_parent_alias_set_from): Handle
30142 TYPE_TYPELESS_STORAGE.
30143 (get_alias_set): Likewise.
30144 * tree-core.h (tree_type_common): Add typeless_storage flag.
30145 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
30146 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
30147 for types containing members with TYPE_TYPELESS_STORAGE.
30148 (place_field): Likewise.
30149 (layout_type): Likewise for ARRAY_TYPE.
30150 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
30151 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
30152 TYPE_TYPELESS_STORAGE.
30153 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
30155 2017-04-12 Jakub Jelinek <jakub@redhat.com>
30158 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
30159 first argument to type.
30161 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30165 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
30166 CONST0_RTX (mode) rather than const0_rtx where appropriate.
30167 (rs6000_expand_binop_builtin): Likewise.
30168 (rs6000_expand_ternop_builtin): Likewise; also add missing
30169 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
30170 vshasigma built-ins.
30171 * doc/extend.texi: Document that vec_xxpermdi's third argument
30172 must be a constant.
30174 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
30176 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
30177 Use shift_const cost parameter when calculating gain of STV shifts.
30179 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
30181 PR rtl-optimization/70478
30182 * lra-constraints.c (process_alt_operands): Check memory for
30183 disfavoring memory insn operand.
30185 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30187 PR middle-end/80100
30188 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
30189 left shift in unsigned HOST_WIDE_INT type.
30191 PR rtl-optimization/80385
30192 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
30193 (not (neg X)) into (plus X -1) for complex or non-integral modes.
30196 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
30197 if they have any depend clauses.
30199 2017-04-11 Martin Liska <mliska@suse.cz>
30202 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
30203 * ipa-split.c (split_function): Create a local comdat symbol
30204 if caller is in a comdat group.
30206 2017-04-11 Martin Liska <mliska@suse.cz>
30209 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
30212 2017-04-11 Martin Sebor <msebor@redhat.com>
30214 PR middle-end/80364
30215 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
30216 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
30218 (directive::set_width, directive::set_precision, format_character):
30220 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
30223 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
30226 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
30227 conflict, set target->arch_name instead of target->cpu_name.
30229 2017-04-11 Richard Biener <rguenther@suse.de>
30231 PR tree-optimization/80374
30232 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
30233 build_zero_cst, remove fold_convertible_p check again.
30235 2017-04-11 Martin Liska <mliska@suse.cz>
30238 * ubsan.c (instrument_object_size): Do not instrument register
30241 2017-04-11 Jakub Jelinek <jakub@redhat.com>
30244 * config/i386/i386-builtin-types.def
30245 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
30246 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
30247 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
30248 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
30249 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
30250 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
30251 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
30252 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
30253 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
30254 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
30255 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
30256 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
30257 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
30258 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
30259 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
30260 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
30261 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
30262 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
30263 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
30264 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
30265 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
30266 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
30267 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
30268 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
30269 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
30270 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
30271 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
30272 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
30273 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
30274 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
30275 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
30276 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
30277 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
30278 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
30279 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
30280 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
30281 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
30282 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
30283 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
30284 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
30285 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
30286 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
30287 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
30288 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
30289 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
30290 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
30292 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
30293 flag to second_arg_count, handle 4 argument function type _COUNT
30294 aliases, handle second_arg_count on second argument rather than last.
30296 2017-04-10 Jeff Law <law@redhat.com>
30298 PR tree-optimization/80374
30299 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
30300 record anything if we can not convert integer_zero_node to the
30303 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30306 * config/rs6000/rs6000.c (rs6000_option_override_internal):
30307 Enhance special handling given to the TARGET_P9_MINMAX option in
30308 relation to certain other options.
30310 2017-04-10 Bin Cheng <bin.cheng@arm.com>
30312 PR tree-optimization/80153
30313 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
30314 remove POINTER_PLUS_EXPR's base part directly, rather than through
30317 2017-04-10 Richard Biener <rguenther@suse.de>
30318 Bin Cheng <bin.cheng@arm.com>
30320 PR tree-optimization/80153
30321 * tree-affine.c (aff_combination_to_tree): Get base pointer from
30322 the first element of pointer type aff_tree. Build result expr in
30324 (add_elt_to_tree): Convert to type unconditionally. Remove other
30325 fold_convert calls.
30326 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
30327 (rewrite_use_nonlinear_expr): Check invariant using iv information.
30329 2017-04-10 Richard Biener <rguenther@suse.de>
30331 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30334 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
30336 PR rtl-optimization/70478
30337 * lra-constraints.c (curr_small_class_check): New.
30338 (update_and_check_small_class_inputs): New.
30339 (process_alt_operands): Update curr_small_class_check. Disfavor
30340 alternative insn memory operands. Check available regs for small
30343 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30346 * config/mips/mips.opt (-mvirt): Update description.
30347 * doc/invoke.texi (-mvirt): Likewise.
30349 2017-04-10 Richard Biener <rguenther@suse.de>
30351 PR middle-end/80362
30352 * fold-const.c (fold_binary_loc): Look at unstripped ops when
30353 looking for NEGATE_EXPR in -A / -B to A / B folding.
30355 2017-04-10 Martin Liska <mliska@suse.cz>
30357 PR gcov-profile/80224
30358 * gcov.c (print_usage): Fix usage string.
30359 (get_gcov_intermediate_filename): Remove.
30360 (output_gcov_file): Use both for normal and intermediate format.
30361 (generate_results): Do not initialize special file for
30362 intermediate format.
30364 2017-04-10 Richard Biener <rguenther@suse.de>
30366 PR tree-optimization/80304
30367 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
30370 2017-04-10 Nathan Sidwell <nathan@acm.org>
30373 * config/rs6000/rs6000.c (rs6000_vector_type): New.
30374 (rs6000_init_builtins): Use it.
30376 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30378 * config/arm/arm.md (<mrc>): Add mode to SET source.
30379 (<mrrc>): Likewise.
30381 2017-04-10 Richard Biener <rguenther@suse.de>
30383 PR middle-end/80344
30384 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
30386 2017-04-10 Jakub Jelinek <jakub@redhat.com>
30389 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
30390 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
30391 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
30392 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
30393 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
30394 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
30395 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
30396 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
30397 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
30398 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
30399 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
30400 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
30401 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
30402 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
30403 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
30404 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
30405 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
30406 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
30407 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
30408 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
30409 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
30410 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
30411 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
30413 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
30415 PR rtl-optimization/70478
30416 * lra-constraints.c: Reverse the last patch.
30418 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
30420 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
30421 Add comment for WCHAR_T.
30423 2017-04-08 Martin Liska <mliska@suse.cz>
30426 2017-04-07 Martin Liska <mliska@suse.cz>
30429 * ipa-split.c (split_function): Add function part to a same comdat
30432 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30435 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
30437 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
30439 * rs6000/rs6000.c (vec_load_pendulum): Rename...
30440 (vec_pairing): ...to this.
30441 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
30442 (rs6000_sched_init): Adjust for name change.
30443 (struct rs6000_sched_context): Likewise.
30444 (rs6000_init_sched_context): Likewise.
30445 (rs6000_set_sched_context): Likewise.
30447 2017-04-07 Jakub Jelinek <jakub@redhat.com>
30453 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
30455 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
30456 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
30457 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
30459 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
30461 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
30463 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30465 PR rtl-optimization/70703
30466 * ira-color.c (update_conflict_hard_regno_costs): Use
30467 int64_t instead of HOST_WIDE_INT.
30469 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
30471 PR rtl-optimization/70478
30472 * lra-constraints.c (process_alt_operands): Disfavor alternative
30473 insn memory operands.
30475 2017-04-07 Jeff Law <law@redhat.com>
30477 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
30478 CALL and NOTE_INSN_CALL_ARG_LOCATION.
30480 2017-04-07 Martin Liska <mliska@suse.cz>
30483 * config/aarch64/aarch64.c (aarch64_process_target_attr):
30484 Show error message instead of an ICE.
30486 2017-04-07 Martin Liska <mliska@suse.cz>
30489 * ipa-split.c (split_function): Add function part to a same comdat
30492 2017-04-07 Richard Biener <rguenther@suse.de>
30494 PR middle-end/80341
30495 * tree.c (get_unwidened): Also handle ! for_type case for
30497 * convert.c (do_narrow): Split out from ...
30498 (convert_to_integer_1): ... here. Do not pass final truncation
30499 type to get_unwidened for TRUNC_DIV_EXPR.
30501 2017-04-07 Richard Biener <rguenther@suse.de>
30503 * tree-affine.c (wide_int_ext_for_comb): Take type rather
30505 (aff_combination_const): Adjust.
30506 (aff_combination_scale): Likewise.
30507 (aff_combination_add_elt): Likewise.
30508 (aff_combination_add_cst): Likewise.
30509 (aff_combination_convert): Likewise.
30510 (add_elt_to_tree): Likewise. Remove unused argument.
30511 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
30513 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
30515 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
30517 * config/arm/arm.c (arm_default_short_enums): Use
30518 ARM_DEFAULT_SHORT_ENUMS.
30519 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
30521 2017-04-06 Jakub Jelinek <jakub@redhat.com>
30524 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
30525 members with redundant out-of-class redeclaration.
30527 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30530 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
30531 * config/i386/i386.md (*zero_extendsidi2):
30532 Add (?*x,*x) and (?*v,*v) alternatives.
30534 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
30537 * config/i386/i386.c (ix86_expand_builtin)
30538 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
30539 mode from insn data. Convert operands to insn operand mode.
30540 Copy operands that don't satisfy insn predicate to a register.
30542 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
30544 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
30547 2017-04-06 Richard Biener <rguenther@suse.de>
30549 PR tree-optimization/80334
30550 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
30551 preserve alignment of accesses.
30553 2017-04-06 Richard Biener <rguenther@suse.de>
30555 PR tree-optimization/80262
30556 * tree-sra.c (build_ref_for_offset): Preserve address-space
30558 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
30559 Drop useless address-space information on MEM_REF offsets.
30561 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
30563 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
30565 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30567 PR rtl-optimization/70703
30568 * ira-color.c (update_conflict_hard_regno_costs): Use
30569 HOST_WIDE_INT instead of long.
30571 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
30574 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
30575 not defined for x86_64 target. Add -mmmx target option when __SSE2__
30577 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
30578 for x86_64 target. Handle -m3dnowa option.
30580 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
30582 PR rtl-optimization/70703
30583 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
30584 (update_conflict_hard_regno_costs): Use long instead of unsigned
30585 arithmetic for cost calculation.
30587 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30588 Bernd Edlinger <bernd.edlinger@hotmail.de>
30591 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
30594 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
30597 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
30598 ptr_mode with Pmode throughout.
30599 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
30600 into probe_stack_range and use DImode.
30602 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
30605 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
30606 call_eh_return is true.
30608 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30610 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
30611 Initialize last_match_fntype_index.
30613 2017-04-05 Jakub Jelinek <jakub@redhat.com>
30616 * tree-nvr.c: Include internal-fn.h.
30617 (pass_return_slot::execute): Ignore internal calls without
30620 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30621 Richard Biener <rguenther@suse.de>
30624 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
30625 captures used multiple times, except for the last use.
30626 * generic-match-head.c: Include gimplify.h.
30628 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30630 PR tree-optimization/79390
30631 * target.h (struct noce_if_info): Declare.
30632 * targhooks.h (default_noce_conversion_profitable_p): Declare.
30633 * target.def (noce_conversion_profitable_p): New target hook.
30634 * ifcvt.h (struct noce_if_info): New type, moved from ...
30635 * ifcvt.c (struct noce_if_info): ... here.
30636 (noce_conversion_profitable_p): Renamed to ...
30637 (default_noce_conversion_profitable_p): ... this. No longer
30639 (noce_try_store_flag_constants, noce_try_addcc,
30640 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
30641 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
30642 instead of noce_conversion_profitable_p.
30643 * config/i386/i386.c: Include ifcvt.h.
30644 (ix86_option_override_internal): Don't override
30645 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
30646 (ix86_noce_conversion_profitable_p): New function.
30647 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
30648 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
30649 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
30650 * doc/tm.texi: Regenerated.
30652 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30654 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
30657 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
30660 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
30661 instructions for small multiply cores.
30663 2017-04-04 Jeff Law <law@redhat.com>
30665 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
30667 (mips_expand_vec_perm_const): Initialize elements in orig_perm
30668 that are not set by the loop over the elements.
30670 2017-04-04 Jakub Jelinek <jakub@redhat.com>
30673 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
30674 int mode, convert_modes it to mode as unsigned, otherwise use
30675 lowpart_subreg to mode rather than SImode.
30676 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
30677 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
30678 Use DImode instead of SImode for the shift count operand.
30679 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
30682 2017-04-04 Richard Biener <rguenther@suse.de>
30684 PR middle-end/80281
30685 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
30686 arithmetic done for the negate or the plus. Simplify.
30687 (A - (-B) -> A + B): Likewise.
30688 * fold-const.c (split_tree): Make sure to not negate pointers.
30690 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
30692 PR rtl-optimization/60818
30693 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
30694 a compare of comparisons with the thing compared if this results
30695 in a different machine mode.
30697 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
30699 * alias.c (base_alias_check): Fix typo in comment.
30700 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
30701 * cgraphunit.c (symbol_table::compile): Likewise.
30702 * collect2.c (maybe_run_lto_and_relink): Likewise.
30703 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
30704 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
30705 * config/avr/avr.c (avr_map_op_t): Likewise.
30706 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
30707 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
30708 * config/epiphany/epiphany.md (movcc): Likewise.
30709 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
30710 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
30712 * config/mips/mips.c (mips_save_restore_reg): Likewise.
30713 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
30714 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
30715 * config/sh/sh.c (sh_rtx_costs): Likewise.
30716 * fold-const.c (fold_truth_andor): Likewise.
30717 * genautomata.c (collapse_flag): Likewise.
30718 * gengtype.h (struct type::u::s): Likewise.
30719 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
30720 * input.c (FORMAT_AMOUNT): Likewise.
30721 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
30722 (known_aggs_to_agg_replacement_list): Likewise.
30723 * ipa-inline-analysis.c: Likewise.
30724 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
30725 * ipa-polymorphic-call.c
30726 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
30727 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
30728 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
30730 * modulo-sched.c (apply_reg_moves): Likewise.
30731 * omp-expand.c (build_omp_regions_1): Likewise.
30732 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
30733 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
30734 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
30735 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
30736 * value-prof.c: Likewise.
30737 * var-tracking.c (val_reset): Likewise.
30739 2017-04-03 Richard Biener <rguenther@suse.de>
30741 PR tree-optimization/80275
30742 * fold-const.c (split_address_to_core_and_offset): Handle
30745 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
30747 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
30748 descriptors is at least equal to that of functions.
30750 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30752 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
30754 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
30757 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
30758 (mov<IMOD4:mode>): New expander.
30759 (*mov<IMOD4:mode>_internal): New insn and split pattern.
30761 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
30763 PR rtl-optimization/79405
30764 * fwprop.c (propagations_left): New variable.
30765 (forward_propagate_into): Decrement it.
30766 (fwprop_init): Initialize it.
30767 (fw_prop): If the variable has reached zero, stop propagating.
30768 (fwprop_addr): Ditto.
30770 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30773 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
30774 a FUNCTION_DECL, pass it as decl instead of origin to
30777 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
30779 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
30782 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
30785 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
30786 TARGET_VSX_SMALL_INTEGER.
30788 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30790 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30791 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
30793 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
30795 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
30796 extraction from odd-numbered MSA register.
30798 2017-03-31 Jakub Jelinek <jakub@redhat.com>
30800 PR middle-end/80173
30801 * expmed.c (store_bit_field_1): Don't attempt to create
30802 a word subreg out of hard registers wider than word if they
30803 have HARD_REGNO_NREGS of 1 for their mode.
30805 PR middle-end/80163
30806 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
30807 conversions to integer types wider than word and pointer.
30810 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
30811 (rtx_equal_for_cselib_p): Pass 0 to it.
30812 * cselib.c (cselib_hasher::equal): Likewise.
30813 (rtx_equal_for_cselib_1): Add depth argument. If depth
30814 is 128, don't look up VALUE locs and punt. Increment
30815 depth in recursive calls when walking VALUE locs.
30817 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
30819 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
30820 (make_gcov_file_name): Use the canonical path name for generating
30822 (read_line): Fix handling of files with ascii null bytes.
30824 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
30826 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
30827 to initialise a vector register instead
30828 of using a const_int.
30830 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30832 PR translation/80189
30833 * gimplify.c (omp_default_clause): Use %qs instead of %s in
30834 diagnostic messages.
30836 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
30839 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
30840 (dfp_diex_<mode>): Update mode of operand 1.
30841 * doc/extend.texi (dxex, dxexq): Document change to return type.
30842 (diex, diexq): Document change to argument type.
30844 2017-03-30 Martin Jambor <mjambor@suse.cz>
30847 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
30848 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
30849 it reflects the signature changes performed at the callee side.
30850 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
30851 to cgraph_build_function_type_skip_args.
30852 (build_function_decl_skip_args): Adjust call to the above function.
30854 2017-03-30 Jakub Jelinek <jakub@redhat.com>
30857 * config/i386/sse.md
30858 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
30859 register as dest whenever it is a MEM not rtx_equal_p to the
30860 corresponding dup operand, and when forcing into reg move the
30861 reg into the memory afterwards.
30862 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
30863 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
30864 for the force_reg mode.
30865 (avx512vl_vextractf128<mode>): Use register as dest either
30866 always when a MEM, or when it is a MEM not rtx_equal_p to the
30867 corresponding dup operand, or even not when it is a CONST_VECTOR
30868 depending on the mode and lo vs. hi.
30869 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
30871 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
30872 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
30873 Likewise. Require that operands[2] is even.
30874 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
30875 Remove extraneous parens. Require that operands[2] is a multiple
30877 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
30878 operands[0] is a MEM if <mask_applied>, the predicates/constraints
30879 disallow memory then.
30881 2017-03-30 Richard Biener <rguenther@suse.de>
30883 PR tree-optimization/77498
30884 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
30885 to non-constants over backedges.
30887 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
30889 PR rtl-optimization/80233
30890 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
30891 as last_combined_insn. Do not test for BARRIER_P separately.
30893 2017-03-29 Andreas Schwab <schwab@suse.de>
30896 * calls.c (prepare_call_address): Convert funexp to Pmode before
30897 copying to temp reg.
30899 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30901 PR tree-optimization/80158
30902 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
30903 Handle possible future case of more than one alternate
30905 (replace_rhs_if_not_dup): Likewise.
30906 (replace_one_candidate): Likewise.
30908 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
30910 PR rtl-optimization/80193
30911 * ira.c (ira): Do not check allocation for LRA.
30913 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
30915 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
30916 (nvptx_output_simt_exit): Declare.
30917 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
30918 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
30919 (init_softstack_frame): Move initialization of crtl->is_leaf to...
30920 (nvptx_declare_function_name): ...here. Emit declaration of local
30921 memory space buffer for omp_simt_enter insn.
30922 (nvptx_output_unisimt_switch): New.
30923 (nvptx_output_softstack_switch): New.
30924 (nvptx_output_simt_enter): New.
30925 (nvptx_output_simt_exit): New.
30926 * config/nvptx/nvptx.h (struct machine_function): New fields
30927 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
30928 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
30929 (UNSPECV_SIMT_EXIT): Ditto.
30930 (omp_simt_enter_insn): New insn.
30931 (omp_simt_enter): New expansion.
30932 (omp_simt_exit): New insn.
30933 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
30935 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
30936 (expand_GOMP_SIMT_ENTER_ALLOC): New.
30937 (expand_GOMP_SIMT_EXIT): New.
30938 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
30939 (GOMP_SIMT_ENTER_ALLOC): Ditto.
30940 (GOMP_SIMT_EXIT): Ditto.
30941 * target-insns.def (omp_simt_enter): New insn.
30942 (omp_simt_exit): Ditto.
30943 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
30945 (lower_rec_simd_input_clauses): Implement SIMT privatization.
30946 (lower_rec_input_clauses): Likewise.
30947 (lower_lastprivate_clauses): Handle SIMT privatization.
30949 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
30950 (ompdevlow_adjust_simt_enter): New.
30951 (find_simtpriv_var_op): New.
30952 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
30953 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
30955 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
30956 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
30957 (copy_decl_for_dup_finish): Ditto.
30959 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
30961 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
30964 * config/i386/i386.c (ix86_option_override_internal): Always
30965 allow -mpreferred-stack-boundary=3 for 64-bit targets.
30967 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30969 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
30971 2017-03-28 Bin Cheng <bin.cheng@arm.com>
30973 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
30974 mark new edge's irreducible flag accordign to it.
30975 (vect_do_peeling): Check loop preheader edge's irreducible flag
30976 and pass it to function slpeel_add_loop_guard.
30978 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
30980 PR tree-optimization/80218
30981 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30982 Update block frequencies and counts.
30984 2017-03-28 Richard Biener <rguenther@suse.de>
30986 PR tree-optimization/78644
30987 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
30988 of a simplification result we may not use it at all.
30990 2017-03-28 Richard Biener <rguenther@suse.de>
30993 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
30994 without arguments, generate default definition of a SSA name.
30996 2017-03-28 Richard Biener <rguenther@suse.de>
30998 PR middle-end/80222
30999 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
31000 TYPE_REF_CAN_ALIAS_ALL references.
31001 * fold-const.c (fold_indirect_ref_1): Likewise.
31003 2017-03-28 Martin Liska <mliska@suse.cz>
31006 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
31007 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
31009 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31010 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
31012 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
31013 (EXTRA_SPECS): Define.
31014 (SUBTARGET_EXTRA_SPECS): Likewise.
31015 (SUBTARGET_CPP_SPEC): Likewise.
31016 * config/arc/elf.h (EXTRA_SPECS): Renamed to
31017 SUBTARGET_EXTRA_SPECS.
31018 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
31020 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
31022 * config/arc/simdext.md (vst64_insn): Update pattern.
31023 (vld32wh_insn): Likewise.
31024 (vld32wl_insn): Likewise.
31025 (vld64_insn): Likewise.
31026 (vld32_insn): Likewise.
31028 2017-03-28 Marek Polacek <polacek@redhat.com>
31031 * fold-const.c (fold_comparison): Use protected_set_expr_location
31032 instead of SET_EXPR_LOCATION.
31034 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
31036 * tree.c (add_expr): Avoid name lookup warning.
31038 2017-03-27 Jeff Law <law@redhat.com>
31040 PR tree-optimization/80216
31041 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
31042 function name. Limit recursion depth.
31043 (record_temporary_equivalences): Corresponding changes.
31045 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
31047 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
31050 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31053 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
31055 * cfgcleanup.c (reg_note_cfa_p): New array.
31056 (insns_have_identical_cfa_notes): New function.
31057 (old_insns_match_p): Don't cross-jump in between /f
31058 and non-/f instructions. If both i1 and i2 are frame related,
31059 verify all CFA notes, their order and content.
31061 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
31064 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
31065 HImode and SImode with zero extend to DImode to one insn.
31066 (bswap<mode>2_extenddi): Likewise.
31067 (bswapsi2_extenddi): Likewise.
31068 (bswaphi2_extendsi): Likewise.
31069 (bswaphi2): Combine bswap HImode and SImode into one insn.
31070 Separate memory insns from swapping register.
31071 (bswapsi2): Likewise.
31072 (bswap<mode>2): Likewise.
31073 (bswaphi2_internal): Delete, no longer used.
31074 (bswapsi2_internal): Likewise.
31075 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
31076 store, and gpr<-gpr swap insns.
31077 (bswap<mode>2_store): Likewise.
31078 (bswaphi2_reg): Register only splitter, combine with the splitter.
31079 (bswaphi2 splitter): Likewise.
31080 (bswapsi2_reg): Likewise.
31081 (bswapsi2 splitter): Likewise.
31082 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
31083 the insns into load, store, and register/register insns.
31084 (bswapdi2_ldbrx): Likewise.
31085 (bswapdi2_load): Likewise.
31086 (bswapdi2_store): Likewise.
31087 (bswapdi2_reg): Likewise.
31089 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
31091 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
31092 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
31094 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
31097 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
31099 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31100 special handling for target option conflicts between dform
31101 options (-mpower9-dform, -mpower9-dform-vector,
31102 -mpower9-dform-scalar) and -mno-direct-move.
31104 2017-03-27 Richard Biener <rguenther@suse.de>
31106 PR tree-optimization/80181
31107 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
31109 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31111 * config/arc/predicates.md (move_double_src_operand): Replace the
31112 call to move_double_src_operand with a call to address_operand.
31114 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31116 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
31117 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
31118 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
31120 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31122 * config/arc/predicates.md (long_immediate_loadstore_operand):
31123 Consider scaled addresses cases.
31125 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
31127 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
31128 restored when in interrupt.
31129 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
31130 doesn't have delay slot.
31132 2017-03-27 Richard Biener <rguenther@suse.de>
31135 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
31136 inlined thunk clones.
31138 2017-03-27 Jakub Jelinek <jakub@redhat.com>
31141 * asan.c (instrument_derefs): Copy over last operand from
31142 original COMPONENT_REF to the new COMPONENT_REF with
31143 DECL_BIT_FIELD_REPRESENTATIVE.
31144 * ubsan.c (instrument_object_size): Likewise.
31146 2017-03-27 Richard Biener <rguenther@suse.de>
31148 PR tree-optimization/80170
31149 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
31150 sure DR/SCEV didnt fold in constants we do not see when looking
31151 at the reference base alignment.
31153 2017-03-27 Richard Biener <rguenther@suse.de>
31155 PR middle-end/80171
31156 * gimple-fold.c (fold_ctor_reference): Properly guard against
31157 NULL return value from canonicalize_constructor_val.
31159 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
31162 * config/i386/i386.c (ix86_expand_builtin)
31163 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
31164 flags reg setting and flags reg using instructions.
31165 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
31166 clobbering instructions to zero extend op2.
31168 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
31170 * doc/install.texi (Configuration) <--with-aix-soname>:
31171 Update link to AIX ld.
31173 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
31175 PR rtl-optimization/80160
31176 PR rtl-optimization/80159
31177 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
31178 reg_alternate_class into account.
31180 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
31183 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
31184 to consider in curr_insn_transform.
31186 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31188 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
31189 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
31190 and emit_mode_inner.
31192 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31194 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
31195 argument to the overloaded builtin variants. Use the new flag to
31196 deprecate certain builtin variants.
31197 * config/s390/s390-builtin-types.def: Add new builtin types.
31198 * config/s390/s390-builtins.h: Support new flags field for
31199 overloaded builtins.
31200 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
31201 (s390_macro_to_expand): Enable vector float data type.
31202 (s390_cpu_cpp_builtins_internal): Indicate support of the new
31203 builtins by incrementing the __VEC__ version number.
31204 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
31206 (s390_resolve_overloaded_builtin): Emit error messages depending
31207 on the builtin flags.
31208 * config/s390/s390.c (s390_expand_builtin): Support additional
31209 flags argument. Change error message to match the messages
31210 emitted in s390-c.c.
31211 * config/s390/s390.md: New UNSPEC_* constants.
31212 (op_type): Add new instruction types.
31213 * config/s390/vecintrin.h: Add new builtins and test data class
31215 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
31216 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
31217 (VEC_INEXACT, VEC_NOINEXACT): New constants.
31218 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
31219 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
31220 ("vec_mergel<mode>"): V_HW -> VEC_HW.
31222 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
31223 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
31224 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
31225 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
31227 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
31228 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
31229 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
31230 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
31232 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
31233 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
31234 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
31235 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
31236 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
31237 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
31238 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
31240 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
31241 ("vec_scatter_element<V_HW_4:mode>_DI")
31242 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
31243 ("vec_fpint<mode>", "vflls")
31244 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
31245 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
31246 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
31247 ("*vec_cmphe<mode>_cc"): ... these.
31249 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
31250 mode constant instead of magic value.
31252 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31254 * config/s390/s390.c (s390_expand_vec_compare): Support other
31255 vector floating point modes than just V2DF.
31256 (s390_expand_vcond): Likewise.
31257 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
31258 (s390_cannot_change_mode_class): Prevent mode changes between TF
31259 and V1TF in vector registers.
31260 * config/s390/s390.md (DF, SF): New mode attributes.
31261 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
31262 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
31263 SFmode support for VRs.
31264 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
31266 (VFT, VF_HW): New mode iterators.
31267 (vw, sdx): New mode attributes.
31268 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
31269 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
31270 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
31271 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
31272 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
31273 also the new vector floating point modes. Renaming to ...
31275 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
31276 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
31277 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
31278 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
31279 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
31280 ("vec_unordered<mode>"): ... these.
31282 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
31283 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
31284 ("*vec_extendv2df"): New insn definitions.
31286 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31288 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
31289 ("mulditi3_2", "*muldi3_sign"): New patterns.
31290 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
31291 rename the pattern definition.
31293 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31295 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
31297 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
31299 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31301 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
31302 instruction if possible.
31303 * config/s390/vector.md (vec_halfnumelts): New mode
31305 ("*vec_vllezlf<mode>"): New pattern.
31307 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31309 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
31310 ("popcountv4si2", "popcountv2di2"): Rename to ...
31311 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
31312 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
31314 ("popcount<mode>2_vxe"): New pattern.
31316 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31318 * common/config/s390/s390-common.c (processor_flags_table): Add
31320 * config.gcc: Add arch12.
31321 * config/s390/driver-native.c (s390_host_detect_local_cpu):
31322 Default to arch12 for unknown CPU model numbers.
31323 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
31324 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
31325 PROCESSOR_max sanity check.
31326 * config/s390/s390-opts.h (enum processor_type): Add
31328 * config/s390/s390.c (processor_table): Add arch12.
31329 (s390_expand_builtin): Add check for B_VXE flag.
31330 (s390_issue_rate): Add PROCESSOR_ARCH12.
31331 (s390_get_sched_attrmask): Likewise.
31332 (s390_get_unit_mask): Likewise.
31333 (s390_sched_score): Enable z13 scheduling for arch12.
31334 (s390_sched_reorder): Likewise.
31335 (s390_sched_variable_issue): Likewise.
31336 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
31338 (s390_tune_attr): Use z13 scheduling also for arch12.
31339 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
31340 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
31341 (TARGET_VXE_P): New macros.
31342 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
31343 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
31344 * config/s390/s390.opt: Add arch12 as processor_type.
31346 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31348 * config/s390/s390.md
31349 ("fixuns_truncdddi2", "fixuns_trunctddi2")
31350 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
31351 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
31353 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
31354 Rename expanders to ...
31356 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
31357 ("fixuns_truncdddi2_emu"): ... these.
31359 ("fixuns_trunc<mode>si2_emu"): New expander.
31361 ("*fixuns_truncdfdi2_z13"): Rename to ...
31362 ("*fixuns_truncdfdi2_vx"): ... this.
31364 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31366 * config/s390/2964.md: Remove the single element vector compare
31367 instructions which are no longer used.
31368 * config/s390/s390.c (s390_select_ccmode): Remove handling of
31370 (s390_canonicalize_comparison): Remove handling of DFmode
31372 (s390_expand_vec_compare_scalar): Remove function.
31373 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
31374 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
31376 ("*cmp<mode>_ccs"): Add wfcdb instruction.
31378 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31380 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
31382 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
31383 will anyway by matched by mov<mode>_64dfp.
31385 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31387 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
31388 vlef/vstef. Add missing operand to vleif.
31390 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31392 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
31393 pair for all vector types with 64 bit elements.
31394 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
31395 * config/s390/vector.md (V_HW_64): ... here.
31396 (V_128_NOSINGLE): New mode iterator.
31397 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
31398 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
31399 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
31400 ("*vec_load_pairv2di"): Change to ...
31401 ("*vec_load_pair<mode>"): ... this one.
31403 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31405 * config/s390/constraints.md: Add comments.
31406 (jKK): Reject element sizes > 8 bytes.
31407 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
31409 * config/s390/s390.md: Add the s_operand checks formerly in
31410 s390_split_ok_p to various splitters where they are still
31412 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
31413 for 128 bit vectors. Plus two splitters.
31415 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31417 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
31420 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31423 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
31424 error if the boundary argument is not constant.
31426 2017-03-24 Jakub Jelinek <jakub@redhat.com>
31428 PR rtl-optimization/80112
31429 * loop-doloop.c (doloop_condition_get): Don't check condition
31430 if cmp isn't SET with IF_THEN_ELSE src.
31432 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31434 PR tree-optimization/80158
31435 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
31436 replacing a candidate statement, also replace it for the
31437 candidate's alternate interpretation.
31438 (replace_rhs_if_not_dup): Likewise.
31439 (replace_one_candidate): Likewise.
31441 2017-03-24 Richard Biener <rguenther@suse.de>
31443 PR tree-optimization/80167
31444 * graphite-isl-ast-to-gimple.c
31445 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
31447 (translate_isl_ast_to_gimple::get_rename): Likewise.
31449 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31451 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
31452 handling of certain combinations of target options, including the
31453 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
31454 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
31456 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31459 * config/arm/arm.md (*load_multiple): Add reload_completed to
31460 matching condition.
31462 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31463 Richard Biener <rguenth@suse.de>
31465 PR tree-optimization/79908
31466 PR tree-optimization/80136
31467 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31468 been cast away, gimplify_and_add suffices.
31470 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
31472 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
31474 2017-03-23 Richard Biener <rguenther@suse.de>
31476 PR tree-optimization/80032
31477 * gimplify.c (gimple_push_cleanup): Forced unconditional
31478 cleanups still have to go to the conditional_cleanups
31481 2017-03-22 Jakub Jelinek <jakub@redhat.com>
31483 PR tree-optimization/80072
31484 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
31486 (next_operand_entry_id): Change type to unsigned int.
31487 (sort_by_operand_rank): Make sure to return the right return value
31488 even if unsigned fields are bigger than INT_MAX.
31489 (struct oecount): Change cnt and id type to unsigned int.
31490 (oecount_hasher::equal): Formatting fix.
31491 (oecount_cmp): Make sure to return the right return value
31492 even if unsigned fields are bigger than INT_MAX.
31493 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
31496 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
31497 TREE_READONLY on result if writing it more than once.
31500 * doc/invoke.texi (-fsanitize=thread): Document that with
31501 -fnon-call-exceptions atomics are not able to throw
31505 * tsan.c: Include tree-eh.h.
31506 (instrument_builtin_call): Call maybe_clean_eh_stmt or
31507 maybe_clean_or_replace_eh_stmt where needed.
31508 (instrument_memory_accesses): Add cfg_changed argument.
31509 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
31510 if it returned true.
31511 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
31513 PR rtl-optimization/63191
31514 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
31515 wrapper function, moved the whole old content into ...
31516 (ix86_delegitimize_address_1): ... this. New inline function.
31517 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
31518 true as last argument instead of ix86_delegitimize_address.
31520 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31522 * config/aarch64/aarch64.c (generic_branch_cost): Copy
31523 cortexa57_branch_cost.
31525 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
31527 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
31529 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31532 * doc/md.texi (Constraints): Document wA constraint.
31533 * config/rs6000/constraints.md (wA): New.
31534 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
31535 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
31536 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
31537 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
31539 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
31542 * gimplify.c (is_oacc_declared): New function.
31543 (oacc_default_clause): Use it to set default flags for acc declared
31544 variables inside parallel regions.
31545 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
31546 declared variables.
31547 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
31548 declare attribute to any decl as necessary.
31550 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
31553 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
31554 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
31555 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
31556 (arm_arch_lpae): This.
31557 * config/arm/arm.c (arm_arch7ve): Rename into ...
31558 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
31559 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
31562 2017-03-22 Martin Liska <mliska@suse.cz>
31565 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
31566 error message instead of an ICE.
31568 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31570 * doc/extend.texi (6.11 Additional Floating Types): Revise.
31572 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
31574 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
31576 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31579 2017-03-21 Martin Sebor <msebor@redhat.com>
31581 * doc/extend.texi: Use "cannot" instead of "can't."
31582 * doc/hostconfig.texi: Same.
31583 * doc/install.texi: Same.
31584 * doc/invoke.texi: Same.
31585 * doc/loop.texi: Same.
31586 * doc/md.texi: Same.
31587 * doc/objc.texi: Same.
31588 * doc/rtl.texi: Same.
31589 * doc/tm.texi: Same.
31590 * doc/tm.texi.in: Same.
31591 * doc/trouble.texi: Same.
31593 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
31596 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
31597 (collect_checksum_attributes): Set it.
31598 (die_checksum_ordered): Use it.
31600 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31602 PR tree-optimization/79908
31603 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
31604 change: For a VA_ARG whose LHS has been cast away, use
31605 force_gimple_operand to construct the side effects.
31607 2017-03-21 David Malcolm <dmalcolm@redhat.com>
31609 PR translation/80001
31610 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
31611 more amenable to translation.
31612 (oacc_loop_auto_partitions): Likewise.
31614 2017-03-21 Marek Polacek <polacek@redhat.com>
31615 Martin Sebor <msebor@redhat.com>
31617 PR tree-optimization/80109
31618 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
31619 on INTEGRAL_TYPE_P.
31621 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31622 Segher Boessenkool <segher@kernel.crashing.org>
31625 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
31626 check reg_used_between_p between insn and one of succ or succ2
31627 depending on if succ is artificial insn not inserted into insn
31630 2017-03-21 Martin Liska <mliska@suse.cz>
31632 PR gcov-profile/80081
31633 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
31634 * doc/gcc.texi: Include gcov-dump stuff.
31635 * doc/gcov-dump.texi: New file.
31637 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
31639 PR rtl-optimization/79150
31640 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
31641 conditional jump, if the jump is the last insn of the loop.
31643 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31644 Richard Biener <rguenth@suse.de>
31646 PR tree-optimization/79908
31647 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
31648 been cast away, use force_gimple_operand to construct the side
31651 2017-03-21 Martin Liska <mliska@suse.cz>
31653 PR libfortran/79956
31654 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
31657 2017-03-21 Brad Spengler <spender@grsecurity.net>
31660 * plugin.c (htab_hash_plugin): New function.
31661 (add_new_plugin): Use it and adjust.
31662 (parse_plugin_arg_opt): Adjust.
31663 (init_one_plugin): Likewise.
31665 2017-03-21 Richard Biener <rguenther@suse.de>
31667 PR tree-optimization/80032
31668 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
31669 if set force the cleanup to happen unconditionally.
31670 (gimplify_target_expr): Push inserted clobbers with force_uncond
31671 to avoid them being removed by control-dependent DCE.
31673 2017-03-21 Richard Biener <rguenther@suse.de>
31675 PR tree-optimization/80122
31676 * tree-inline.c (copy_bb): Do not expans va-arg packs or
31677 va_arg_pack_len when the inlined call stmt requires pack
31679 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
31681 2017-03-21 Jakub Jelinek <jakub@redhat.com>
31684 * tsan.c (instrument_builtin_call): If the memory model argument
31685 is not a constant, assume it is valid.
31688 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
31691 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31693 PR rtl-optimization/79910
31694 * combine.c (can_combine_p): Do not allow combining an I0 or I1
31695 if its dest is used by an insn before I2 (other than the combined
31696 insns themselves, which are properly handled already).
31698 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
31701 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31703 * combine.c (record_used_regs): New static function.
31704 (try_combine): Handle situations where there is an additional
31705 instruction between I2 and I3 which needs to have a LOG_LINK
31709 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31711 * combine.c (try_combine): Delete redundant i1 test. Call
31712 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31714 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31717 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
31718 alternatives 13/14.
31720 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31722 PR tree-optimization/80054
31723 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
31724 the optimization if a PHI or any of its arguments is not dominated
31725 by the candidate's basis. Use gphi* rather than gimple* as
31727 (replace_profitable_candidates): Clean up a gimple* variable that
31728 should be a gphi* variable.
31730 2017-03-20 Martin Sebor <msebor@redhat.com>
31733 * doc/extend.texi (attribute constructor): Document present limitation.
31735 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
31738 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
31739 __POWER9_VECTOR__ #ifdef control, change template definition to
31740 use Power9-specific built-in function.
31741 (vec_any_eq): Likewise.
31742 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
31743 to control outcomes from this test.
31744 (vector_ae_<mode>p): For VEC_F modes, likewise.
31746 2017-03-20 Ian Lance Taylor <iant@google.com>
31748 * config/i386/i386.c (ix86_function_regparm): Save an extra
31749 register for -fsplit-stack with DECL_STATIC_CHAIN.
31751 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31754 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
31755 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
31757 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
31759 * config/riscv/riscv.c (riscv_print_operand): Use "fence
31761 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
31764 2017-03-20 Marek Polacek <polacek@redhat.com>
31767 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
31769 2017-03-20 Richard Biener <rguenther@suse.de>
31771 PR tree-optimization/80113
31772 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
31773 allocate extra SSA name for PHI def.
31774 (add_close_phis_to_outer_loops): Likewise.
31775 (add_close_phis_to_merge_points): Likewise.
31776 (copy_loop_close_phi_args): Likewise.
31777 (copy_cond_phi_nodes): Likewise.
31779 2017-03-20 Martin Liska <mliska@suse.cz>
31781 PR middle-end/79753
31782 * tree-chkp.c (chkp_build_returned_bound): Do not build
31783 returned bounds for a LHS that's not a BOUNDED_P type.
31785 2017-03-20 Martin Liska <mliska@suse.cz>
31789 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
31790 COMPLEX_CST and VECTOR_CST.
31792 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31795 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
31796 target operand. A new splitter adds the clobber statement in case
31797 the target operand is dead anyway.
31799 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
31801 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
31802 to age-old versions of binutils and glibc.
31804 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
31806 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
31808 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31810 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
31812 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
31814 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
31815 requirement for binutils 2.13.
31817 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
31819 * combine.c (try_combine): Delete redundant i1 test. Call
31820 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
31822 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
31824 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
31825 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
31827 <riscv64-*-elf>: Re-arrange section
31828 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
31829 <riscv32-*-linux>: Likewise.
31830 <riscv64-*-elf>: Likewise
31831 <riscv64-*-linux>: Likewise.
31833 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
31836 * aarch64.opt(verbose-cost-dump): Fix typo.
31838 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
31841 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
31842 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
31844 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
31846 * reload.c (find_reloads): When reloading a nonoffsettable address,
31847 use RELOAD_OTHER for it and its address reloads.
31849 PR rtl-optimization/79910
31850 * combine.c (record_used_regs): New static function.
31851 (try_combine): Handle situations where there is an additional
31852 instruction between I2 and I3 which needs to have a LOG_LINK
31855 2017-03-17 Jeff Law <law@redhat.com>
31857 PR tree-optimization/71437
31858 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
31859 conditional in the hash table first.
31860 (vrp_dom_walker::before_dom_children): Extract condition from
31861 ASSERT_EXPR. Record condition, its inverion and any implied
31862 conditions as well.
31864 2017-03-17 Marek Polacek <polacek@redhat.com>
31865 Markus Trippelsdorf <markus@trippelsdorf.de>
31867 PR tree-optimization/80079
31868 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
31871 2017-03-17 Richard Biener <rguenther@suse.de>
31873 PR middle-end/80075
31874 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
31875 Properly verify the LHS before the RHS possibly claims to be
31877 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
31880 2017-03-17 Martin Jambor <mjambor@suse.cz>
31882 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
31883 (List of -O2 options): Likewise.
31884 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
31887 2017-03-17 Tom de Vries <tom@codesourcery.com>
31889 * gcov-dump.c (print_usage): Print bug_report_url.
31891 2017-03-17 Richard Biener <rguenther@suse.de>
31893 PR middle-end/80050
31894 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
31895 (parser::peek): Likewise.
31897 2017-03-17 Richard Biener <rguenther@suse.de>
31899 PR tree-optimization/80048
31900 * sese.c (free_sese_info): Properly release rename_map and
31901 copied_bb_map elements.
31903 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
31905 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
31906 Add linked-list forward and backlinks. Insert on
31907 construction, remove on destruction.
31908 (class pass_store_merging): Add m_stores_head field.
31909 (pass_store_merging::terminate_and_process_all_chains):
31910 Iterate over m_stores_head list.
31911 (pass_store_merging::terminate_all_aliasing_chains):
31913 (pass_store_merging::execute): Check for debug stmts first.
31914 Push new chains onto the m_stores_head stack.
31916 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
31919 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
31920 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
31921 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
31923 2017-03-16 Jeff Law <law@redhat.com>
31925 PR tree-optimization/71437
31926 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
31927 member function. Implementation moved into after_dom_children
31928 member function and into the threader's thread_outgoing_edges
31930 (dom_opt_dom_walker::after_dom_children): Simplify by moving
31931 some code into new thread_outgoing_edges.
31932 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
31933 definition. Simplify marker handling (do it here). Assume we always
31934 have the available expression and the const/copies tables.
31935 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
31937 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
31938 * tree-vrp.c (equiv_stack): No longer file scoped.
31939 (vrp_dom_walker): New class.
31940 (vrp_dom_walker::before_dom_children): New member function.
31941 (vrp_dom_walker::after_dom_children): Likewise.
31942 (identify_jump_threads): Setup domwalker. Use it rather than
31943 walking edges in a random order by hand. Simplify setup/finalization.
31944 (finalize_jump_threads): Remove.
31945 (vrp_finalize): Do not call identify_jump_threads here.
31946 (execute_vrp): Do it here instead and call thread_through_all_blocks
31949 PR tree-optimization/71437
31950 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
31952 (simplify_stmt_for_jump_threading): Add basic_block argument. All
31954 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
31955 (dom_opt_dom_walker::thread_across_edge): Remove
31956 handle_dominating_asserts argument. All callers changed.
31957 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
31958 changes. Remove calls to lhs_of_dominating_assert. Other
31959 uses of handle_dominating_asserts turn into unconditional code
31960 (simplify_control_stmt_condition_1): Likewise.
31961 (simplify_control_stmt_condition): Likewise.
31962 (thread_through_normal_block, thread_across_edge): Likewise.
31963 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
31964 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
31965 object if it is not an SSA_NAME.
31966 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
31967 before calling into the VRP specific simplifiers.
31968 (identify_jump_threads): Remove handle_dominating_asserts
31971 2017-03-16 Jakub Jelinek <jakub@redhat.com>
31974 * tree-diagnostic.c (default_tree_printer): No longer static.
31975 * tree-diagnostic.h (default_tree_printer): New prototype.
31977 2017-03-16 Tamar Christina <tamar.christina@arm.com>
31979 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
31980 Change ins into fmov.
31982 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31984 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
31985 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
31986 Use h_con constraint for operand 1.
31987 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
31988 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
31990 2017-03-15 Jeff Law <law@redhat.com>
31992 PR tree-optimization/71437
31993 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
31994 (record_temporary_equivalences): Use it.
31996 PR tree-optimization/71437
31997 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
31998 tree-ssa-scopedtables.
31999 (lookup_avail_expr, build_and_record_new_cond): Likewise.
32000 (record_conditions, record_cond, vuse_eq): Likewise.
32001 (record_edge_info): Adjust to API tweak of record_conditions.
32002 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
32003 (record_temporary_equivalences, optimize_stmt): Likewise.
32004 (eliminate_redundant_computations): Likewise.
32005 (record_equivalences_from_stmt): Likewise.
32006 * tree-ssa-scopedtables.c: Include options.h and params.h.
32007 (vuse_eq): New function, moved from tree-ssa-dom.c
32008 (build_and_record_new_cond): Likewise.
32009 (record_conditions): Likewise. Accept vector of conditions rather
32010 than edge_equivalence structure for first argument.
32011 for the first argument.
32012 (avail_exprs_stack::lookup_avail_expr): New member function, moved
32013 from tree-ssa-dom.c.
32014 (avail_exprs_stack::record_cond): Likewise.
32015 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
32016 from tree-ssa-dom.c.
32017 (avail_exprs_stack): Add new member functions lookup_avail_expr
32019 (record_conditions): Declare.
32021 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
32024 * lra-constraints.c (process_alt_operands): Increase reject for
32025 reloading an input/output operand.
32027 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
32030 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
32031 insns to convert from signed/unsigned char/short to IEEE 128-bit
32033 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
32035 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
32038 * config/i386/i386.c (ix86_vector_duplicate_value): Create
32039 subreg of inner mode for values already in registers.
32041 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
32043 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
32044 iteration reg is used after the loop.
32046 2017-03-14 Martin Sebor <msebor@redhat.com>
32048 PR tree-optimization/79800
32049 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
32050 precision in negative-positive range.
32051 (format_floating): Call non-const overload with adjusted precision.
32053 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32056 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
32059 2017-03-14 Martin Sebor <msebor@redhat.com>
32061 PR middle-end/80020
32062 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
32063 * builtins.def (aligned_alloc): Use it.
32066 * Makefile.in (GTFILES): Add calls.c.
32067 * calls.c: Include "gt-calls.h".
32069 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
32071 PR rtl-optimization/79728
32072 * regs.h (struct target_regs): New field
32073 x_contains_allocatable_regs_of_mode.
32074 (contains_allocatable_regs_of_mode): New macro.
32075 * reginfo.c (init_reg_sets_1): Initialize it, and change
32076 contains_reg_of_mode so it includes global regs as well.
32077 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
32078 rather than contains_regs_of_mode.
32080 2017-03-14 Martin Liska <mliska@suse.cz>
32082 * doc/invoke.texi: Document options that can't be combined with
32083 -fcheck-pointer-bounds.
32085 2017-03-14 Martin Liska <mliska@suse.cz>
32087 PR middle-end/79831
32088 * doc/invoke.texi (-Wchkp): Document the option.
32090 2017-03-14 Martin Liska <mliska@suse.cz>
32092 * Makefile.in: Install gcov-dump.
32094 2017-03-14 Martin Liska <mliska@suse.cz>
32096 * multiple_target.c (expand_target_clones): Bail out for
32097 an invalid attribute.
32099 2017-03-14 Richard Biener <rguenther@suse.de>
32101 * alias.c (struct alias_set_entry): Pack properly.
32102 * cfgloop.h (struct loop): Likewise.
32103 * cse.c (struct set): Likewise.
32104 * ipa-utils.c (struct searchc_env): Likewise.
32105 * loop-invariant.c (struct invariant): Likewise.
32106 * lra-remat.c (struct cand): Likewise.
32107 * recog.c (struct change_t): Likewise.
32108 * rtl.h (struct address_info): Likewise.
32109 * symbol-summary.h (function_summary): Likewise.
32110 * tree-loop-distribution.c (struct partition): Likewise.
32111 * tree-object-size.c (struct object_size_info): Likewise.
32112 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
32113 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
32114 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
32115 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
32116 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
32117 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
32118 (struct _stmt_vec_info): Likewise.
32120 2017-03-14 Martin Liska <mliska@suse.cz>
32123 * multiple_target.c (create_dispatcher_calls): Check that
32124 a target can create a function dispatcher.
32126 2017-03-14 Martin Liska <mliska@suse.cz>
32129 * multiple_target.c (expand_target_clones): Drop local.local
32130 flag for default implementation.
32132 2017-03-14 Richard Biener <rguenther@suse.de>
32134 PR tree-optimization/80030
32135 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
32137 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
32139 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
32140 gcc_fallthrough() instead of __attribute__((fallthrough));
32142 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32144 * doc/gcc.texi: Remove "up" link to (DIR).
32145 * doc/gccint.texi: Ditto.
32147 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
32149 * doc/install.texi (Specific) <avr>: Remove reference to
32152 2017-03-13 Jeff Law <law@redhat.com>
32154 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
32155 attribute rather than comments.
32157 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
32158 match_scratch operand is highest.
32160 2017-03-13 Martin Liska <mliska@suse.cz>
32162 PR middle-end/78339
32163 * ipa-pure-const.c (warn_function_noreturn): If the declarations
32164 is a CHKP clone, use original declaration.
32166 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32168 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
32169 (arc_conditional_register_usage): Use a different allocation order
32170 when optimizing for size.
32171 * common/config/arc/arc-common.c (arc_option_optimization_table):
32172 Section anchors default on when optimizing for size.
32174 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32176 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
32178 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32180 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
32181 * config/arc/arc.md (cpu_facility): Add cd variant.
32182 (*movqi_insn): Add code density variant.
32183 (*movhi_insn): Likewise.
32184 (*movqi_insn): Likewise.
32185 (*addsi3_mixed): Likewise.
32186 (subsi3_insn): Likewise.
32188 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32190 * config/arc/arc.md (movsi_cond_exec): Update constraint.
32192 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
32194 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
32195 expressions with MINUS and UNARY ops.
32197 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32200 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
32202 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
32203 between vec_select and vector argument.
32204 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
32205 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
32206 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
32207 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
32208 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
32209 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
32211 2017-03-13 Richard Biener <rguenther@suse.de>
32214 * params.def (vect-max-peeling-for-alignment): Fix typo.
32216 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32218 * doc/install.texi (Specific) <mips-*-*>: Remove description of
32219 issue that only occurred with binutils below 2.18.
32221 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32223 * doc/install.texi (Specific) <cris-axis-elf>: No longer
32224 refer to binutils 2.11/2.12 minimum.
32226 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
32228 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
32229 ftp.kernel.org and simplify binutils requirement.
32231 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
32233 * doc/invoke.texi (Warning Options): Fix spelling of link-time
32235 (Optimize Options): Ditto. Also remove redundancy.
32237 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32239 PR translation/79848
32240 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
32242 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
32243 to G_ to avoid double translation.
32245 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32247 PR translation/79923
32248 * auto-profile.c (get_combined_location): Convert leading
32249 character of diagnostics to lower case and remove trailing period.
32250 (read_profile): Likewise for various diagnostics.
32251 * config/arm/arm.c (arm_option_override): Remove trailing period
32252 from various diagnostics.
32253 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
32254 (msp430_expand_delay_cycles): Likewise.
32256 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32259 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
32260 full command-line argument, rather than just "str".
32261 (aarch64_validate_march): Likewise.
32262 (aarch64_validate_mtune): Likewise.
32264 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
32266 PR rtl-optimization/78911
32267 * lra-assigns.c (must_not_spill_p): New function.
32268 (spill_for): Use it.
32270 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32272 PR tree-optimization/79981
32273 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
32274 ATOMIC_COMPARE_EXCHANGE ifn result.
32275 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
32276 IFN_ATOMIC_COMPARE_EXCHANGE.
32278 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32281 * opts.c (parse_sanitizer_options): Add missing question mark to
32282 "did you mean" message.
32284 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32286 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
32288 (VMULEUH_UNS): Likewise.
32289 (VMULOUB_UNS): Likewise.
32290 (VMULOUH_UNS): Likewise.
32291 * config/rs6000/rs6000.c (builtin_function_type): Remove
32292 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
32294 2017-03-10 David Malcolm <dmalcolm@redhat.com>
32297 * read-rtl-function.c (function_reader::read_rtx_operand): Update
32298 x with result of extra_parsing_for_operand_code_0.
32299 (function_reader::extra_parsing_for_operand_code_0): Convert
32300 return type from void to rtx, returning x. When reading
32301 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
32302 larger size containing struct block_symbol.
32304 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
32306 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
32307 -mfloat128-hardware without -m64.
32309 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
32312 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
32313 entries to the case statement that marks unsigned arguments to
32314 overloaded functions.
32316 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32318 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
32319 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
32321 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
32324 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
32325 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
32327 2017-03-10 Martin Liska <mliska@suse.cz>
32331 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
32332 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
32335 2017-03-10 Olivier Hainque <hainque@adacore.com>
32337 * tree-switch-conversion (array_value_type): Start by resetting
32338 candidate type to it's main variant.
32340 2017-03-10 Jakub Jelinek <jakub@redhat.com>
32342 PR rtl-optimization/79909
32343 * combine.c (try_combine): Use simplify_replace_rtx on individual
32344 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
32345 of the whole CALL_INSN_FUNCTION_USAGE.
32347 PR tree-optimization/79972
32348 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
32349 get_range_info on SSA_NAMEs. Formatting fixes.
32351 2017-03-10 Richard Biener <rguenther@suse.de>
32352 Jakub Jelinek <jakub@redhat.com>
32354 PR tree-optimization/77975
32355 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
32356 edge to be constant.
32357 (get_val_for): For constant x return it. Formatting fix.
32358 (loop_niter_by_eval): Avoid pointless looping if the next iteration
32359 would use the same bases as the current one.
32361 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32363 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
32364 instead of vec_select for V1TImode.
32365 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
32367 (VSX_LE_128): Add V1TI to this mode iterator.
32368 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
32369 (*vsx_le_perm_store_<mode>): Likewise.
32370 (pre-reload splitter for VSX stores): Likewise.
32371 (post-reload splitter for VSX stores): Likewise.
32372 (*vsx_xxpermdi2_le_<mode>): Likewise.
32373 (*vsx_lxvd2x2_le_<mode>): Likewise.
32374 (*vsx_stxvd2x2_le_<mode>): Likewise.
32376 2017-03-09 Michael Eager <eager@eagercon.com>
32378 Correct failures with --enable-checking=yes,rtl.
32380 * config/microblaze/microblaze.c (microblaze_expand_shift):
32381 Replace GET_CODE test with CONST_INT_P and INTVAL test with
32382 test for const0_rtx.
32383 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
32384 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
32386 2017-03-09 Richard Biener <rguenther@suse.de>
32388 PR tree-optimization/79977
32389 * graphite-scop-detection.c (scop_detection::merge_sese):
32390 Handle the case of extra exits to blocks dominating the entry.
32392 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
32394 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
32397 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
32399 PR rtl-optimization/79949
32400 * lra-constraints.c (process_alt_operands): Check memory when
32401 trying to predict a cycle. Print about the overall increase.
32403 2017-03-09 Richard Biener <rguenther@suse.de>
32405 PR middle-end/79971
32406 * gimple-expr.c (useless_type_conversion_p): Preserve
32407 TYPE_SATURATING for fixed-point types.
32409 2017-03-09 Richard Biener <rguenther@suse.de>
32412 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
32413 alignment of BLKmode params.
32415 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32418 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
32419 (VALL_NO_V2Q): Likewise.
32421 * config/aarch64/aarch64-simd.md
32422 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
32424 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
32425 VALL_NO_V2Q mode iterator.
32426 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
32428 2017-03-09 Martin Liska <mliska@suse.cz>
32430 PR tree-optimization/79631
32431 * tree-chkp-opt.c (chkp_is_constant_addr): Call
32432 tree_int_cst_sign_bit just for INTEGER constants.
32434 2017-03-09 Martin Liska <mliska@suse.cz>
32438 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
32441 2017-03-09 Marek Polacek <polacek@redhat.com>
32444 * tree.c (inchash::add_expr): Handle TREE_VEC.
32446 2017-03-09 Martin Liska <mliska@suse.cz>
32449 (chkp_narrow_size_and_offset): New function.
32450 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
32451 (void chkp_parse_bit_field_ref): New function.
32452 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
32453 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
32455 2017-03-09 Martin Liska <mliska@suse.cz>
32458 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
32459 (chkp_find_bounds_1): Remove gcc_unreachable.
32461 2017-03-09 Jakub Jelinek <jakub@redhat.com>
32464 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
32465 BUILT_IN_SYNC*, determine the access type from the size suffix and
32466 always build a MEM_REF with that type. Handle forgotten
32467 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
32470 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
32471 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
32472 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
32473 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
32474 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
32475 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
32476 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
32477 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
32478 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
32479 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
32480 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
32481 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
32482 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
32483 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
32484 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
32485 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
32486 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
32487 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
32488 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
32489 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
32490 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
32491 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
32492 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
32493 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
32494 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
32495 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
32496 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
32497 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
32498 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
32499 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
32500 definitions outside of __OPTIMIZE__ guarded section.
32503 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
32504 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
32505 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
32506 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
32509 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32511 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
32512 ("vfenez<mode>"): Add missing constraints.
32514 2017-03-08 Martin Sebor <msebor@redhat.com>
32517 * config/nds32/nds32.c (nds32_option_override):
32518 Fix misspelled diagnostic.
32520 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32523 * gimplify.c (gimplify_omp_for): Replace index var in outer
32524 taskloop statement with an artificial variable and add
32525 OMP_CLAUSE_PRIVATE clause for it.
32527 2017-03-08 Richard Biener <rguenther@suse.de>
32529 PR tree-optimization/79955
32530 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
32531 for accesses that are completely outside of the variable.
32533 2017-03-08 Andrew Haley <aph@redhat.com>
32535 PR tree-optimization/79943
32536 * tree-ssa-loop-split.c (compute_new_first_bound): When
32537 calculating the new upper bound, (END-BEG) should be added, not
32540 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32542 * config/avr/avr.md (setmemhi): Make sure match_dup
32543 operand number comes before match_scratch.
32545 2017-03-08 Richard Biener <rguenther@suse.de>
32547 PR tree-optimization/79920
32548 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
32549 with ncopies == 1 to ...
32550 (vect_transform_slp_perm_load): ... here. Properly compute
32551 all element loads by iterating VF times over the group. Do
32552 not handle ncopies (computed in a broken way) in
32553 vect_create_mask_and_perm.
32555 2017-03-08 Jakub Jelinek <jakub@redhat.com>
32558 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
32559 is a uniform vector, use uniform_vector_p return value instead of
32560 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
32562 2017-03-07 Marek Polacek <polacek@redhat.com>
32564 PR middle-end/79809
32565 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
32566 (alloca_call_type): Likewise.
32568 2017-03-07 Martin Liska <mliska@suse.cz>
32570 * gcov.c (process_args): Put comment to correct location.
32572 2017-03-07 Martin Liska <mliska@suse.cz>
32574 PR middle-end/68270
32575 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
32576 Use array_at_struct_end_p instead of DECL_CHAIN (field).
32577 (chkp_narrow_bounds_for_field): Likewise.
32578 (chkp_parse_array_and_component_ref): Pass one more argument to
32581 2017-03-07 Richard Biener <rguenther@suse.de>
32583 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
32586 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
32588 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
32589 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
32591 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32594 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
32595 to end of description.
32596 (PARAM_MAX_STORES_TO_MERGE): Likewise.
32598 2017-03-07 Jakub Jelinek <jakub@redhat.com>
32600 PR rtl-optimization/79901
32601 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
32603 (*avx512f_<code><mode>3<mask_name>): ... this.
32604 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
32605 iterator instead of VI8_AVX2_AVX512BW.
32607 PR rtl-optimization/79901
32608 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
32609 min/max expander, expand it using expand_vec_cond_expr.
32612 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
32615 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32618 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
32619 to void * for PCH reasons.
32620 * dwarf2out.c (output_loc_operands, output_die): Cast
32621 v.val_vec.array to unsigned char *.
32623 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
32626 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
32629 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
32631 PR rtl-optimization/79571
32632 * lra-constraints.c (process_alt_operands): Calculate static
32633 reject and subtract it from overall when only addresses will be
32636 2017-03-06 Julia Koval <julia.koval@intel.com>
32639 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
32640 incoming stack boundary to 128 for 64-bit targets.
32642 2017-03-06 Richard Biener <rguenther@suse.de>
32644 PR tree-optimization/79894
32645 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
32646 to NULL after folding it.
32648 2017-03-06 Richard Biener <rguenther@suse.de>
32650 PR tree-optimization/79824
32651 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
32652 check disabling peeling for gaps.
32654 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
32656 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
32657 attributes): Document gettimeofday.
32659 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32661 * config/s390/s390.c (s390_option_override_internal): Set
32662 PARAM_MIN_VECT_LOOP_BOUND
32664 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
32666 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
32667 * config/s390/s390.md: Likewise.
32669 2017-03-06 Jakub Jelinek <jakub@redhat.com>
32672 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
32673 (<avx2_avx512>_perm<mode>): Rename to ...
32674 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
32676 (<avx512>_perm<mode>_mask): Rename to ...
32677 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
32679 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
32680 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
32681 instead of VI8F_256_512.
32682 (avx512f_perm<mode>): New define_expand.
32683 (avx512f_perm<mode>_mask): Likewise.
32684 (avx512f_perm<mode>_1<mask_name>): New define_insn.
32685 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
32687 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32689 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
32690 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
32692 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
32694 2017-03-06 Martin Liska <mliska@suse.cz>
32697 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
32698 when having a SSA NAME w/o VAR_DECL assigned to it.
32700 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32702 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
32703 msa_dpsub_<su>_d): Fix MODE for vec_select.
32705 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
32707 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
32709 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
32711 2017-03-06 Richard Biener <rguenther@suse.de>
32713 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
32714 * plugin.c (register_plugin_info): Likewise.
32715 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
32717 2017-03-05 Jakub Jelinek <jakub@redhat.com>
32719 * config/i386/sse.md (sse_storehps, sse_storelps,
32720 avx_<castmode><avxsizesuffix>_<castmode>,
32721 avx512f_<castmode><avxsizesuffix>_<castmode>,
32722 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
32723 in condition that at least one operand is not a MEM.
32725 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32727 PR middle-end/79805
32728 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
32729 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
32731 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
32732 gimple_call_nothrow_p flag based on whether original builtin can throw.
32733 If it can, emit following stmts on the fallthrough edge.
32734 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
32735 don't create new bb if inserting just debug stmts on the edge, try to
32736 insert them on the fallthru bb or just reset debug stmts.
32738 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
32741 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
32742 restore recog_data (including the operand rtxes inside it) around
32743 the call to get_insn_template.
32745 2017-03-03 Martin Sebor <msebor@redhat.com>
32747 PR tree-optimization/79699
32748 * context.c (context::~context): Free MPFR caches to avoid
32749 a memory leak on program exit.
32751 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32753 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
32754 Use wide_int::ulow () instead of .elt (0).
32756 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32758 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
32759 (*pushxf): Limit oF constraint to 32bit targets and add oC
32760 constraint for 64bit targets.
32761 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
32762 (*pushdf): Change rmF constraint to rmC.
32764 2017-03-03 Martin Liska <mliska@suse.cz>
32766 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
32767 Remove unused variable.
32769 2017-03-03 Jakub Jelinek <jakub@redhat.com>
32772 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
32773 is a memory operand, increase num_memory.
32774 (ix86_expand_args_builtin): Likewise.
32776 2017-03-03 Jan Hubicka <jh@suse.cz>
32779 * ipa-devirt.c (maybe_record_node): Properly handle
32780 __cxa_pure_virtual visibility.
32782 2017-03-03 Martin Liska <mliska@suse.cz>
32784 PR tree-optimization/79803
32785 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
32787 (pass_loop_prefetch::execute): Disabled optimization if an
32788 assumption about L1 cache size is not met.
32790 2017-03-03 Martin Liska <mliska@suse.cz>
32792 PR rtl-optimization/79574
32793 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
32794 (hash_scan_set): Likewise.
32795 (dump_hash_table): Likewise.
32796 (hoist_code): Likewise.
32798 2017-03-03 Richard Biener <rguenther@suse.de>
32800 * fixed-value.c (fixed_from_string): Restore use of elt (1)
32801 in place of uhigh ().
32802 (fixed_convert_from_real): Likewise.
32804 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
32807 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
32809 2017-03-03 Richard Biener <rguenther@suse.de>
32811 PR middle-end/79818
32812 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
32813 TYPE_OVERFLOW_UNDEFINED check.
32815 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32817 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
32819 (vector_ae_<mode>_p): Likewise.
32820 (vector_nez_<mode>_p): Likewise.
32821 (vector_ne_v2di_p): Likewise.
32822 (vector_ae_v2di_p): Likewise.
32823 (vector_ne_<mode>_p): Likewise.
32824 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
32826 (vsx_tsqrt<mode>2_fe): Likewise.
32828 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
32831 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
32833 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32835 PR rtl-optimization/79780
32836 * cprop.c (one_cprop_pass): When second and further conditional trap
32837 in a single basic block is turned into an unconditional trap, turn it
32838 into a deleted note to avoid RTL verification failures.
32840 2017-03-02 Richard Biener <rguenther@suse.de>
32842 * fold-const.c (const_binop): Use ulow () instead of elt (0).
32844 2017-03-02 Richard Biener <rguenther@suse.de>
32846 PR tree-optimization/79345
32848 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
32849 param and abort the walk, returning -1 if it is hit.
32850 (walk_aliased_vdefs): Take a limit param and pass it on.
32851 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
32852 defaulting to 0 and return a signed int.
32853 * tree-ssa-uninit.c (struct check_defs_data): New struct.
32854 (check_defs): New helper.
32855 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
32856 about uninitialized memory.
32857 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
32858 bogus uninitialized warning.
32859 (fixed_convert_from_real): Likewise.
32861 2017-03-02 Bin Cheng <bin.cheng@arm.com>
32863 PR tree-optimization/66768
32864 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
32865 iv_use if base object can't be determined.
32867 2017-03-02 Jakub Jelinek <jakub@redhat.com>
32869 PR tree-optimization/79345
32870 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
32871 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
32872 (get_pattern_stats): Initialize it.
32873 * genemit.c (gen_expand): Verify match_scratch numbers come after
32874 match_operand/match_dup numbers.
32875 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
32876 match_scratch numbers.
32877 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
32879 * config/s390/s390.md (trunctdsd2): Likewise.
32881 2017-03-02 Richard Biener <rguenther@suse.de>
32883 * wide-int.h (wide_int_storage::operator=): Implement in terms
32886 2017-03-02 Richard Biener <rguenther@suse.de>
32888 PR tree-optimization/79777
32889 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
32890 the to insert expression to sth existing.
32892 2017-03-01 Martin Sebor <msebor@redhat.com>
32894 PR middle-end/79692
32895 * gimple-ssa-sprintf.c
32896 (directive::known_width_and_precision): New function.
32897 (format_integer): Use it.
32898 (get_mpfr_format_length): Consider the full range of precision
32899 when computing %g output with the # flag. Set the likely byte
32900 count to 3 rather than 1 when precision is indeterminate.
32901 (format_floating): Correct the lower bound of precision.
32903 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32905 * doc/invoke.texi: Document default code model for 64-bit Linux.
32907 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32910 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
32911 udiv rather than div since input pattern is unsigned.
32913 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
32915 * config/i386/i386.c (print_reg): Warn for values of
32916 unsupported size in integer register.
32918 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
32921 * config/rs6000/predicates.md (current_file_function_operand): Do
32922 not allow self calls to be local if the function is replaceable.
32924 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
32927 * config/rs6000/altivec.h (vec_ctz and others): Change the
32928 preprocessor macro that controls conditional compilation from
32929 _ARCH_PWR9 to __POWER9_VECTOR__.
32930 (vec_all_ne): Change parameterization of __altivec_scalar_pred
32931 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
32932 control (instead of _ARCH_PWR9 control) so that template
32933 definition uses power9-specific function.
32934 (vec_any_eq): Likewise.
32935 (vec_all_ne): Change macro definition to use a power9-specific
32936 expansion under #ifdef __POWER9_VECTOR__ control (instead of
32937 _ARCH_PWR9 control).
32938 (vec_any_eq) Likewise.
32939 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
32940 expansion for CMPNEF to remove support for xvcmpnesp instruction.
32941 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
32942 support for xvcmpnedp instruction.
32943 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
32944 macro expansion so that Power9 implementation of vec_all_ne does
32945 not use the AltiVec predicate framework.
32946 (VCMPNEH_P): Likewise.
32947 (VCMPNEW_P): Likewise.
32948 (VCMPNED_P): Likewise.
32949 (VCMPNEFP_P): Likewise.
32950 (VCMPNEDP_P): Likewise.
32951 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
32952 implementation of vec_any_eq to not use AltiVec predicate
32954 (VCMPAEH_P): Likewise.
32955 (VCMPAEW_P): Likewise.
32956 (VCMPAED_P): Likewise.
32957 (VCMPAEFP_P): Likewise.
32958 (VCMPAEDP_P): Likewise.
32959 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
32960 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
32961 not use the AltiVec predicate framework.
32962 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
32963 of vec_any_eq to not use AltiVec predicate framework.
32964 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
32965 support for predefined __POWER9_VECTOR__ macro to indicate that
32966 Power9 instruction selection is enabled.
32967 (altivec_overloaded_builtins): Remove extraneous
32968 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
32969 function argument types RS6000_BTI_bool_V16QI and
32970 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
32971 entry for overloaded function argument types RS6000_BTI_bool_V4SI
32972 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
32973 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
32974 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
32975 Power9 for implementations of vec_cmpne. Change the signature for
32976 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
32977 (representing vec_all_ne) to remove the previously described first
32978 argument of type RS6000_BTI_INTSI, as this was an artifact of
32979 reliance on the AltiVec predicate framework, which is no longer
32980 used in the implementation of these functions. Add
32981 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
32982 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
32983 since, unlike the AltiVec predicate framework implementation, we
32984 do not share function descriptors between vec_alle and vec_anyeq.
32985 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
32986 set of modes that receive special treatment even when
32987 TARGET_P9_VECTOR is true. The special treatment emits code that
32988 does not depend on Power9 instructions.
32989 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
32990 define_expand to not rely on AltiVec predicate framework.
32991 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
32993 (vector_ne_v2di_p): Change this define_expand to not rely on
32994 AltiVec predicate framework.
32995 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
32997 (vector_ne_<mode>_p): Change this define_expand to not rely on
32998 AltiVec predicate framework.
32999 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
33001 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
33002 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
33003 define_insn pattern.
33004 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
33005 define_insn pattern because the xvcmpne<VSs>. instruction is not
33007 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
33008 instruction is not supported.
33010 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33012 * config/nvptx/nvptx.c: Include intl.h.
33014 2017-03-01 Martin Jambor <mjambor@suse.cz>
33017 * ipa-prop.h (ipa_bits): Removed field known.
33018 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
33019 to pointers. Adjusted their comments to warn about their sharing.
33020 (ipcp_transformation_summary): Change bits to a vector of pointers.
33021 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
33022 (ipa_get_ipa_bits_for_value): Declare.
33023 * tree-vrp.h (value_range): Mark as GTY((for_user)).
33024 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
33025 (ipa_bits_hash_table): Likewise.
33026 (ipa_vr_ggc_hash_traits): Likewise.
33027 (ipa_vr_hash_table): Likewise.
33028 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
33029 being pointers and vr_known being removed.
33030 (ipa_set_jf_unknown): Likewise.
33031 (ipa_get_ipa_bits_for_value): New function.
33032 (ipa_set_jfunc_bits): Likewise.
33033 (ipa_get_value_range): New overloaded functions.
33034 (ipa_set_jfunc_vr): Likewise.
33035 (ipa_compute_jump_functions_for_edge): Use the above functions to
33036 construct bits and vr parts of jump functions.
33037 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
33038 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33040 (ipcp_grow_transformations_if_necessary): Also allocate
33041 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
33043 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
33044 them. Fix too long lines.
33045 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
33046 vr_known being removed.
33047 (ipa_read_jump_function): Use new setter functions to construct bits
33048 and vr parts of jump functions or set them to NULL.
33049 (write_ipcp_transformation_info): Adjust for bits being pointers.
33050 (read_ipcp_transformation_info): Likewise.
33051 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
33053 Include gt-ipa-prop.h.
33054 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
33056 (ipcp_store_bits_results): Likewise.
33057 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
33058 Do not write to existing jump functions but use a temporary instead.
33060 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33063 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
33064 attempt to use its first operand as BIT_FIELD_REF base.
33066 2017-03-01 Richard Biener <rguenther@suse.de>
33068 PR middle-end/79721
33069 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
33070 interpolating formula in wrapping arithmetic.
33071 (chrec_apply): Convert chrec_evaluate return value to wanted type.
33073 2017-03-01 Jakub Jelinek <jakub@redhat.com>
33075 PR tree-optimization/79734
33076 * tree-vect-generic.c (expand_vector_condition): Optimize
33077 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
33078 Handle VEC_COND_EXPR where comparison has different inner width from
33079 type's inner width.
33081 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
33083 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
33084 markup, and similar issues. Remove @opindex entries for things
33085 that aren't options. Add missing -mmpy-option entries.
33087 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33089 PR tree-optimization/79737
33090 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
33091 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
33092 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
33093 instead of byte_size. Formatting fix.
33094 (shift_bytes_in_array_right): Formatting fix.
33096 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
33099 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
33100 condition on optimize for the leaf function test.
33102 2017-02-28 Martin Liska <mliska@suse.cz>
33105 * read-rtl-function.c (function_reader::handle_unknown_directive):
33106 Bail out when one uses -flto.
33108 2017-02-28 Martin Liska <mliska@suse.cz>
33110 * common.opt: Replace space with tabular for options of <number>
33112 * config/i386/i386.opt: Show <number> value for
33113 -mlarge-data-threshold.
33114 * opts.c (print_filtered_help): Do not display number in hexadecimal
33117 2017-02-28 Martin Liska <mliska@suse.cz>
33119 * common.opt: Fix --help=option -Q for options which are of
33122 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
33124 * config/i386/i386.c (print_reg): Error out for values
33125 of 8-bit size in invalid integer register.
33127 2017-02-28 Martin Sebor <msebor@redhat.com>
33129 PR tree-optimization/79691
33130 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
33132 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33135 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
33136 gcc_unreachable with output_operand_lossage.
33138 2017-02-28 Richard Biener <rguenther@suse.de>
33140 PR tree-optimization/79740
33141 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
33143 (visit_nary_op): Insert the nary into the hashtable if we
33144 pattern-matched sth.
33145 * tree-ssa-pre.c (eliminate_insert): Robustify.
33147 2017-02-28 Richard Biener <rguenther@suse.de>
33149 PR middle-end/79731
33150 * fold-const.c (decode_field_reference): Reject out-of-bound
33153 2017-02-28 Jakub Jelinek <jakub@redhat.com>
33155 * config/i386/i386.c: Include intl.h.
33156 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
33157 instead of just cond ? "..." : "...".
33158 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
33159 * coverage.c (read_counts_file): Likewise.
33160 * omp-offload.c: Include intl.h.
33161 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
33162 of just cond ? "..." : "...".
33163 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
33164 of just cond ? "..." : "...".
33166 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
33169 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
33171 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
33172 'tune for' CPU name.
33173 * config/arm/arm-cpu-data.h: Regenerated.
33175 2017-02-28 Richard Biener <rguenther@suse.de>
33177 PR tree-optimization/79732
33178 * tree-inline.c (expand_call_inline): Do not shadow var.
33180 2017-02-28 Richard Biener <rguenther@suse.de>
33182 PR tree-optimization/79723
33183 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
33184 address-space properly.
33186 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
33188 * doc/optinfo.texi (Optimization groups): Fix option used for
33190 * doc/invoke.texi (-fopt-info): Document "omp".
33191 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
33192 (OPTGROUP_ALL): Add OPTGROUP_OMP.
33193 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
33194 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
33195 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
33197 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
33199 * dumpfile.c (optgroup_options): Instead of "openmp", associate
33200 OPTGROUP_OMP with "omp".
33202 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
33205 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
33206 for arithmetic shift of unsigned V2DI.
33208 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
33210 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
33211 arc/linux.h headers.
33212 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
33213 (LINK_SPEC): Likewise.
33214 (ARC_TLS_EXTRA_START_SPEC): Likewise.
33215 (EXTRA_SPECS): Likewise.
33216 (STARTFILE_SPEC): Likewise.
33217 (ENDFILE_SPEC): Likewise.
33218 (LIB_SPEC): Likewise.
33219 (TARGET_SDATA_DEFAULT): Likewise.
33220 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
33221 (MULTILIB_DEFAULTS): Likewise.
33222 (DWARF2_UNWIND_INFO): Likewise.
33223 * config/arc/big.h: New file.
33224 * config/arc/elf.h: Likewise.
33225 * config/arc/linux.h: Likewise.
33226 * config/arc/t-uClibc: Remove.
33228 2017-02-27 Bin Cheng <bin.cheng@arm.com>
33230 PR tree-optimization/77536
33231 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
33232 (tree_transform_and_unroll_loop): Use above function to compute the
33233 estimated niter of unrolled loop and use it when scaling profile.
33234 Also use count info rather than frequency if it's non-zero.
33235 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
33236 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
33237 (vect_transform_loop): Call above function.
33239 2017-02-27 Richard Biener <rguenther@suse.de>
33241 PR tree-optimization/45397
33242 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
33243 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
33244 (visit_nary_op): Add pattern matching for CSEing sign-changed
33245 or truncated operations with wider ones.
33247 2017-02-27 Richard Biener <rguenther@suse.de>
33249 PR tree-optimization/79690
33250 * tree-vect-stmts.c (vectorizable_store): Use vector type
33251 built from the DR with address-space.
33253 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
33255 * doc/invoke.texi (Optimize Options): Refine the description
33256 of asan-use-after-return.
33258 2017-02-25 Alan Modra <amodra@gmail.com>
33260 PR rtl-optimization/79584
33261 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
33262 base, not ad->base_term, the reg within base. Remove assertion
33263 that ad->base == ad->base_term. Replace gen_int_mode using
33264 bogus mode with const0_rtx.
33266 2017-02-25 Jakub Jelinek <jakub@redhat.com>
33268 PR middle-end/79396
33269 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
33270 FMA_EXPR like tcc_binary or tcc_unary.
33272 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
33275 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
33277 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
33278 (output_loc_operands): Handle DW_OP_call_ref and
33279 DW_OP_GNU_variable_value.
33280 (struct variable_value_struct): New type.
33281 (struct variable_value_hasher): Likewise.
33282 (variable_value_hash): New variable.
33283 (string_types): Remove.
33284 (copy_loc_descr): New function.
33285 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
33286 (prepend_loc_descr_to_each): New function.
33287 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
33288 instead of add_loc_descr_to_each if the first argument is single
33289 location list and the second has multiple.
33290 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
33291 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
33292 when looking for variable value which doesn't have other location info.
33293 (loc_list_from_tree): Formatting fix.
33294 (gen_array_type_die): Simplify DW_AT_string_length handling.
33295 (adjust_string_types): Remove.
33296 (gen_subprogram_die): Don't call adjust_string_types nor test/set
33297 string_types. Call resolve_variable_values.
33298 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
33299 (resolve_addr_in_expr): Likewise. Add A argument.
33300 (copy_deref_exprloc): Remove deref argument. Adjust for the
33301 original expression being DW_OP_GNU_variable_value with optionally
33302 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
33303 optionally after it.
33304 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
33305 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
33306 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
33307 (variable_value_hasher::hash, variable_value_hasher::equal): New
33309 (resolve_variable_value_in_expr, resolve_variable_value,
33310 resolve_variable_values, note_variable_value_in_expr,
33311 note_variable_value): New functions.
33312 (dwarf2out_early_finish): Call note_variable_value on all toplevel
33315 2017-02-24 Jakub Jelinek <jakub@redhat.com>
33318 * opts.h (handle_generated_option): Add GENERATED_P argument.
33319 * opts-common.c (handle_option): Adjust function comment.
33320 (handle_generated_option): Add GENERATED_P argument, pass it to
33322 (control_warning_option): Pass false to handle_generated_option
33324 * opts.c (maybe_default_option): Pass true to handle_generated_option
33326 * optc-gen.awk: Likewise.
33328 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33330 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
33331 a REG, look at the REG it is a SUBREG of.
33332 (splitter for cmpeqsi_t): Ditto.
33334 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33336 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
33337 the special USEs with the pattern of the insn, not the insn itself.
33339 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
33342 * doc/invoke.texi: Document -mload-store-pairs.
33344 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33345 Sandra Loosemore <sandra@codesourcery.com>
33347 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
33348 argument isn't a CONST_INT.
33349 (nios2_alternate_compare_const): Assert op is a CONST_INT.
33350 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
33351 (nios2_validate_compare): Bypass alternate compare logic if *op2
33352 is not a CONST_INT.
33353 (ldstwm_operation_p): Return false if first_base is not a REG or
33354 if first_offset is not a CONST_INT.
33356 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33358 * config/cris/cris.md: Use correct operand in a define_peephole2.
33360 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33362 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
33364 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
33366 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
33367 this_insn if it is an INSN or JUMP_INSN.
33368 (force_offsettable): Look at base, not at addr.
33369 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
33370 on things that aren't necessarily CONST_INTs.
33372 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
33374 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
33375 -mfpmath=sse is the default also for x86-32 targets with SSE2
33376 instruction set when @option{-ffast-math} is enabled
33378 2017-02-24 Jeff Law <law@redhat.com>
33380 PR rtl-optimizatoin/79286
33381 * ira.c (update_equiv_regs): Drop may_trap_p exception to
33384 2017-02-24 Richard Biener <rguenther@suse.de>
33386 PR tree-optimization/79389
33387 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
33390 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
33392 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
33393 function comment to reflect reality.
33394 (loop_exits_before_overflow): Fix typo in function description.
33396 2017-02-24 Richard Biener <rguenther@suse.de>
33398 PR tree-optimization/79389
33399 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
33400 properly that a threading opportunity exists. Detect conditional
33401 copy/constant propagation opportunities.
33403 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
33405 * config/visium/visium.md (type): Add trap.
33406 (b): New mode attribute.
33407 (*btst): Rename into...
33408 (*btst<mode>): ...this and adjust.
33409 (*cbranchsi4_btst_insn): Rename into...
33410 (*cbranch<mode>4_btst_insn): ...this and adjust.
33411 (trap): New define_insn.
33413 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33415 PR tree-optimization/79389
33416 * ifcvt.c (struct noce_if_info): Add rev_cond field.
33417 (noce_reversed_cond_code): New function.
33418 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
33419 reversed_comparison_code. Formatting fix.
33420 (noce_try_store_flag): Test rev_cond != NULL in addition to
33421 reversed_comparison_code.
33422 (noce_try_store_flag_constants): Likewise.
33423 (noce_try_store_flag_mask): Likewise.
33424 (noce_try_addcc): Use rev_cond if non-NULL instead of
33425 reversed_comparison_code.
33426 (noce_try_cmove_arith): Likewise. Formatting fixes.
33427 (noce_try_minmax, noce_try_abs): Clear rev_cond.
33428 (noce_find_if_block): Initialize rev_cond.
33429 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
33430 instead of false as last argument never attempt to reverse it
33433 2017-02-23 Bin Cheng <bin.cheng@arm.com>
33435 PR tree-optimization/79663
33436 * tree-predcom.c (combine_chains): Process refs in reverse order
33437 only for ZERO length chains, and add explaining comment.
33439 2017-02-23 Jeff Law <law@redhat.com>
33441 PR tree-optimization/79578
33442 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
33443 in call to operand_equal_p.
33445 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
33448 * config/i386/cpuid.h: Fix another undefined behavior.
33450 2017-02-23 Richard Biener <rguenther@suse.de>
33452 PR tree-optimization/79683
33453 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
33454 vector types for data-refs.
33456 2017-02-23 Martin Liska <mliska@suse.cz>
33458 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
33460 2017-02-23 Jakub Jelinek <jakub@redhat.com>
33462 PR middle-end/79665
33463 * internal-fn.c (get_range_pos_neg): Moved to ...
33464 * tree.c (get_range_pos_neg): ... here. No longer static.
33465 * tree.h (get_range_pos_neg): New prototype.
33466 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
33467 are known to be in between 0 and signed maximum inclusive, try to
33468 expand both unsigned and signed divmod and use the cheaper one from
33471 2017-02-22 Jeff Law <law@redhat.com>
33473 PR tree-optimization/79578
33474 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
33475 to compare base operands.
33477 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
33480 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
33481 gpc_reg_operand instead of fpr_reg_operand.
33483 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
33485 * config/mips/mips.c (mips_return_in_memory): Force FP
33486 vector types to be returned in memory for o32 ABI.
33488 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33490 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
33491 instead of DW_TAG_member for static data member declarations and don't
33492 set no_linkage_name for static inline data members.
33493 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
33496 2017-02-22 Martin Liska <mliska@suse.cz>
33498 * doc/invoke.texi: Replace inequality signs with square brackets
33501 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
33504 * lra-constraints.c (simplify_operand_subreg): Handle
33505 WORD_REGISTER_OPERATIONS targets.
33507 2017-02-22 Jakub Jelinek <jakub@redhat.com>
33510 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
33511 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
33512 elimination by swapping fld*.
33514 2017-02-22 Richard Biener <rguenther@suse.de>
33516 PR tree-optimization/79673
33517 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
33518 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
33519 irrelevant address-space qualifiers and avoiding a
33520 ADDR_SPACE_CONVERT_EXPR from fold_convert.
33522 2017-02-22 Richard Biener <rguenther@suse.de>
33524 PR tree-optimization/79666
33525 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
33526 to not symbolically negate if that may introduce undefined
33529 2017-02-22 Martin Liska <mliska@suse.cz>
33532 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
33533 * data-streamer-out.c (streamer_write_gcov_count_stream):
33535 * value-prof.c (stream_out_histogram_value): Make assert more
33536 precise based on type of counter.
33538 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
33541 * config/i386/i386.md (standard_x87sse_constant_load splitter):
33542 Use nonimmediate_operand instead of memory_operand for operand 1.
33543 (float-extend standard_x87sse_constant_load splitter): Ditto.
33545 2017-02-21 Jeff Law <law@redhat.com>
33547 PR tree-optimization/79621
33548 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
33549 blocks with edges to themselves.
33551 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33554 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
33555 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
33556 Use gimple_call_builtin_p.
33559 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
33560 on temporarily removed DEBUG_INSNs.
33562 PR tree-optimization/79649
33563 * tree-loop-distribution.c (classify_partition): Give up on
33564 non-generic address space loads/stores.
33566 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
33568 * doc/loop.texi (Loop manipulation): Remove nonexistent
33569 tree_ssa_loop_version from the documentation.
33570 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
33572 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33575 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
33576 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
33577 * config/rs6000/rs6000.c: Include except.h.
33578 (rs6000_expand_split_stack_prologue): Call
33579 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
33581 2017-02-21 Martin Jambor <mjambor@suse.cz>
33584 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
33585 have been analyzed.
33587 2017-02-21 Martin Jambor <mjambor@suse.cz>
33589 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
33590 for backward compatibility only.
33591 * doc/invoke.texi (Option Summary): Remove all references to
33592 -fipa-cp-alignment.
33594 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
33598 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33600 * lra-constraints.c (curr_insn_transform): Handle
33601 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33603 2017-02-21 Martin Liska <mliska@suse.cz>
33605 * config/i386/i386.opt: Replace -masm-dialect with -masm.
33607 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
33609 PR translation/79638
33610 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
33612 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
33615 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
33616 (arm_function_ok_for_sibcall): Return false for an indirect call by
33617 descriptor if all the argument registers are used.
33618 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
33619 alignment of the function.
33621 2017-02-21 Jakub Jelinek <jakub@redhat.com>
33623 PR tree-optimization/61441
33624 * simplify-rtx.c (simplify_const_unary_operation): For
33625 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
33626 the sNaN unmodified.
33628 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33630 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33631 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33632 instead of SYSTEM_HEADER_DIR.
33634 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
33635 Martin Liška <mliska@suse.cz>
33637 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
33638 Fix typos and grammar, use active voice, and clarify.
33640 2017-02-20 Marek Polacek <polacek@redhat.com>
33642 PR middle-end/79537
33643 * gimplify.c (gimplify_expr): Handle unused *&&L;.
33646 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
33648 2017-02-20 Jakub Jelinek <jakub@redhat.com>
33651 * config/i386/i386.c (ix86_expand_builtin): Handle
33652 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
33653 ix86_builtins_isa[fcode].isa as a requirement of those
33654 flags and any other flag in the bitmask.
33655 (ix86_init_mmx_sse_builtins): Use 0 instead of
33656 ~OPTION_MASK_ISA_64BIT as mask.
33657 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
33658 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
33659 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
33660 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
33662 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33665 * lra-constraints.c (split_reg): Check requested split mode
33666 is supported by the register.
33668 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33670 * lra-constraints.c (simplify_operand_subreg): Remove early
33673 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33676 * lra-constraints.c (curr_insn_transform): Tighten condition
33677 for converting SUBREG reloads from OP_OUT to OP_INOUT.
33679 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
33682 * lra-constraints.c (curr_insn_transform): Handle
33683 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
33685 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
33688 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
33690 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
33692 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
33695 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
33698 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33700 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
33703 2017-02-18 Jakub Jelinek <jakub@redhat.com>
33705 * final.c (last_columnnum, override_columnnum): New variables.
33706 (final_start_function): Set last_columnnum, pass it to begin_prologue
33707 hook and pass 0 to dwarf2out_begin_prologue.
33708 (final_scan_insn): Update override_columnnum. Pass last_columnnum
33709 to source_line debug hook.
33710 (notice_source_line): Compute last_columnnum and for debug_column_info
33711 return true on column changes.
33712 * debug.h (struct gcc_debug_hooks): Add column argument to
33713 source_line and begin_prologue hooks.
33714 (debug_nothing_int_charstar_int_bool): Remove prototype.
33715 (debug_nothing_int_int_charstar,
33716 debug_nothing_int_int_charstar_int_bool): New prototypes.
33717 (dwarf2out_begin_prologue): Add column argument.
33718 * debug.c (do_nothing_debug_hooks): Adjust source_line and
33719 begin_prologue hooks.
33720 (debug_nothing_int_charstar_int_bool): Remove.
33721 (debug_nothing_int_int_charstar,
33722 debug_nothing_int_int_charstar_int_bool): New functions.
33723 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
33724 through to dwarf2out_source_line.
33725 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
33726 (dwarf2out_source_line): Add column argument, emit it if requested.
33727 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
33729 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33730 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
33731 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
33732 through to dwarf2out_begin_prologue.
33733 (vmsdbgout_source_line): Add column argument, pass it through to
33734 dwarf2out_source_line.
33735 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
33736 dbxout_source_line caller.
33737 (dbxout_source_line): Add column argument.
33739 * common.opt (gno-column-info, gcolumn-info): New options.
33740 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
33741 (check_die): Also test for multiple DW_AT_decl_column attributes.
33742 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
33743 DW_AT_decl_column if requested.
33744 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
33746 (gen_variable_die): Likewise.
33747 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
33748 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
33751 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
33752 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
33753 (ix86_handle_option): Handle OPT_m3dnowa.
33754 * doc/invoke.texi (-m3dnowa): Document.
33755 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
33756 -m3dnowa instead of -m3dnow -march=athlon.
33759 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
33760 instead of gcc_assert for K, r and R code checks. Formatting fixes.
33762 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33765 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
33766 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
33767 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
33768 generator for vsx_xxpermdi_<mode>_be.
33769 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
33770 force big-endian semantics.
33771 (vsx_xxpermdi_<mode>_be): New define_expand with same
33772 implementation as previous version of vsx_xxpermdi_<mode>.
33774 2017-02-17 Jakub Jelinek <jakub@redhat.com>
33776 PR tree-optimization/79327
33777 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
33778 variable, its initialization and use.
33780 2017-02-17 Julia Koval <julia.koval@intel.com>
33782 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
33783 (OPTION_MASK_ISA_PKU_UNSET): New.
33784 (ix86_handle_option): Handle -mrdpid.
33785 * config/i386/cpuid.h (bit_RDPID): New.
33786 * config/i386/driver-i386.c (host_detect_local_cpu):
33787 Detect RDPID feature.
33788 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
33789 * config/i386/i386-c.c (ix86_target_macros_internal):
33791 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
33792 (ix86_valid_target_attribute_inner_p): Add "rdpid".
33793 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
33794 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
33795 * config/i386/i386.md (define_insn "rdpid"): New.
33796 * config/i386/i386.opt Add -mrdpid.
33797 * config/i386/immintrin.h (_rdpid_u32): New.
33799 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
33801 PR rtl-optimization/79541
33802 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
33803 instead of transforming it into USE.
33805 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
33807 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
33808 If HONOR_SNANS (SFmode) force the input to a register.
33809 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
33810 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
33811 an frsp or similar insn.
33813 2017-02-17 Martin Liska <mliska@suse.cz>
33815 PR rtl-optimization/79577
33816 * params.def (selsched-max-sched-times): Increase minimum to 1.
33818 2017-02-17 Martin Liska <mliska@suse.cz>
33820 PR rtl-optimization/79574
33821 * gcse.c (want_to_gcse_p): Prevent integer overflow.
33823 2017-02-17 Martin Liska <mliska@suse.cz>
33825 PR tree-optimization/79529
33826 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
33827 ssa_defined_default_def_p to handle cases which are implicitly
33829 * tree-ssa.c (ssa_defined_default_def_p): New function.
33830 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
33831 which are implicitly defined.
33832 * tree-ssa.h (ssa_defined_default_def_p): Declare.
33834 2017-02-17 Richard Biener <rguenther@suse.de>
33836 PR middle-end/79576
33837 * params.def (max-ssa-name-query-depth): Limit to 10.
33839 2017-02-17 Richard Biener <rguenther@suse.de>
33841 PR tree-optimization/79552
33842 * tree-ssa-structalias.c (visit_loadstore): Properly verify
33845 2017-02-17 Richard Biener <rguenther@suse.de>
33848 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
33850 2017-02-17 Marek Polacek <polacek@redhat.com>
33852 PR middle-end/79536
33853 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
33854 (fold_negate_expr): New wrapper.
33856 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
33858 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
33859 Correct terminology and de-emphasize pre-standard behavior.
33861 2017-02-16 Alan Modra <amodra@gmail.com>
33863 PR rtl-optimization/79286
33864 * ira.c (def_dominates_uses): New function.
33865 (update_equiv_regs): Don't create an equivalence for insns that
33866 may trap where the register def does not dominate the use.
33868 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
33870 PR rtl-optimization/78127
33871 * lra.c (lra): Call lra_eliminate before finish the loop after
33874 2017-02-16 Richard Biener <rguenther@suse.de>
33876 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
33878 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
33879 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
33880 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
33881 (isl_val_int_from_wi): New function.
33882 (extract_affine_gmp): Rename to ...
33883 (extract_affine_wi): ... this, take a widest_int.
33884 (extract_affine_int): Just wrap extract_affine_wi.
33885 (add_param_constraints): Use isl_val_int_from_wi.
33886 (add_loop_constraints): Likewise, and extract_affine_wi.
33888 2017-02-15 Jeff Law <law@redhat.com>
33890 PR middle-end/79521
33891 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
33892 ira_init_register_move_cost_if_necessary.
33894 2017-02-15 Martin Sebor <msebor@redhat.com>
33896 PR middle-end/32003
33897 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
33898 removed in a prior commit.
33900 2017-02-15 Bin Cheng <bin.cheng@arm.com>
33902 PR tree-optimization/79347
33903 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
33904 counters during peeling.
33906 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
33908 * Makefile.in (site.exp): Remove "set ISLVER".
33910 2017-02-15 Jakub Jelinek <jakub@redhat.com>
33913 * real.c (real_from_integer): Call real_convert even for decimal.
33915 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
33918 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
33920 2017-02-14 Andrew Pinski <apinski@cavium.com>
33922 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
33923 cores and change the partno/implementer to be correct.
33924 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
33925 the 'B" as the implementer.
33926 * config/aarch64/aarch64-tune.md: Regenerate.
33928 2017-02-14 Carl Love <cel@us.ibm.com>
33930 * config/rs6000/rs6000.c: Add case statement entry to make the
33931 xvcvuxdsp built-in argument unsigned.
33932 * config/rs6000/vsx.md: Fix the source and return operand types so they
33933 match the instruction definitions from the ISA document. Fix typo
33934 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
33937 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
33940 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
33941 member early_clobber_alts.
33942 * lra-lives.c (reg_early_clobber_p): New.
33943 (process_bb_lives): Use it.
33944 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
33945 (debug_operand_data): Initialize early_clobber_alts.
33946 (setup_operand_alternative): Set up early_clobber_alts.
33947 (collect_non_operand_hard_regs): Ditto. Pass early clobber
33948 alternatives to new_insn_reg.
33949 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
33951 (lra_update_insn_regno_info): Pass the new arg.
33953 2017-02-14 Jakub Jelinek <jakub@redhat.com>
33955 PR middle-end/79505
33956 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
33957 (new_oacc_loop_raw): Don't clear already cleared fields.
33960 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
33961 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
33962 _mm512_prefetch_i64gather_ps): New inline functions and macros.
33964 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
33967 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
33969 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
33972 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
33973 the extra instruction to the right place to store 128-bit constant
33976 2017-02-14 Martin Sebor <msebor@redhat.com>
33978 PR middle-end/79448
33979 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
33980 warning for strings of unknown length.
33982 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
33984 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
33986 2017-02-14 Jeff Law <law@redhat.com>
33989 * ira-costs.c (scan_one_insn): Initialize register move costs
33990 for pseudos seen in USE/CLOBBER insns.
33992 PR tree-optimization/79095
33993 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
33994 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
33995 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
33996 if the operands are known to be not equal, then the resulting range
33998 (intersect_ranges): If the new range is ~[0,0] and the old range is
33999 wide, then prefer ~[0,0].
34000 * tree-vrp.c (overflow_comparison_p_1): New function.
34001 (overflow_comparison_p): New function.
34002 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
34003 if NAME is used in an overflow test.
34004 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
34005 overflow check that can be expressed as an equality test, then adjust
34006 ops to be that equality test.
34008 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34010 * config/s390/s390-builtin-types.def: Remove flags argument.
34011 * config/s390/s390.c (s390_init_builtins): Likewise.
34013 2017-02-14 Martin Liska <mliska@suse.cz>
34015 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
34016 vector. Fix trailing white spaces.
34018 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
34020 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
34023 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34025 PR rtl-optimization/68664
34026 * config/arm/arm.c (arm_sched_can_speculate_insn):
34027 New function. Declare prototype.
34028 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34030 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34032 PR rtl-optimization/68664
34033 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
34035 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
34037 2017-02-14 Amit Pawar <amit.pawar@amd.com>
34039 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
34040 max skip bytes for function, loop and jump.
34042 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
34044 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
34045 ABS_EXPR for gimple dump.
34047 2017-02-14 Jakub Jelinek <jakub@redhat.com>
34050 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
34052 PR tree-optimization/79408
34053 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
34054 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
34055 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
34056 also if rhs1 is INTEGER_CST.
34058 2017-02-14 Richard Biener <rguenther@suse.de>
34060 PR middle-end/79432
34061 * tree-into-ssa.c (insert_phi_nodes): When the function can
34062 have abnormal edges rewrite SSA names with broken use-def
34063 dominance out of SSA and register them for PHI insertion.
34065 2017-02-13 Martin Sebor <msebor@redhat.com>
34067 PR middle-end/79496
34068 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
34069 clearing info.nowrite flag when snprintf size argument is a range.
34071 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34073 * cprop.c (cprop_jump): Add missing space in string literal.
34074 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
34075 (get_constraint_for_component_ref): Likewise.
34076 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
34077 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
34078 * lra-constraints.c (process_alt_operands): Likewise.
34079 * ipa-inline.c (inline_small_functions): Likewise.
34080 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
34081 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
34082 * trans-mem.c (diagnose_tm_1_op): Likewise.
34083 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
34084 (grid_parallel_clauses_gridifiable): Likewise.
34086 * config/nvptx/mkoffload.c (process): Add space in between
34089 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
34090 "MOD4_SSE_REGS" and "ALL_REGS".
34092 * spellcheck.c (test_data): Add , in between "foo" and "food".
34094 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34097 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
34098 boundary crossing check and subsequent code generation agree.
34100 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34102 * config/aarch64/aarch64.c (has_memory_op): Delete.
34103 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
34106 2017-02-13 Jakub Jelinek <jakub@redhat.com>
34108 PR rtl-optimization/79388
34109 PR rtl-optimization/79450
34110 * combine.c (distribute_notes): When removing TEM_INSN for which
34111 corresponding dest has last value recorded, invalidate that last
34114 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34116 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
34117 of explicit '@'. Add missing assembly comment marker on branch costs
34120 2017-02-13 Nathan Sidwell <nathan@acm.org>
34122 * gengtype-lex.l (<in_struct>): Add '/'.
34124 2017-02-13 Martin Liska <mliska@suse.cz>
34127 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
34129 2017-02-13 Richard Biener <rguenther@suse.de>
34131 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
34133 * configure: Re-generate.
34134 * config.in: Likewise.
34135 * graphite-dependences.c: Simplify as if
34136 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
34137 * graphite-isl-ast-to-gimple.c: Likewise.
34138 * graphite-optimize-isl.c: Likewise.
34139 * graphite-poly.c: Likewise.
34140 * graphite-sese-to-poly.c: Likewise.
34141 * graphite.h: Likewise.
34142 * toplev.c: Include isl/version.h and use isl_version () for
34143 printing the ISL version.
34144 * doc/install.texi: Update ISL requirement.
34146 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34148 * doc/standards.texi (Standards): Update reference to
34151 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
34153 * doc/extend.texi (Named Address Spaces): sourceware.org now
34155 * doc/install.texi (Binaries): Ditto.
34158 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34160 * doc/cpp.texi: Replace "stringify"/"stringification" with C
34161 standard terminology "stringize"/"stringizing" throughout.
34162 * doc/cppinternals.texi: Likewise.
34164 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
34166 * doc/extend.texi: Fix some spelling mistakes and typos.
34167 * doc/invoke.texi: Likewise.
34169 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34172 * params.def (inline-min-speedup) Change from 10 to 8.
34174 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34176 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
34179 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
34182 * ipa-inline-analysis.c (get_minimal_bb): New function.
34183 (record_modified): Use it.
34184 (remap_edge_change_prob): Handle also ancestor functions.
34186 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
34188 * doc/contrib.texi (Contributors): Remove broken link into
34189 the Mauve CVS repository.
34191 2017-02-11 Jakub Jelinek <jakub@redhat.com>
34193 PR middle-end/79454
34194 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
34195 result computation whenever lhs doesn't have vector mode, not
34196 just when it has BLKmode.
34198 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34200 * doc/makefile.texi (profiledbootstrap): Refer to the
34201 installation instructions only in textual form.
34203 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34206 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
34208 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
34210 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
34211 (Specific): Update mingw-w64 reference.
34213 (Specific): Remove broken link to Renesas RX processor.
34215 2017-02-10 Richard Biener <rguenther@suse.de>
34217 * toplev.c (process_options): Do not mention obsolete graphite
34218 options when printing sorry message about missing graphite support.
34219 Mention -floop-nest-optimize.
34221 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
34223 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
34224 (vtst_p16): Likewise.
34225 (vtstq_p8): Likewise.
34226 (vtstq_p16): Likewise.
34228 (vtstq_p64): Likewise.
34229 * config/arm/arm_neon.h (vgetq_lane_p64): New.
34230 (vset_lane_p64): New.
34231 (vsetq_lane_p64): New.
34233 2017-02-10 Jakub Jelinek <jakub@redhat.com>
34235 PR tree-optimization/79411
34236 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
34237 stmt operands are SSA_NAMEs used in abnormal phis.
34238 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
34241 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34244 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
34247 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
34249 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
34251 2017-02-09 Jakub Jelinek <jakub@redhat.com>
34253 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
34257 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
34258 not arbitrary TREE_CONSTANT.
34261 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
34262 "omp declare target link" attribute unless is_global_var.
34263 * omp-offload.c (find_link_var_op): Likewise.
34265 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
34266 Chung-Lin Tang <cltang@codesourcery.com>
34268 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
34270 (gimplify_adjust_omp_clauses): Don't delete TILE.
34271 (gimplify_omp_for): Deal with TILE.
34272 * internal-fn.c (expand_GOACC_TILE): New function.
34273 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
34275 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
34276 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
34278 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
34279 avoid DIV for outermost collapse var.
34280 (expand_oacc_for): Insert tile element loop as needed. Adjust.
34281 Remove out of date comments, fix whitespace.
34282 * omp-general.c (omp_extract_for_data): Deal with tiling.
34283 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
34284 adjust OLF_DIM_BASE value.
34285 (struct omp_for_data): Add tiling field.
34286 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
34287 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
34288 for auto loops. Remove default auto determining, moved to
34289 oacc_loop_fixed_partitions.
34290 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
34291 stmts, add e_mask field.
34292 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
34293 (oacc_thread_numbers): Use oacc_dim_call.
34294 (oacc_xform_tile): New.
34295 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
34296 (finish_oacc_loop): Adjust for ifns vector.
34297 (oacc_loop_discover_walk): Append loop abstraction sites to list,
34298 add case for GOACC_TILE fns.
34299 (oacc_loop_xform_loop): Delete.
34300 (oacc_loop_process): Iterate over call list directly, and add
34301 handling for GOACC_TILE fns.
34302 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
34304 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
34305 vector partitioning to outer loops. Assign 2 partitions to loops
34306 when available. Add TILE handling.
34307 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
34308 (execite_oacc_device_lower): Process GOACC_TILE fns,
34309 ignore unknown specs.
34310 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
34311 * tree.c (omp_clause_num_ops): Adjust TILE ops.
34312 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
34314 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
34316 * configure.ac (ACX_BUGURL): Update.
34317 * configure: Regenerate.
34319 2017-02-09 Richard Biener <rguenther@suse.de>
34321 PR tree-optimization/69823
34322 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
34323 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
34325 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34327 * config/arc/arc-c.def: Add __NPS400__ definition.
34328 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
34329 (TARGET_NPS400): Define.
34331 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
34333 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
34335 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
34336 pointer, arch_info.
34337 (arc_cpu_types): Fill the arch_info field with a pointer into the
34338 arc_arch_types table.
34339 (arc_selected_cpu): Declare.
34340 * config/arc/arc.c (arc_selected_cpu): Make global.
34341 (arc_selected_arch): Delete.
34342 (arc_base_cpu): Delete.
34343 (arc_override_options): Remove references to deleted variables,
34344 update access to arch information.
34345 (ARC_OPT): Update access to arch information.
34346 (ARC_OPTX): Likewise.
34347 * config/arc/arc.h (arc_base_cpu): Remove declaration.
34348 (TARGET_ARC600): Update access to arch information.
34349 (TARGET_ARC601): Likewise.
34350 (TARGET_ARC700): Likewise.
34351 (TARGET_EM): Likewise.
34352 (TARGET_HS): Likewise.
34353 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
34356 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
34359 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
34360 condition/operands for integer GE/LE/GEU/LEU operations.
34362 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
34364 PR translation/79397
34365 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
34368 2017-02-08 Martin Jambor <mjambor@suse.cz>
34371 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
34372 whether allocation happened.
34373 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
34374 nothing was allocated.
34376 2017-02-08 Jakub Jelinek <jakub@redhat.com>
34378 PR tree-optimization/79408
34379 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
34380 constant, but SSA_NAME with a known integer range, use the minimum
34381 of that range instead of op1 to determine if modulo can be replaced
34382 with its first operand.
34384 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34386 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
34388 2017-02-08 Richard Biener <rguenther@suse.de>
34390 PR tree-optimization/71824
34391 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34392 Check all loops contained in the merged region.
34394 2017-02-07 Andrew Pinski <apinski@cavium.com>
34396 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
34398 2017-02-07 Andrew Pinski <apinski@cavium.com>
34400 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
34401 (thunderxt88): Likewise.
34402 (thunderxt81): Disable LSE and change v8.1 to v8.
34403 (thunderxt83): Likewise.
34405 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34406 Richard Biener <rguenther@suse.de>
34408 PR middle-end/79399
34409 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
34410 type from int to size_t.
34411 * ira-costs.c (struct_costs_size): Change type from int to size_t.
34413 2017-02-07 Jakub Jelinek <jakub@redhat.com>
34415 PR rtl-optimization/79386
34416 * cprop.c (bypass_conditional_jumps): Initialize
34417 bypass_last_basic_block already before splitting bbs after
34418 unconditional traps...
34419 (bypass_conditional_jumps): ... rather than here.
34422 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
34423 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
34424 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
34425 fix -masm=intel patterns.
34427 2017-02-07 Richard Biener <rguenther@suse.de>
34429 PR tree-optimization/79256
34430 PR middle-end/79278
34431 * builtins.c (get_object_alignment_2): Use min_align_of_type
34432 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
34433 and ADJUST_FIELD_ALIGN.
34435 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
34437 * doc/tm.texi: Regenerate.
34438 * stor-layout.c (layout_decl): Adjust.
34439 (update_alignment_for_field): Likewise.
34440 (place_field): Likewise.
34441 (min_align_of_type): Likewise.
34442 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
34443 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
34444 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
34445 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
34446 * config/frv/frv.c (frv_adjust_field_align): Likewise.
34447 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
34448 * config/i386/i386.c (x86_field_alignment): Likewise.
34449 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
34450 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
34451 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
34452 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
34453 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
34454 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
34458 2017-01-30 Richard Biener <rguenther@suse.de>
34460 PR tree-optimization/79256
34461 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
34462 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
34465 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
34467 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
34468 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
34469 builtins to SImode and emit a zero-extend, if necessary.
34471 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34473 * docs/invoke.texi (RISC-V Options): Alphabetize.
34475 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34477 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
34480 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
34482 * config/riscv/riscv.c: New file.
34483 * common/config/riscv/riscv-common.c: Likewise.
34484 * config.gcc: Likewise.
34485 * config/riscv/constraints.md: Likewise.
34486 * config/riscv/elf.h: Likewise.
34487 * config/riscv/generic.md: Likewise.
34488 * config/riscv/linux.h: Likewise.
34489 * config/riscv/multilib-generator: Likewise.
34490 * config/riscv/peephole.md: Likewise.
34491 * config/riscv/pic.md: Likewise.
34492 * config/riscv/predicates.md: Likewise.
34493 * config/riscv/riscv-builtins.c: Likewise.
34494 * config/riscv/riscv-c.c: Likewise.
34495 * config/riscv/riscv-ftypes.def: Likewise.
34496 * config/riscv/riscv-modes.def: Likewise.
34497 * config/riscv/riscv-opts.h: Likewise.
34498 * config/riscv/riscv-protos.h: Likewise.
34499 * config/riscv/riscv.h: Likewise.
34500 * config/riscv/riscv.md: Likewise.
34501 * config/riscv/riscv.opt: Likewise.
34502 * config/riscv/sync.md: Likewise.
34503 * config/riscv/t-elf-multilib: Likewise.
34504 * config/riscv/t-linux: Likewise.
34505 * config/riscv/t-linux-multilib: Likewise.
34506 * config/riscv/t-riscv: Likewise.
34507 * configure.ac: Likewise.
34508 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
34509 Waterman as RISC-V maintainers.
34510 * doc/install.texi: Add RISC-V entries.
34511 * doc/invoke.texi: Add RISC-V options section.
34512 * doc/md.texi: Add RISC-V constraints section.
34513 * configure: Regenerated.
34515 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
34518 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
34519 false values to be constant vectors with all 0 or all 1 bits set.
34520 (vcondu<mode><mode>): Likewise.
34521 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
34523 (fpmask_comparison_operator): Update comment.
34524 (vecint_comparison_operator): New predicate.
34525 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
34526 vector conditionals when the true and false values are constant
34527 vectors with all 0 bits or all 1 bits set.
34529 2017-02-06 Martin Sebor <msebor@redhat.com>
34531 PR tree-optimization/79376
34532 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
34534 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
34536 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
34537 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
34538 to simplify split condition.
34540 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34542 * omp-expand.c (oxpand_omp_atomic_fetch_op,
34543 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
34546 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
34548 PR rtl-optimization/68664
34549 * target.def (can_speculate_insn): New hook.
34550 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
34551 * doc/tm.texi: Regenerate.
34552 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
34553 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
34554 (rs6000_sched_can_speculate_insn): New function.
34556 2017-02-06 Jakub Jelinek <jakub@redhat.com>
34558 PR tree-optimization/79284
34559 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
34560 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
34561 vectorizable_mask_load_store, vectorizable_operation,
34562 vect_is_simple_cond, get_same_sized_vectype): Use it instead
34563 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
34564 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
34565 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
34566 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
34567 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
34568 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
34569 is_gimple_assign (stmt). Replace another such test with
34570 is_gimple_assign (stmt).
34572 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34575 * config/avr/avr.c (rtl-iter.h): Include it.
34576 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
34577 (avr_legitimate_combined_insn): ...and implementation.
34579 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
34581 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
34582 * config/s390/s390.c (s390_const_operand_ok)
34583 (s390_canonicalize_comparison, s390_extract_part)
34584 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
34585 (s390_contiguous_bitmask_p, s390_rtx_costs)
34586 (legitimize_pic_address): Likewise.
34587 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
34588 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
34589 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
34590 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
34591 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
34593 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
34595 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
34596 REGNO($0) == REGNO($1).
34598 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34600 * config/s390/linux.h(SIZE_TYPE): Add comment.
34602 2017-02-06 Julian Brown <julian@codesourcery.com>
34603 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34604 Virendra Pathak <virendra.pathak@broadcom.com>
34606 * config/aarch64/aarch64-cores.def: Change the scheduler
34608 * config/aarch64/aarch64.md: Include thunderx2t99.md.
34609 * config/aarch64/thunderx2t99.md: New file.
34611 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34613 * doc/standards.texi (Go Language): Update link to language
34616 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
34618 * tree-eh.c (lower_resx): Sanitize profile.
34619 (cleanup_empty_eh_move_lp): Likewise.
34621 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
34624 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
34626 * cfgloopmanip.h (loop_version): Update prototype.
34627 * modulo-sched.c (sms_schedule): Update call of loop_version.
34628 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
34629 * tree-parloops.c (gen_parallel_loop): Likewise.
34630 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
34631 * tree-ssa-loop-split.c (split_loop): Likewise.
34632 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
34633 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
34635 2017-02-05 Martin Liska <mliska@suse.cz>
34638 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
34640 (print_operand_address): Initialize a struct to zero.
34642 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34644 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
34645 garbage collector only in textual form.
34647 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
34649 * doc/extend.texi (x86 specific memory model extensions for
34650 transactional memory): Simplify a phrase.
34652 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
34655 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
34656 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
34657 (atomic_storedi_1): Likewise.
34659 2017-02-04 Jakub Jelinek <jakub@redhat.com>
34661 PR tree-optimization/79338
34662 * tree-parloops.c (gather_scalar_reductions): Don't call
34663 vect_analyze_loop_form for loop->inner before destroying loop's
34666 2017-02-03 Martin Sebor <msebor@redhat.com>
34668 PR tree-optimization/79327
34669 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
34670 when precision has resulted in leading zeros.
34671 (format_integer): Adjust the likely counter to assume an unknown
34672 argument that may be zero is non-zero.
34674 2017-02-03 Jason Merrill <jason@redhat.com>
34677 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
34678 avoid copying non-taken branch.
34680 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34682 PR tree-optimization/79340
34683 * tree-vect-loop.c (vectorizable_reduction): Release
34684 vec_defs elements after safe_splicing them into other vectors.
34687 PR tree-optimization/79327
34688 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
34689 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
34691 (format_integer): Use wide_int_to_tree instead of build_int_cst
34692 + to_?hwi. If argmin is NULL, just set argmin and argmax to
34693 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
34694 of shortest and longest sequence.
34696 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
34698 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
34699 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
34701 2017-02-03 Walter Lee <walt@tilera.com>
34704 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
34705 after initial stackframe link reg save.
34706 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
34708 2017-02-03 Jakub Jelinek <jakub@redhat.com>
34711 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
34712 wu for stxssp alternative.
34714 2017-02-03 Martin Sebor <msebor@redhat.com>
34716 PR tree-optimization/79352
34717 * gimple-fold.c (get_range_strlen): Add argument.
34718 (get_range_strlen): Change return type to bool.
34719 (get_maxval_strlen): Pass in a dummy argument.
34720 * gimple-fold.h (get_range_strlen): Change return type to bool.
34721 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
34722 * tree.h (array_at_struct_end_p): Add argument.
34723 * tree.c (array_at_struct_end_p): Handle it.
34725 2017-02-03 Martin Liska <mliska@suse.cz>
34728 * multiple_target.c (create_dispatcher_calls): Redirect edge
34729 from a caller of a dispatcher.
34730 (expand_target_clones): Make the clones local.
34731 (ipa_target_clone): Do both target clones and resolvers.
34732 (ipa_dispatcher_calls): Remove the pass.
34733 (pass_dispatcher_calls::gate): Likewise.
34734 (make_pass_dispatcher_calls): Likewise.
34735 * passes.def (pass_target_clone): Put as very first IPA early
34738 2017-02-03 Martin Liska <mliska@suse.cz>
34740 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
34741 in case of a function with ifunc attribute.
34743 2017-02-03 Martin Liska <mliska@suse.cz>
34745 * cgraph.c (cgraph_node::dump): Dump function version info.
34746 * symtab.c (symtab_node::dump_base): Add missing new line.
34748 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34750 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
34751 (ifcombine_ifandif): Use it.
34753 2017-02-03 Martin Liska <mliska@suse.cz>
34755 * doc/invoke.texi: Document default value for
34756 use-after-scope-direct-emission-threshold.
34758 2017-02-03 Martin Liska <mliska@suse.cz>
34760 PR tree-optimization/79339
34761 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
34762 (format_floating): Likewise.
34764 2017-02-03 Martin Liska <mliska@suse.cz>
34767 * ipa-prop.c (ipa_node_params_t::insert): Remove current
34769 (ipa_node_params_t::remove): Likewise.
34770 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
34771 initialization from removed ipa_node_params_t::insert.
34772 (ipa_node_params::~ipa_node_params): Move from removed
34773 ipa_node_params_t::release.
34774 * symbol-summary.h (symbol_summary::m_released): New member.
34775 Do not release a summary twice. Do not allow to call finalizer
34776 for types of a summary that live in GGC memory.
34778 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34780 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
34783 2017-02-02 Martin Sebor <msebor@redhat.com>
34785 PR middle-end/79275
34786 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
34787 (format_string): Tighten up the range of output for non-constant
34788 strings and correct the expected range for wide non-constant strings.
34790 2017-02-02 Martin Sebor <msebor@redhat.com>
34792 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
34794 PR middle-end/32003
34795 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
34797 (-fdump-tree-@var): Add to index and document how to come up
34798 with pass-specific option and dump file names.
34799 (-fdump-passes): Clarify where to look for output.
34801 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
34803 PR middle-end/77445
34804 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
34805 statistics of the analyzed path; allow threading for speed when
34806 any of BBs along the path are optimized for speed.
34808 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
34810 PR middle-end/78468
34811 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
34812 settings of the virtual registers.
34815 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
34817 * explow.c (get_dynamic_stack_size): Take known alignment of stack
34818 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
34821 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34823 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
34824 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
34826 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34828 * config/s390/s390.md: Add missing comments with the expanded
34830 * config/s390/vector.md: Likewise.
34831 * config/s390/vx-builtins.md: Likewise.
34833 2017-02-02 Jakub Jelinek <jakub@redhat.com>
34836 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
34837 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
34838 conditions on a single line.
34840 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34842 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34843 __S390_VX__ to __VX__.
34845 2017-02-01 Andrew Pinski <apinski@cavium.com>
34847 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
34848 stmt_info to record_stmt_cost.
34849 (vect_get_known_peeling_cost): Pass stmt_info if known to
34851 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
34852 cpu_vector_cost field into
34853 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
34854 field into vec_int_stmt_cost and vec_fp_stmt_cost.
34855 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
34856 splitting of scalar_stmt_cost and vec_stmt_cost.
34857 (thunderx_vector_cost): Likewise.
34858 (cortexa57_vector_cost): LIkewise.
34859 (exynosm1_vector_cost): Likewise.
34860 (xgene1_vector_cost): Likewise.
34861 (thunderx2t99_vector_cost): Improve after the splitting of the two
34863 (aarch64_builtin_vectorization_cost): Update for the splitting of
34864 scalar_stmt_cost and vec_stmt_cost.
34866 2017-02-01 Torvald Riegel <triegel@redhat.com>
34867 Richard Henderson <rth@redhat.com>
34869 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
34870 conditional on existance of a fast atomic load.
34871 * optabs-query.c (can_atomic_load_p): New function.
34872 * optabs-query.h (can_atomic_load_p): Declare it.
34873 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
34874 no fast atomic load is available for the particular size of access.
34875 (expand_atomic_compare_and_swap): Likewise.
34876 (expand_atomic_load): Likewise.
34877 (expand_atomic_store): Likewise.
34878 (expand_atomic_fetch_op): Likewise.
34879 * testsuite/lib/target-supports.exp
34880 (check_effective_target_sync_int_128): Remove x86 because it provides
34881 no fast atomic load.
34882 (check_effective_target_sync_int_128_runtime): Likewise.
34884 2017-02-01 Richard Biener <rguenther@suse.de>
34886 * graphite.c: Include tree-vectorizer.h for find_loop_location.
34887 (graphite_transform_loops): Provide opt-info for optimized nests.
34888 * tree-parloop.c (parallelize_loops): Provide opt-info for
34889 parallelized loops.
34891 2017-02-01 Richard Biener <rguenther@suse.de>
34893 PR middle-end/79315
34894 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
34895 was not set before.
34897 2017-02-01 Richard Biener <rguenther@suse.de>
34899 PR tree-optimization/71824
34900 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
34901 Verify the loops are valid in the merged SESE region.
34902 (scop_detection::can_represent_loop_1): Check analyzing the
34903 evolution of the number of iterations in the region succeeds.
34905 2017-01-31 Ian Lance Taylor <iant@golang.org>
34907 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
34908 REG_ARGS_SIZE note to 32-bit push insns and call insn.
34910 2017-01-31 David Malcolm <dmalcolm@redhat.com>
34912 PR preprocessor/79210
34913 * input.c (get_substring_ranges_for_loc): Replace line_width
34914 assertion with error-handling.
34916 2017-01-31 Richard Biener <rguenther@suse.de>
34918 PR tree-optimization/77318
34919 * graphite-sese-to-poly.c (extract_affine): Fix assert.
34920 (create_pw_aff_from_tree): Take loop parameter.
34921 (add_condition_to_pbb): Pass loop of the condition to
34922 create_pw_aff_from_tree.
34924 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34926 * config/s390/s390.c (s390_asan_shadow_offset): New function.
34927 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
34929 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
34933 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
34935 (convert_int_to_float128): Likewise.
34936 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
34937 (convert_int_to_float128): Likewise.
34938 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
34939 (UNSPEC_IEEE128_CONVERT): Likewise.
34940 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
34941 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
34942 Use local variables for IBM extended format.
34943 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
34944 (fix_trunc<mode>si2_fprs): Likewise.
34945 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
34946 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
34947 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
34948 to know that we can now have integers of all sizes in vector
34950 (fix<uns>_<mode>di2_hw): Likewise.
34951 (float<uns>_<mode>si2_hw): Likewise.
34952 (fix_<mode>si2_hw): Likewise.
34953 (fixuns_<mode>si2_hw): Likewise.
34954 (float<uns>_<mode>di2_hw): Likewise.
34955 (float_<mode>di2_hw): Likewise.
34956 (float_<mode>si2_hw): Likewise.
34957 (floatuns_<mode>di2_hw): Likewise.
34958 (floatuns_<mode>si2_hw): Likewise.
34959 (xscvqp<su>wz_<mode>): Delete, no longer used.
34960 (xscvqp<su>dz_<mode>): Likewise.
34961 (xscv<su>dqp_<mode>): Likewise.
34962 (ieee128_mfvsrd_64bit): Likewise.
34963 (ieee128_mfvsrd_32bit): Likewise.
34964 (ieee128_mfvsrwz): Likewise.
34965 (ieee128_mtvsrw): Likewise.
34966 (ieee128_mtvsrd_64bit): Likewise.
34967 (ieee128_mtvsrd_32bit): Likewise.
34969 2017-01-31 Martin Liska <mliska@suse.cz>
34972 * ipa-prop.c (ipa_free_all_node_params): Call release method
34973 instead of ~sumbol_summary to not to trigger double times
34976 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
34978 PR tree-optimization/71691
34979 * bitmap.h (class auto_bitmap): New.
34980 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
34981 is_maybe_undefined instead of ssa_undefined_value_p.
34983 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34985 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
34986 __S390_ARCH_LEVEL__ to __ARCH__.
34988 2017-01-31 Jakub Jelinek <jakub@redhat.com>
34990 PR tree-optimization/79267
34991 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
34992 if should_remove_lhs_p is true.
34994 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
34997 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
34998 (add_alignment_attribute): New.
34999 (base_type_die): Add alignment attribute.
35000 (subrange_type_die): Likewise.
35001 (modified_type_die): Likewise.
35002 (gen_array_type_die): Likewise.
35003 (gen_descr_array_type_die: Likewise.
35004 (gen_enumeration_type_die): Likewise.
35005 (gen_subprogram_die): Likewise.
35006 (gen_variable_die): Likewise.
35007 (gen_field_die): Likewise.
35008 (gen_ptr_to_mbr_type_die): Likewise.
35009 (gen_struct_or_union_type_die): Likewise.
35010 (gen_subroutine_type_die): Likewise.
35011 (gen_typedef_die): Likewise.
35012 (base_type_cmp): Compare alignment attribute.
35014 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35017 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
35018 (setb_unsigned) New pattern for setb with CCUNS.
35019 * config/rs6000/rs6000.c (expand_block_compare): Use a different
35020 subfc./subfe sequence to avoid overflow problems. Generate a
35021 shorter sequence with cmpld/setb for power9.
35022 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
35023 for generating subfc. instruction.
35024 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
35025 now uses this instruction.
35027 2017-01-30 Ian Lance Taylor <iant@google.com>
35030 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
35031 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
35033 2017-01-30 Martin Sebor <msebor@redhat.com>
35035 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
35036 Move constant to the right of a relational operator.
35037 (get_mpfr_format_length, format_character, format_string): Ditto.
35038 (should_warn_p, maybe_warn): Same.
35040 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
35042 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
35045 * asan.c (get_translation_unit_decl): Remove function.
35046 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
35048 2017-01-30 Martin Liska <mliska@suse.cz>
35050 PR gcov-profile/79259
35051 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
35052 -fprofile-generate.
35054 2017-01-30 Martin Liska <mliska@suse.cz>
35057 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
35058 Initialize variables with NULL value.
35060 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
35063 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
35065 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
35067 2017-01-30 Richard Biener <rguenther@suse.de>
35069 PR tree-optimization/79276
35070 * tree-vrp.c (process_assert_insertions): Properly adjust common
35071 when removing a duplicate.
35073 2017-01-30 Richard Biener <rguenther@suse.de>
35075 PR tree-optimization/79256
35076 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
35077 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
35079 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
35081 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
35084 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
35085 ("*r<noxa>sbg_<mode>_sll_bitmask")
35086 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
35087 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
35088 Use contiguous_bitmask_nowrap_operand.
35090 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35093 * config/rs6000/altivec.h (vec_xl): Revise #define.
35094 (vec_xst): Likewise.
35096 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
35098 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
35100 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
35102 PR rtl-optimization/79194
35103 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
35104 traps before call to bypass_conditional_jumps.
35106 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35108 PR tree-optimization/71374
35109 * lra-constraints.c (check_conflict_input_operands): New.
35110 (match_reload): Use it.
35112 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
35115 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
35116 account to calculate conflict_set.
35118 2017-01-27 Bin Cheng <bin.cheng@arm.com>
35120 PR rtl-optimization/78559
35121 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
35122 other_insn in combine.
35124 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
35126 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
35127 uint16_type_node for BT_UINT16.
35129 2017-01-27 David Malcolm <dmalcolm@redhat.com>
35131 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
35132 "RTL Tests" to menu.
35133 (GIMPLE Tests): New node.
35134 (RTL Tests): New node.
35136 2017-01-27 Richard Biener <rguenther@suse.de>
35138 PR tree-optimization/79245
35139 * tree-loop-distribution.c (distribute_loop): Apply cost
35140 modeling also to detected patterns.
35142 2017-01-27 Richard Biener <rguenther@suse.de>
35144 PR tree-optimization/71433
35145 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
35146 (compare_assert_loc): New function.
35147 (process_assert_insertions): Sort and optimize assert locations
35148 to remove duplicates and push down identical assertions on
35149 edges to their destination block.
35151 2017-01-27 Richard Biener <rguenther@suse.de>
35153 PR tree-optimization/79244
35154 * tree-vrp.c (remove_range_assertions): Forcefully propagate
35155 out SSA names even if abnormal.
35157 2017-01-27 Jakub Jelinek <jakub@redhat.com>
35159 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
35160 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
35161 instead of MPFR_RNDN.
35163 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
35166 * arm.c (arm_option_override): Don't call build_target_option_node
35167 until after doing all option overrides.
35168 (arm_valid_target_attribute_tree): Likewise.
35170 2017-01-27 Martin Liska <mliska@suse.cz>
35172 * doc/invoke.texi (-fprofile-arcs): Document profiling support
35173 for {cd}tors and C++ {cd}tors.
35175 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
35177 * config/s390/s390.md ("*setmem_long_and")
35178 ("*setmem_long_and_31z"): Use zero_extend instead of and.
35180 2017-01-26 Martin Sebor <msebor@redhat.com>
35182 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
35185 2017-01-26 Martin Sebor <msebor@redhat.com>
35187 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
35188 HAVE_DFmode before using XFmode or DFmode.
35189 (parse_directive): Avoid using the z length modifier to avoid
35190 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
35192 PR middle-end/78703
35193 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
35194 to accept adjustment as an array.
35195 (get_int_range): New function.
35196 (struct directive): Make width and prec arrays.
35197 (directive::set_width, directive::set_precision): Call get_int_range.
35198 (format_integer, format_floating): Handle width and precision ranges.
35199 (format_string, parse_directive): Same.
35201 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35204 * dwarf2out.c (generate_skeleton_bottom_up): For children with
35205 comdat_type_p set, just clone them, but keep the children in the
35209 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
35210 which have direct callers with -fvar-tracking-assignments enabled
35212 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
35213 inside of type units.
35215 2017-01-26 Martin Sebor <msebor@redhat.com>
35217 PR middle-end/78703
35218 * gimple-ssa-sprintf.c (struct result_range): Add likely and
35220 (struct format_result): Replace number_chars, number_chars_min,
35221 and number_chars_max with a single member of struct result_range.
35223 (format_result::operator+=): Adjust.
35224 (struct fmtresult): Remove bounded. Handle likely and unlikely
35226 (fmtresult::adjust_for_width_or_precision): New function.
35227 (fmtresult:type_max_digits): New function.
35228 (bytes_remaining): Handle likely and unlikely counters.
35229 (min_bytes_remaining): Remove.
35230 (format_percent): Simplify.
35231 (format_integer, format_floating): Set likely and unlikely counters.
35232 (get_string_length, format_character, format_string): Same.
35233 (format_plain, should_warn_p): New function.
35234 (maybe_warn): Call should_warn_p. Update diagnostic messages
35235 and handle those for all directives, including plain strings.
35236 (format_directive): Handle likely and unlikely counters.
35237 Remove unnecessary quoting from diagnostics. Add an informational
35239 (add_bytes): Remove.
35240 (pass_sprintf_length::compute_format_length): Simplify.
35241 (try_substitute_return_value): Handle likely and unlikely counters.
35243 2017-01-26 Carl Love <cel@us.ibm.com>
35245 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
35246 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
35248 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
35251 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
35252 endianess for subregs into account.
35253 * lra-constraints.c (lra_constraints): Do risky transformations
35254 always on the first iteration.
35255 * lra-lives.c (check_pseudos_live_through_calls): Add arg
35256 last_call_used_reg_set.
35257 (process_bb_lives): Define and use last_call_used_reg_set.
35258 * lra.c (lra): Always continue after lra_constraints on the first
35261 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
35263 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
35265 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
35267 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35269 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
35270 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
35271 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
35272 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
35273 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
35274 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
35275 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
35276 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
35277 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
35279 2017-01-26 Marek Polacek <polacek@redhat.com>
35282 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
35283 for the third operand.
35285 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35287 PR middle-end/79236
35288 * omp-low.c (struct omp_context): Add simt_stmt field.
35289 (scan_omp_for): Return omp_context *.
35290 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
35291 context to the _simt_ SIMD stmt.
35292 (lower_omp_for): For combined SIMD with sibling _simt_
35293 SIMD, make sure to use the same decls in _looptemp_
35294 clauses as in the sibling.
35296 2017-01-26 David Sherwood <david.sherwood@arm.com>
35298 PR middle-end/79212
35299 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
35302 2017-01-26 Jakub Jelinek <jakub@redhat.com>
35305 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
35306 emit fld b; fld a; if possible.
35308 * brig-builtins.def: Update copyright years.
35309 * config/arm/arm_acle_builtins.def: Update copyright years.
35311 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
35314 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
35315 constraint instead of o for the stxsd instruction.
35317 2017-01-25 Carl Love <cel@us.ibm.com>
35319 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
35320 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
35322 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
35324 * doc/invoke.texi (C++ Dialect Options): Fix typo.
35326 2017-01-25 Richard Biener <rguenther@suse.de>
35328 PR tree-optimization/69264
35329 * target.def (vector_alignment_reachable): Improve documentation.
35330 * doc/tm.texi: Regenerate.
35331 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
35333 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
35334 earlier changes with respect to TYPE_USER_ALIGN.
35335 (vector_alignment_reachable_p): Likewise. Improve dumping.
35337 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35340 * config/arm/arm.md (xordi3): Force constant operand into a register
35343 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35345 * doc/invoke.texi (-fstore-merging): Correct default optimization
35346 levels at which it is enabled.
35347 (-O): Move -fstore-merging from list to...
35350 2017-01-25 Richard Biener <rguenther@suse.de>
35353 * omp-expand.c: Include debug.h.
35354 (expand_omp_taskreg): Make sure to generate early debug before
35355 outlining anything from a function.
35356 (expand_omp_target): Likewise.
35357 (grid_expand_target_grid_body): Likewise.
35359 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
35362 * asan.c (get_translation_unit_decl): New function.
35363 (asan_add_global): Extract modules file name from globals
35364 TRANSLATION_UNIT_DECL name.
35366 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
35369 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
35370 for long calls with APCS frame and VFP.
35372 2017-01-24 David Malcolm <dmalcolm@redhat.com>
35374 * cfg.c (original_copy_tables_initialized_p): New function.
35375 * cfg.h (original_copy_tables_initialized_p): New decl.
35376 * cfgrtl.c (relink_block_chain): Guard the call to
35377 free_original_copy_tables with a call to
35378 original_copy_tables_initialized_p.
35379 * cgraph.h (symtab_node::native_rtl_p): New decl.
35380 * cgraphunit.c (symtab_node::native_rtl_p): New function.
35381 (symtab_node::needed_p): Don't assert for early assembly output
35382 for __RTL functions.
35383 (cgraph_node::finalize_function): Set "force_output" for __RTL
35385 (cgraph_node::analyze): Bail out early for __RTL functions.
35386 (analyze_functions): Update assertion to support __RTL functions.
35387 (cgraph_node::expand): Bail out early for __RTL functions.
35388 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
35390 * function.h (struct function): Update comment for field
35392 * gimple-expr.c: Include "tree-pass.h".
35393 (gimple_has_body_p): Return false for __RTL functions.
35394 * Makefile.in (OBJS): Add run-rtl-passes.o.
35395 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
35397 (gcc::pass_manager::get_clean_slate): New accessor.
35398 * passes.c: Include "insn-addr.h".
35399 (should_skip_pass_p): Add logging. Update logic for running
35400 "expand" to be compatible with both __GIMPLE and __RTL. Guard
35401 property-provider override so it is only done for gimple passes.
35403 (skip_pass): New function.
35404 (execute_one_pass): Call skip_pass when skipping passes.
35405 * read-md.c (md_reader::read_char): Support filtering
35406 the input to a subset of line numbers.
35407 (md_reader::md_reader): Initialize fields
35408 m_first_line and m_last_line.
35409 (md_reader::read_file_fragment): New function.
35410 * read-md.h (md_reader::read_file_fragment): New decl.
35411 (md_reader::m_first_line): New field.
35412 (md_reader::m_last_line): New field.
35413 * read-rtl-function.c (function_reader::create_function): Only
35414 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
35415 curr_properties. Set DECL_INITIAL to a dummy block.
35416 (read_rtl_function_body_from_file_range): New function.
35417 * read-rtl-function.h (read_rtl_function_body_from_file_range):
35419 * run-rtl-passes.c: New file.
35420 * run-rtl-passes.h: New file.
35422 2017-01-24 Jeff Law <law@redhat.com>
35424 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
35427 2017-01-24 Bin Cheng <bin.cheng@arm.com>
35429 PR tree-optimization/79159
35430 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
35431 (record_nonwrapping_iv): Improve boundary using above function if no
35432 value range information.
35434 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
35435 Martin Jambor <mjambor@suse.cz>
35437 * brig-builtins.def: New file.
35438 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
35439 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
35440 (DEF_HSAIL_SAT_BUILTIN): Likewise.
35441 (DEF_HSAIL_INTR_BUILTIN): Likewise.
35442 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
35443 * builtin-types.def (BT_INT8): New.
35444 (BT_INT16): Likewise.
35445 (BT_UINT8): Likewise.
35446 (BT_UINT16): Likewise.
35447 (BT_FN_ULONG): Likewise.
35448 (BT_FN_UINT_INT): Likewise.
35449 (BT_FN_UINT_ULONG): Likewise.
35450 (BT_FN_UINT_LONG): Likewise.
35451 (BT_FN_UINT_PTR): Likewise.
35452 (BT_FN_ULONG_PTR): Likewise.
35453 (BT_FN_INT8_FLOAT): Likewise.
35454 (BT_FN_INT16_FLOAT): Likewise.
35455 (BT_FN_UINT32_FLOAT): Likewise.
35456 (BT_FN_UINT16_FLOAT): Likewise.
35457 (BT_FN_UINT8_FLOAT): Likewise.
35458 (BT_FN_UINT64_FLOAT): Likewise.
35459 (BT_FN_UINT16_UINT32): Likewise.
35460 (BT_FN_UINT32_UINT16): Likewise.
35461 (BT_FN_UINT16_UINT16_UINT16): Likewise.
35462 (BT_FN_INT_PTR_INT): Likewise.
35463 (BT_FN_UINT_PTR_UINT): Likewise.
35464 (BT_FN_LONG_PTR_LONG): Likewise.
35465 (BT_FN_ULONG_PTR_ULONG): Likewise.
35466 (BT_FN_VOID_UINT64_UINT64): Likewise.
35467 (BT_FN_UINT8_UINT8_UINT8): Likewise.
35468 (BT_FN_INT8_INT8_INT8): Likewise.
35469 (BT_FN_INT16_INT16_INT16): Likewise.
35470 (BT_FN_INT_INT_INT): Likewise.
35471 (BT_FN_UINT_FLOAT_UINT): Likewise.
35472 (BT_FN_FLOAT_UINT_UINT): Likewise.
35473 (BT_FN_ULONG_UINT_UINT): Likewise.
35474 (BT_FN_ULONG_UINT_PTR): Likewise.
35475 (BT_FN_ULONG_ULONG_ULONG): Likewise.
35476 (BT_FN_UINT_UINT_UINT): Likewise.
35477 (BT_FN_VOID_UINT_PTR): Likewise.
35478 (BT_FN_UINT_UINT_PTR: Likewise.
35479 (BT_FN_UINT32_UINT64_PTR): Likewise.
35480 (BT_FN_INT_INT_UINT_UINT): Likewise.
35481 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
35482 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
35483 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
35484 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
35485 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
35486 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
35487 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
35488 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
35489 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
35490 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
35491 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
35492 * doc/frontends.texi: List BRIG FE.
35493 * doc/install.texi (Testing): Add BRIG tesring requirements.
35494 * doc/invoke.texi (Overall Options): Mention BRIG.
35495 * doc/standards.texi (Standards): Doucment BRIG HSA version.
35497 2017-01-24 Richard Biener <rguenther@suse.de>
35499 PR translation/79208
35500 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
35502 2017-01-24 Martin Jambor <mjambor@suse.cz>
35505 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
35506 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
35507 and known_contexts.
35509 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
35511 PR middle-end/79123
35512 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
35513 casts from signed to unsigned really don't have a range.
35515 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
35517 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
35518 GMP_RNDx for compatiblity.
35520 2017-01-24 Martin Liska <mliska@suse.cz>
35523 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
35524 that would prevent us to call alloca with -1 as argument.
35526 2017-01-24 Jakub Jelinek <jakub@redhat.com>
35528 * dwarf2out.c (output_compilation_unit_header, output_file_names):
35529 Avoid -Wformat-security warning.
35531 2017-01-23 Andrew Pinski <apinski@cavium.com>
35533 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
35536 2017-01-23 Martin Sebor <msebor@redhat.com>
35538 PR middle-end/78703
35539 * gimple-ssa-sprintf.c (warn_level): New global.
35540 (format_integer): Use it here and throughout the rest of the file.
35541 Use the same switch to compute sign as base.
35542 (maybe_warn): New function.
35543 (format_directive): Factor out warnings into maybe_warn.
35544 Add debugging output. Use warn_level.
35545 (add_bytes): Use warn_level.
35546 (pass_sprintf_length::compute_format_length): Add debugging output.
35547 (try_substitute_return_value): Same.
35548 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
35550 PR middle-end/78703
35551 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
35552 (struct fmtresult, format_integer, format_floating): Adjust.
35553 (fmtresult::fmtresult): Set max correctly in two argument ctor.
35554 (get_string_length, format_string,format_directive): Same.
35555 (pass_sprintf_length::compute_format_length): Same.
35556 (try_substitute_return_value): Simplify slightly.
35558 PR middle-end/78703
35559 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
35560 (fmtresult::operator+=): Outlined.
35561 (struct fmtresult): Add ctors.
35562 (struct conversion_spec): Rename...
35563 (struct directive): ...to this. Add and remove data members.
35564 (directive::set_width, directive::set_precision): New functions.
35565 (format_percent): Use fmtresult ctor.
35566 (get_width_and_precision): Remove.
35567 (format_integer): Make naming changes. Avoid computing width and
35569 (format_floating): Same. Adjust indentation.
35570 (format_character, format_none): New functions.
35571 (format_string): Moved character handling to format_character.
35572 (format_directive): Remove arguments, change return type.
35573 (parse_directive): New function.
35574 (pass_sprintf_length::compute_format_length): Move directive
35575 parsing to parse_directive.
35577 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35579 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
35580 (assign_assembler_name_if_needed): ... this.
35581 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
35582 (assign_assembler_name_if_needed): ... this.
35583 (free_lang_data_in_cgraph): Adjust callers.
35584 * cgraphunit.c (cgraph_node::analyze): Likewise.
35585 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
35588 2017-01-23 Richard Biener <rguenther@suse.de>
35590 PR tree-optimization/79088
35591 PR tree-optimization/79188
35592 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
35593 resetting loop bounds after last path deletion. Reset loop
35594 bounds of the target loop, make code match the comments.
35595 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
35596 Make sure loops need no fixups.
35598 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35600 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
35601 exponent support with double type for first argument.
35602 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
35603 type returned by __builtin_vec_extract_sig,
35604 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
35605 functions from "vector int" to "vector unsigned int" or from
35606 "vector long long int" to "vector unsigned long long int".
35607 Changed type returned by __builtin_vec_extract_exp,
35608 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
35609 functions from "vector int" to "vector unsigned int" or from
35610 "vector long long int" to "vector unsigned long long int".
35611 Changed return type of __builtin_vec_test_data_class,
35612 __builtin_vec_test_data_class_sp, and
35613 __builtin_vec_test_data_class_dp from "vector int" to
35614 "vector bool int" or from "vector long long int" to "vector bool
35615 long long int" and changed second argument type from "unsigned
35616 int" to "int". Added new overloaded function forms "vector float
35617 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
35618 "vector float __builtin_vec_insert_exp_sp (vector float, vector
35619 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
35620 double, vector unsigned long long int)" and "vector double
35621 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
35622 long int)". Changed return type of
35623 __builtin_scalar_test_data_class and
35624 __builtin_scalar_test_data_class_sp and
35625 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
35626 int" and changed second argument from "unsigned int" to "int".
35627 Changed type returned by __builtin_scalar_test_neg,
35628 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
35629 from "int" to "bool int". Added new overloaded function form
35630 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
35631 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
35632 exponent double-precision with floating point first argument.
35633 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
35634 documentation of scalar_test_data_class, scalar_test_neg,
35635 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
35636 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
35637 vec_test_data_class built-in functions to reflect refinements in
35638 their type signatures.
35640 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
35642 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
35644 (aarch64_elf_asm_destructor): Likewise.
35646 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
35648 PR rtl-optimization/78634
35649 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
35650 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
35651 * ifcvt.c (noce_try_cmove): Add missing cost check.
35653 PR rtl-optimization/71724
35654 * combine.c (if_then_else_cond): Look for situations where it is
35655 beneficial to undo the work of one of the recursive calls.
35657 2017-01-23 Bin Cheng <bin.cheng@arm.com>
35659 PR tree-optimization/70754
35660 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
35661 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
35662 combined stmt before it if not NULL.
35663 (combine_chains): Process refs reversely and compute dominance point
35666 2017-01-23 Martin Liska <mliska@suse.cz>
35668 PR tree-optimization/79196
35669 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
35670 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
35672 (strlen_optimize_stmt): Call the renamed function.
35674 2017-01-23 Michael Matz <matz@suse.de>
35676 PR tree-optimization/78384
35677 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
35679 2017-01-23 Richard Biener <rguenther@suse.de>
35681 PR tree-optimization/79186
35682 * tree-vrp.c (register_new_assert_for): Make sure we've seen
35683 both incoming edges before moving an assert.
35685 2017-01-23 Martin Jambor <mjambor@suse.cz>
35687 * ipa-prop.c (load_from_param_1): Removed.
35688 (load_from_unmodified_param): Bits from load_from_param_1 put back
35690 (load_from_param): Removed.
35691 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
35692 with stmt. Reverted back to use of load_from_unmodified_param.
35694 2017-01-23 Martin Jambor <mjambor@suse.cz>
35697 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
35698 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
35699 field a pointer to garbage collected vector, mark lattices and
35700 ipcp_orig_node with GTY((skip)).
35701 (ipa_get_param_count): Adjust to descriptors being a pointer.
35702 (ipa_get_param): Likewise.
35703 (ipa_get_type): Likewise.
35704 (ipa_get_param_move_cost): Likewise.
35705 (ipa_set_param_used): Likewise.
35706 (ipa_get_controlled_uses): Likewise.
35707 (ipa_set_controlled_uses): Likewise.
35708 (ipa_is_param_used): Likewise.
35709 (ipa_node_params_t): Move into garbage collector. New methods insert
35711 (ipa_node_params_sum): Annotate wth GTY(()).
35712 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
35714 (ipa_load_from_parm_agg): Adjust declaration.
35715 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
35716 * ipa-profile.c (ipa_profile): Likewise.
35717 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
35718 (ipa_populate_param_decls): Make descriptors parameter garbage
35720 (ipa_dump_param): Adjust to descriptors being a pointer.
35721 (ipa_alloc_node_params): Likewise.
35722 (ipa_initialize_node_params): Likewise.
35723 (load_from_param_1): Make descriptors parameter garbage collected.
35724 (load_from_unmodified_param): Likewise.
35725 (load_from_param): Likewise.
35726 (ipa_load_from_parm_agg): Likewise.
35727 (ipa_node_params::~ipa_node_params): Removed.
35728 (ipa_free_all_node_params): Remove call to delete operator.
35729 (ipa_node_params_t::insert): New.
35730 (ipa_node_params_t::remove): Likewise.
35731 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
35732 copy known_csts and known_contexts vectors.
35733 (ipa_read_node_info): Adjust to descriptors being a pointer.
35734 (ipcp_modif_dom_walker): Make m_descriptors field garbage
35736 (ipcp_transform_function): Make descriptors variable garbage
35739 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
35741 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
35742 * config/i386/avx512dqintrin.h: Ditto.
35743 * config/i386/avx512fintrin.h: Ditto.
35744 * config/i386/i386.c: Handle new builtins.
35745 * config/i386/i386-builtin.def: Add new builtins.
35746 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
35747 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
35749 2017-01-23 Jakub Jelinek <jakub@redhat.com>
35750 Martin Liska <mliska@suse.cz>
35752 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
35753 * asan.c (asan_expand_poison_ifn): Support stores and use
35754 appropriate ASAN report function.
35755 * internal-fn.c (expand_ASAN_POISON_USE): New function.
35756 * internal-fn.def (ASAN_POISON_USE): Declare.
35757 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
35758 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
35759 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
35760 ASAN_POISON calls w/o LHS.
35761 * tree-ssa.c (execute_update_addresses_taken): Create clobber
35762 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
35763 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
35764 * gimplify.c (asan_poison_variables): Add attribute
35765 use_after_scope_memory to variables that really needs to live
35767 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
35768 having the attribute.
35770 2017-01-23 Martin Liska <mliska@suse.cz>
35772 * asan.c (create_asan_shadow_var): New function.
35773 (asan_expand_poison_ifn): Likewise.
35774 * asan.h (asan_expand_poison_ifn): New declaration.
35775 * internal-fn.c (expand_ASAN_POISON): Likewise.
35776 * internal-fn.def (ASAN_POISON): New builtin.
35777 * sanopt.c (pass_sanopt::execute): Expand
35778 asan_expand_poison_ifn.
35779 * tree-inline.c (copy_decl_for_dup_finish): Make function
35781 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
35782 * tree-ssa.c (is_asan_mark_p): New function.
35783 (execute_update_addresses_taken): Rewrite local variables
35784 (identified just by use-after-scope as addressable) into SSA.
35786 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35788 * doc/install.texi (Specific): opensource.apple.com uses https
35789 now. Remove trailing slash.
35791 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
35793 * README.Portability: Remove note on an Irix compatibility issue.
35795 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
35797 * gcov.c (INCLUDE_ALGORITHM): Define.
35798 (INCLUDE_VECTOR): Define.
35799 No longer include <vector> and <algorithm> directly.
35801 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
35803 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
35805 * doc/invoke.texi (Code Gen Options): Ditto.
35807 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
35810 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
35812 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
35814 rtl-optimization/79125
35815 * cprop.c (local_cprop_pass): Handle cases where we make an
35816 unconditional trap.
35818 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
35822 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
35823 read from, for big endian.
35825 2017-01-20 Jiong Wang <jiong.wang@arm.com>
35827 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
35828 register pauth builtins for LP64 only.
35830 2017-01-20 Marek Polacek <polacek@redhat.com>
35833 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
35836 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35838 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
35840 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
35841 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
35842 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
35844 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35847 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
35848 in big-endian mode when they are not a single duplicated value.
35850 2017-01-20 Richard Biener <rguenther@suse.de>
35852 * BASE-VER: Bump to 7.0.1.
35854 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
35856 * omp-low.c (omplow_simd_context): New struct. Use it...
35857 (lower_rec_simd_input_clauses): ...here and...
35858 (lower_rec_input_clauses): ...here to hold common data. Adjust all
35859 references to idx, lane, max_vf, is_simt.
35861 2017-01-20 Graham Markall <graham.markall@embecosm.com>
35863 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
35866 2017-01-20 Martin Jambor <mjambor@suse.cz>
35868 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
35869 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
35871 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
35872 (GTFILES): Rename hsa.c to hsa-common.c.
35873 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
35874 * hsa-dump.c: Likewise.
35875 * hsa-gen.c: Likewise.
35876 * hsa-regalloc.c: Likewise.
35877 * ipa-hsa.c: Likewise.
35878 * omp-expand.c: Likewise.
35879 * omp-low.c: Likewise.
35880 * toplev.c: Likewise.
35882 2017-01-20 Marek Polacek <polacek@redhat.com>
35885 * doc/invoke.texi: Document -Wduplicated-branches.
35886 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
35887 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
35888 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
35889 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
35890 return 0 only when not OEP_LEXICOGRAPHIC.
35891 (fold_build_cleanup_point_expr): Use the expression
35892 location when building CLEANUP_POINT_EXPR.
35893 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
35894 * tree.c (add_expr): Handle error_mark_node.
35896 2017-01-20 Martin Liska <mliska@suse.cz>
35899 * tree-profile.c (init_ic_make_global_vars): Do not call
35901 (gimple_init_gcov_profiler): Likewise.
35903 2017-01-20 Martin Liska <mliska@suse.cz>
35906 * cgraph.h (maybe_create_reference): Remove argument and
35908 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
35910 * ipa-cp.c (create_specialized_node): Likewise.
35911 * symtab.c (symtab_node::maybe_create_reference): Handle
35912 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
35914 2017-01-20 Martin Liska <mliska@suse.cz>
35916 * read-rtl-function.c (function_reader::create_function): Use
35917 build_decl instread of build_decl_stat.
35919 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
35921 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
35922 * config/i386/avx512dqintrin.h: Ditto.
35923 * config/i386/avx512fintrin.h: Ditto.
35924 * config/i386/i386-builtin-types.def: Add new types.
35925 * config/i386/i386.c: Handle new types.
35926 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
35927 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
35928 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
35929 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
35930 (__builtin_ia32_kshiftridi): New.
35931 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
35933 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
35937 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
35938 define to rs6000_init_stack_protect_guard.
35939 (rs6000_init_stack_protect_guard): New function.
35941 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
35942 Yunqiang Su <yunqiang.su@imgtec.com>
35944 * config.gcc (supported_defaults): Add madd4.
35945 (with_madd4): Add validation.
35946 (all_defaults): Add madd4.
35947 * config/mips/mips.opt (mmadd4): New option.
35948 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
35950 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
35952 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
35953 (ISA_HAS_FUSED_MADD4): Likewise.
35954 * doc/invoke.texi (-mmadd4): Document the new option.
35955 * doc/install.texi (--with-madd4): Document the new option.
35957 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35959 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
35960 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
35961 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
35962 (aarch64_init_pauth_hint_builtins): New.
35963 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
35964 (aarch64_expand_builtin): Expand new builtins.
35966 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35968 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
35969 * combine-stack-adj.c (no_unhandled_cfa): Handle
35970 REG_CFA_TOGGLE_RA_MANGLE.
35971 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
35972 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
35973 info for return address signing.
35974 (aarch64_expand_epilogue): Likewise.
35976 2017-01-19 Jiong Wang <jiong.wang@arm.com>
35978 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
35979 * config/aarch64/aarch64-protos.h
35980 (aarch64_return_address_signing_enabled): New declaration.
35981 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
35983 (aarch64_expand_prologue): Sign return address before it's pushed onto
35985 (aarch64_expand_epilogue): Authenticate return address fetched from
35987 (aarch64_override_options): Sanity check for ILP32 and ISA level.
35988 (aarch64_attributes): New function attributes for "sign-return-address".
35989 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
35990 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
35991 ("*do_return"): Generate combined instructions according to key index.
35992 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
35993 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
35995 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
35996 * config/aarch64/aarch64.opt (msign-return-address=): New.
35997 * doc/extend.texi (AArch64 Function Attributes): Documents
35998 "sign-return-address=".
35999 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
36001 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
36003 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
36004 overall option summary.
36006 2017-01-19 Jiong Wang <jiong.wang@arm.com>
36008 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
36009 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
36010 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
36011 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
36013 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
36015 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
36016 -mpower9-minmax by default for -mcpu=power9.
36017 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
36018 128-bit floating point.
36020 2017-01-20 Alan Modra <amodra@gmail.com>
36022 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
36023 optimizing for size.
36025 2017-01-20 Alan Modra <amodra@gmail.com>
36028 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
36029 for strcmp and strncmp from corresponding builtin decl.
36031 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36033 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
36034 instead of i386/rtems-64.h.
36035 * config/i386/rtems-64.h: Remove.
36037 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
36041 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
36043 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
36045 2017-01-19 Tamar Christina <tamar.christina@arm.com>
36047 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
36048 Change int to HOST_WIDE_INT.
36049 * config/aarch64/aarch64-protos.h
36050 (aarch64_simd_gen_const_vector_dup): Likewise.
36051 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
36053 2017-01-19 David Malcolm <dmalcolm@redhat.com>
36055 * langhooks-def.h (lhd_type_for_size): New decl.
36056 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
36057 * langhooks.c (lhd_type_for_size): New function, taken from
36060 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
36062 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
36063 define_bypass for CR latency.
36064 (power9-cracked-alu): Update bypass latency and remove power9-branch.
36065 (power9-alu2): Add define_bypass for CR latency.
36067 (power9-mul): Update insn latency.
36068 (power9-mul-compare): Update insn latency, bypass latency and remove
36071 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36073 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
36075 * config/aarch64/aarch64.md
36076 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
36077 aarch64_nopcrelative_literal_loads.
36078 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
36080 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
36082 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
36083 TARGET_LOONGSON_3A.
36084 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
36086 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
36089 * config.gcc (supported_defaults): Add lxc1-sxc1.
36090 (with_lxc1_sxc1): Add validation.
36091 (all_defaults): Add lxc1-sxc1.
36092 * config/mips/mips.opt (mlxc1-sxc1): New option.
36093 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
36095 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
36096 __mips_no_lxc1_sxc1.
36097 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
36098 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
36099 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
36101 2017-01-19 Richard Biener <rguenther@suse.de>
36103 PR tree-optimization/72488
36104 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
36105 sure to restore SSA info.
36106 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
36108 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
36110 PR rtl-optimization/79121
36111 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
36112 of the inner type when shifting an extended value.
36114 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36117 * symtab.c (symtab_node::equal_address_to): Fix comparing of
36118 interposable aliases.
36120 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
36123 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
36124 Use the evmergelohi instruction.
36125 (mov_si<mode>_e500_subreg4_2_le): Likewise.
36126 (mov_sitf_e500_subreg8_2_be): Likewise.
36127 (mov_sitf_e500_subreg12_2_le): Likewise.
36128 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
36129 (mov_si<mode>_e500_subreg4_2_be): Likewise.
36130 (mov_sitf_e500_subreg8_2_le): Likewise.
36131 (mov_sitf_e500_subreg12_2_be): Likewise.
36133 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36135 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
36136 attribute from vecsimple to vecperm.
36137 (altivec_vbpermq2): Likewise.
36139 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36142 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
36144 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36145 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
36146 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
36147 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
36148 case where N arg is SIZE_MAX.
36149 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
36150 (cmpstrsi): Add pattern.
36152 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
36154 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
36155 __builtin_vec_revb builtins.
36156 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
36157 built-in functions to support generation of the ISA 3.0 XXBR<x>
36158 vector byte reverse instructions.
36159 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
36160 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
36161 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
36162 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
36163 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
36164 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
36165 (P9V_BUILTIN_VEC_REVB): Likewise.
36166 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
36167 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
36168 (p9_xxbrq_v16qi): Likewise.
36169 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
36170 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
36171 (p9_xxbrh_v8hi): Likewise.
36172 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
36173 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
36174 vec_revb built-in functions.
36176 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
36178 PR rtl-optimization/78952
36179 * config/i386/i386.md (any_extract): New code iterator.
36180 (*insvqi_2): Use any_extract for source operand.
36181 (*insvqi_3): Use any_shiftrt for source operand.
36183 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
36185 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
36187 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
36189 2017-01-18 Matthias Klose <doko@ubuntu.com>
36191 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
36193 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36195 * config/rs6000/altivec.h (vec_bperm): Change #define.
36196 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
36197 (altivec_vbpermq2): New define_insn.
36198 (altivec_vbpermd): Likewise.
36199 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
36200 function interface.
36201 (VBPERMD): Likewise.
36202 (VBPERM): New polymorphic function interface.
36203 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
36204 Add entries for P9V_BUILTIN_VEC_VBPERM.
36205 * doc/extend.texi: Add interfaces for vec_bperm.
36207 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36209 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
36210 first letter of error messages.
36211 (s390_resolve_overloaded_builtin): Likewise.
36212 * config/s390/s390.c (s390_expand_builtin): Likewise.
36213 (s390_invalid_arg_for_unprototyped_fn): Likewise.
36214 (s390_valid_target_attribute_inner_p): Likewise.
36215 * config/s390/s390.md ("tabort"): Likewise.
36217 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
36219 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
36220 (ISA_AVOID_DIV_HILO): New macro.
36221 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
36222 (ISA_HAS_DDIV): Likewise.
36224 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36226 * doc/invoke.texi (fabi-version): Correct number of occurrences.
36228 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36230 * doc/invoke.texi (fabi-version): Spelling fix.
36232 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36235 * doc/invoke.texi (fabi-version): Mention mangling fix for
36238 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
36241 * doc/invoke.texi (fabi-version): Document discriminator mangling.
36243 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
36246 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
36247 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
36249 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
36251 (stack_protect_test): Ditto.
36252 * config/rs6000/rs6000.opt (mstack-protector-guard=,
36253 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
36255 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
36256 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
36257 -mstack-protector-guard-offset=.
36258 (RS/6000 and PowerPC Options): Ditto.
36260 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36262 * config/i386/i386.h (MASK_CLASS_P): New define.
36263 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
36264 there are no registers from different register sets also when
36265 mask registers are used. Update function comment.
36266 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
36267 to (*k/*r) and (*k/*km) alternatives.
36269 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
36271 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
36272 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
36273 (EH_RETURN_HANDLER_RTX): New define.
36274 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
36275 Force frame pointer in EH return functions.
36276 (aarch64_expand_epilogue): Add barrier for eh_return.
36277 (aarch64_final_eh_return_addr): Remove.
36278 (aarch64_eh_return_handler_rtx): New function.
36279 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
36281 (aarch64_eh_return_handler_rtx): New prototype.
36283 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36285 * config/rs6000/altivec.h (vec_rlmi): New #define.
36286 (vec_vrlnm): Likewise.
36287 (vec_rlnm): Likewise.
36288 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
36289 (UNSPEC_VRLNM): Likewise.
36290 (VIlong): New mode iterator.
36291 (altivec_vrl<VI_char>mi): New define_insn.
36292 (altivec_vrl<VI_char>nm): Likewise.
36293 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
36295 (VRLDNM): Likewise.
36296 (RLNM): New polymorphic function entry.
36297 (VRLWMI): New monomorphic function entry.
36298 (VRLDMI): Likewise.
36299 (RLMI): New polymorphic function entry.
36300 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
36301 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
36302 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
36305 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36308 * dwarf2out.c (field_byte_offset): Restore the
36309 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
36310 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
36311 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
36314 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36317 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
36319 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36322 * dwarf2out.c (add_data_member_location_attribute): For constant
36323 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
36324 instead of DW_AT_data_member_location, DW_AT_bit_offset and
36325 DW_AT_byte_size attributes.
36327 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
36329 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
36330 after forcing to constant memory when the code model is medium.
36332 2017-01-17 Julia Koval <julia.koval@intel.com>
36335 * config/i386/avx512fintrin.h
36336 (_mm512_i32gather_ps): Change __addr type to void const*.
36337 (_mm512_mask_i32gather_ps): Ditto.
36338 (_mm512_i32gather_pd): Ditto.
36339 (_mm512_mask_i32gather_pd): Ditto.
36340 (_mm512_i64gather_ps): Ditto.
36341 (_mm512_mask_i64gather_ps): Ditto.
36342 (_mm512_i64gather_pd): Ditto.
36343 (_mm512_mask_i64gather_pd): Ditto.
36344 (_mm512_i32gather_epi32): Ditto.
36345 (_mm512_mask_i32gather_epi32): Ditto.
36346 (_mm512_i32gather_epi64): Ditto.
36347 (_mm512_mask_i32gather_epi64): Ditto.
36348 (_mm512_i64gather_epi32): Ditto.
36349 (_mm512_mask_i64gather_epi32): Ditto.
36350 (_mm512_i64gather_epi64): Ditto.
36351 (_mm512_mask_i64gather_epi64): Ditto.
36352 (_mm512_i32scatter_ps): Change __addr type to void*.
36353 (_mm512_mask_i32scatter_ps): Ditto.
36354 (_mm512_i32scatter_pd): Ditto.
36355 (_mm512_mask_i32scatter_pd): Ditto.
36356 (_mm512_i64scatter_ps): Ditto.
36357 (_mm512_mask_i64scatter_ps): Ditto.
36358 (_mm512_i64scatter_pd): Ditto.
36359 (_mm512_mask_i64scatter_pd): Ditto.
36360 (_mm512_i32scatter_epi32): Ditto.
36361 (_mm512_mask_i32scatter_epi32): Ditto.
36362 (_mm512_i32scatter_epi64): Ditto.
36363 (_mm512_mask_i32scatter_epi64): Ditto.
36364 (_mm512_i64scatter_epi32): Ditto.
36365 (_mm512_mask_i64scatter_epi32): Ditto.
36366 (_mm512_i64scatter_epi64): Ditto.
36367 (_mm512_mask_i64scatter_epi64): Ditto.
36368 * config/i386/avx512pfintrin.h
36369 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
36370 (_mm512_mask_prefetch_i32gather_ps): Ditto.
36371 (_mm512_mask_prefetch_i64gather_pd): Ditto.
36372 (_mm512_mask_prefetch_i64gather_ps): Ditto.
36373 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
36374 (_mm512_prefetch_i32scatter_ps): Ditto.
36375 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
36376 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
36377 (_mm512_prefetch_i64scatter_pd): Ditto.
36378 (_mm512_prefetch_i64scatter_ps): Ditto.
36379 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
36380 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
36381 * config/i386/avx512vlintrin.h
36382 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
36383 (_mm_mmask_i32gather_ps): Ditto.
36384 (_mm256_mmask_i32gather_pd): Ditto.
36385 (_mm_mmask_i32gather_pd): Ditto.
36386 (_mm256_mmask_i64gather_ps): Ditto.
36387 (_mm_mmask_i64gather_ps): Ditto.
36388 (_mm256_mmask_i64gather_pd): Ditto.
36389 (_mm_mmask_i64gather_pd): Ditto.
36390 (_mm256_mmask_i32gather_epi32): Ditto.
36391 (_mm_mmask_i32gather_epi32): Ditto.
36392 (_mm256_mmask_i32gather_epi64): Ditto.
36393 (_mm_mmask_i32gather_epi64): Ditto.
36394 (_mm256_mmask_i64gather_epi32): Ditto.
36395 (_mm_mmask_i64gather_epi32): Ditto.
36396 (_mm256_mmask_i64gather_epi64): Ditto.
36397 (_mm_mmask_i64gather_epi64): Ditto.
36398 (_mm256_i32scatter_ps): Change __addr type to void*.
36399 (_mm256_mask_i32scatter_ps): Ditto.
36400 (_mm_i32scatter_ps): Ditto.
36401 (_mm_mask_i32scatter_ps): Ditto.
36402 (_mm256_i32scatter_pd): Ditto.
36403 (_mm256_mask_i32scatter_pd): Ditto.
36404 (_mm_i32scatter_pd): Ditto.
36405 (_mm_mask_i32scatter_pd): Ditto.
36406 (_mm256_i64scatter_ps): Ditto.
36407 (_mm256_mask_i64scatter_ps): Ditto.
36408 (_mm_i64scatter_ps): Ditto.
36409 (_mm_mask_i64scatter_ps): Ditto.
36410 (_mm256_i64scatter_pd): Ditto.
36411 (_mm256_mask_i64scatter_pd): Ditto.
36412 (_mm_i64scatter_pd): Ditto.
36413 (_mm_mask_i64scatter_pd): Ditto.
36414 (_mm256_i32scatter_epi32): Ditto.
36415 (_mm256_mask_i32scatter_epi32): Ditto.
36416 (_mm_i32scatter_epi32): Ditto.
36417 (_mm_mask_i32scatter_epi32): Ditto.
36418 (_mm256_i32scatter_epi64): Ditto.
36419 (_mm256_mask_i32scatter_epi64): Ditto.
36420 (_mm_i32scatter_epi64): Ditto.
36421 (_mm_mask_i32scatter_epi64): Ditto.
36422 (_mm256_i64scatter_epi32): Ditto.
36423 (_mm256_mask_i64scatter_epi32): Ditto.
36424 (_mm_i64scatter_epi32): Ditto.
36425 (_mm_mask_i64scatter_epi32): Ditto.
36426 (_mm256_i64scatter_epi64): Ditto.
36427 (_mm256_mask_i64scatter_epi64): Ditto.
36428 (_mm_i64scatter_epi64): Ditto.
36429 (_mm_mask_i64scatter_epi64): Ditto.
36430 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
36431 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
36432 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
36433 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
36434 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
36435 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
36436 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
36437 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
36438 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
36439 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
36440 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
36441 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
36442 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
36443 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
36444 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
36445 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
36446 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
36447 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
36448 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
36449 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
36450 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
36451 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
36452 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
36453 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
36454 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
36455 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
36456 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
36457 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
36458 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
36459 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
36460 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
36461 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
36462 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
36463 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
36464 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
36465 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
36466 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
36467 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
36468 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
36469 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
36470 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
36471 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
36472 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
36473 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
36474 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
36475 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
36476 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
36477 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
36478 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
36479 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
36480 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
36481 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
36482 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
36483 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
36484 definitions accordingly.
36486 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
36487 Kuan-Lin Chen <kuanlinchentw@gmail.com>
36490 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
36493 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
36496 * ira-conflicts.c (ira_build_conflicts): Update total conflict
36497 hard regs for inner regno.
36499 2017-01-17 Martin Liska <mliska@suse.cz>
36502 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
36503 assumption and add comment.
36505 2017-01-17 Nathan Sidwell <nathan@acm.org>
36507 * ipa-visibility.c (localize_node): New function, broken out of ...
36508 (function_and_variable_visibility): ... here. Call it.
36510 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
36512 PR middle-end/77445
36513 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
36514 correctly set frequency of oudgoing edge.
36515 (duplicate_thread_path): Fix profile updating.
36517 2017-01-17 Jakub Jelinek <jakub@redhat.com>
36520 * configure.ac: Add GCC_BASE_VER.
36521 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
36522 version from BASE-VER file.
36523 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
36524 (gcc.o): Depend on $(BASEVER).
36525 * common.opt (dumpfullversion): New option.
36526 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
36527 * doc/invoke.texi: Document -dumpfullversion.
36528 * doc/install.texi: Document --with-gcc-major-version-only.
36529 * configure: Regenerated.
36531 2017-01-17 Richard Biener <rguenther@suse.de>
36533 PR tree-optimization/71433
36534 * tree-vrp.c (register_new_assert_for): Merge same asserts
36535 on all incoming edges.
36536 (process_assert_insertions_for): Handle insertions at the
36539 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
36541 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
36542 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
36544 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
36547 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
36550 2017-01-17 Alan Modra <amodra@gmail.com>
36553 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
36554 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
36555 symbolic stack limit when pic.
36557 2017-01-16 Martin Sebor <msebor@redhat.com>
36559 PR tree-optimization/78608
36560 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
36562 2017-01-16 Jeff Law <law@redhat.com>
36565 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
36566 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
36567 for several include directories that may be relative to sysroot.
36568 * config/i386/x-mingw32 (gplus_includedir): Define.
36569 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
36570 (native_system_includedir): Likewise.
36571 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
36572 override if TARGET_SYSTEM_ROOT is defined.
36573 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
36575 PR tree-optimization/79090
36576 PR tree-optimization/33562
36577 PR tree-optimization/61912
36578 PR tree-optimization/77485
36579 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
36580 and computed trims into the dump file.
36582 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
36584 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
36586 2017-01-16 Jakub Jelinek <jakub@redhat.com>
36589 * gimplify.c (gimplify_init_constructor): If want_value and
36590 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
36594 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
36595 sequence. Formatting fixes.
36596 (doloop_optimize): Formatting fixes.
36599 * gcc.c (debug_level_greater_than_spec_func): New function.
36600 (static_spec_functions): Add debug-level-gt spec function.
36601 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
36603 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
36604 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
36605 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
36606 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
36607 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
36608 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
36610 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
36612 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
36613 QImode fixups to general and mask registers only.
36615 2017-01-16 Carl Love <cel@us.ibm.com>
36617 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
36618 for built-in functions
36619 vector signed char vec_nabs (vector signed char)
36620 vector signed short vec_nabs (vector signed short)
36621 vector signed int vec_nabs (vector signed int)
36622 vector signed long long vec_nabs (vector signed long long)
36623 vector float vec_nabs (vector float)
36624 vector double vec_nabs (vector double)
36625 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
36627 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
36628 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
36629 * doc/extend.texi: Update the documentation file for the new built-in
36632 2017-01-16 Martin Sebor <msebor@redhat.com>
36634 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
36637 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36639 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
36640 UNSPEC_VSX__XXSPLTD to require special splat handling.
36642 2017-01-16 David Malcolm <dmalcolm@redhat.com>
36645 * system.h: Poison strndup.
36647 2017-01-16 Alan Modra <amodra@gmail.com>
36650 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
36653 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
36655 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
36657 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
36659 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
36660 call recog here. Assert that INSN_CODE (insn) is non-negative.
36662 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36665 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
36667 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
36668 in the currently scheduled RTL fragment.
36670 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
36672 PR rtl-optimization/78751
36673 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
36676 2017-01-14 Jeff Law <law@redhat.com>
36678 PR tree-optimization/79090
36679 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
36680 variable length stores.
36681 (compute_trims): Delete dead assignment to *trim_tail.
36682 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
36685 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
36687 PR rtl-optimization/78626
36688 PR rtl-optimization/78727
36689 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
36690 of a block, and split such blocks after everything else is finished.
36692 2017-01-14 Alan Modra <amodra@gmail.com>
36695 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
36696 target legitimate_combined_insn.
36697 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
36698 (rs6000_legitimate_combined_insn): New function.
36699 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
36701 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
36702 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
36703 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
36705 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
36707 * doc/frontends.texi (G++ and GCC): Remove references to Java.
36709 2017-01-13 Jeff Law <law@redhat.com>
36711 PR tree-optimization/33562
36712 PR tree-optimization/61912
36713 PR tree-optimization/77485
36714 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
36716 (delete_dead_assignment): Likewise.
36717 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
36718 statement to delete_dead_call and delete_dead_assignment.
36720 2017-01-13 David Malcolm <dmalcolm@redhat.com>
36723 * substring-locations.c (format_warning_va): Strengthen case 1 so
36724 that both endpoints of the substring must be within the format
36725 range for just the substring to be printed.
36727 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
36729 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
36730 * config/i386/i386.c (ix86_target_string): Add missing options
36731 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
36732 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
36733 flags_other and ix86_target_other to flags2_other. Display unknown
36735 (ix86_valid_target_attribute_inner_p): Add missing options and
36736 reorder options by implied ISAs, as in ix86_target_string.
36738 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36740 * hash-table.h (hash_table::too_empty_p): New function.
36741 (hash_table::expand): Use it.
36742 (hash_table::traverse): Likewise.
36743 (hash_table::empty_slot): Use sizeof (value_type) instead of
36744 sizeof (PTR) to convert bytes to elements. Shrink the table
36745 if the current size is excessive for the current number of
36748 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
36750 * ira-costs.c (record_reg_classes): Break from the inner loop
36751 early once alt_fail is known to be true. Update outer loop
36752 handling accordingly.
36754 2017-01-13 Jeff Law <law@redhat.com>
36756 * tree-ssa-dse.c (decrement_count): New function.
36757 (increment_start_addr, maybe_trim_memstar_call): Likewise.
36758 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
36759 when we know the partially dead statement is a mem* function.
36761 PR tree-optimization/61912
36762 PR tree-optimization/77485
36763 * tree-ssa-dse.c: Include expr.h.
36764 (maybe_trim_constructor_store): New function.
36765 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
36767 PR tree-optimization/33562
36768 PR tree-optimization/61912
36769 PR tree-optimization/77485
36770 * doc/invoke.texi: Document new dse-max-object-size param.
36771 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
36772 * tree-ssa-dse.c: Include params.h.
36773 (dse_store_status): New enum.
36774 (initialize_ao_ref_for_dse): New, partially extracted from
36776 (valid_ao_ref_for_dse, normalize_ref): New.
36777 (setup_live_bytes_from_ref, compute_trims): Likewise.
36778 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
36779 (maybe_trim_partially_dead_store): Likewise.
36780 (maybe_trim_complex_store): Likewise.
36781 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
36782 Track what bytes live from the original store. Return tri-state
36783 for dead, partially dead or live.
36784 (dse_dom_walker): Add constructor, destructor and new private members.
36785 (delete_dead_call, delete_dead_assignment): New extracted from
36787 (dse_optimize_stmt): Make a member of dse_dom_walker.
36788 Use initialize_ao_ref_for_dse.
36790 PR tree-optimization/33562
36791 PR tree-optimization/61912
36792 PR tree-optimization/77485
36793 * sbitmap.h (bitmap_count_bits): Prototype.
36794 (bitmap_clear_range, bitmap_set_range): Likewise.
36795 * sbitmap.c (bitmap_clear_range): New function.
36796 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
36798 2017-01-13 Martin Liska <mliska@suse.cz>
36801 * function.c (set_cfun): Add new argument force.
36802 * function.h (set_cfun): Likewise.
36803 * ipa-inline-transform.c (inline_call): Use the function when
36804 strict alising from is dropped for function we inline to.
36806 2017-01-13 Richard Biener <rguenther@suse.de>
36808 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
36809 for dumping GIMPLE INTEGER_CSTs.
36811 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
36813 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
36814 to 201112L since C++17.
36816 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
36819 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
36820 if -fsanitize=kernel-address is present.
36822 2017-01-13 Richard Biener <rguenther@suse.de>
36824 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
36825 as _Literal ( type ) number in case usual suffixes do not
36826 preserve all information.
36828 2017-01-13 Richard Biener <rguenther@suse.de>
36830 PR tree-optimization/77283
36831 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
36832 and ssa-iterators.h.
36833 (is_feasible_trace): Implement a cost model based on joiner
36836 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
36839 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
36840 char or short to __float128/_Float128 directly.
36842 2017-01-12 Martin Sebor <msebor@redhat.com>
36844 to -Wformat-overflow.
36845 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
36846 (min_bytes_remaining): Same.
36847 (get_string_length): Same.
36848 (format_string): Same.
36849 (format_directive): Same.
36851 (pass_sprintf_length::handle_gimple_call): Same.
36853 2017-01-12 Jakub Jelinek <jakub@redhat.com>
36855 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
36856 info.nowrite calls with no lhs that can't throw. Return bool
36857 whether gsi_remove has been called or not.
36858 (pass_sprintf_length::handle_gimple_call): Return bool whether
36859 try_substitute_return_value called gsi_remove. Formatting fix.
36860 (pass_sprintf_length::execute): Don't use gsi_remove if
36861 handle_gimple_call returned true.
36864 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
36865 be removed due to side-effects, don't remove following barrier nor
36866 turn the successor edge into fallthru edge.
36868 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36871 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
36872 element-reversing loads and stores as not swappable.
36874 2017-01-12 Nathan Sidwell <nathan@acm.org>
36875 Nicolai Stange <nicstange@gmail.com>
36877 * combine.c (try_combine): Don't ignore result of overlap checking
36878 loop. Combine overlap & asm check into single loop.
36880 2017-01-12 Richard Biener <rguenther@suse.de>
36882 * tree-pretty-print.c (dump_generic_node): Provide -gimple
36883 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
36885 2017-01-12 Richard Biener <rguenther@suse.de>
36887 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
36888 and TS_TARGET_OPTION directly derive from TS_BASE.
36889 * tree-core.h (tree_optimization_option): Derive from tree_base.
36890 (tree_target_option): Likewise.
36892 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
36894 * config/i386/i386.c (memory_address_length): Increase len
36895 only when rip_relative_addr_p returns false.
36897 2017-01-11 Julia Koval <julia.koval@intel.com>
36899 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
36900 (OPTION_MASK_ISA_SGX_SET): New.
36901 (ix86_handle_option): Handle OPT_msgx.
36902 * config.gcc: Added sgxintrin.h.
36903 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
36904 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
36905 * config/i386/i386.c (ix86_target_string): Add -msgx.
36907 (ix86_option_override_internal): Handle new options.
36908 (ix86_valid_target_attribute_inner_p): Add sgx.
36909 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
36910 * config/i386/i386.opt: Add msgx.
36911 * config/i386/sgxintrin.h: New file.
36912 * config/i386/x86intrin.h: Add sgxintrin.h.
36914 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36917 * fold-const.c (maybe_nonzero_address): Return 1 for function
36919 (tree_single_nonzero_warnv_p): Don't handle function local objects
36923 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
36926 2017-01-11 David Malcolm <dmalcolm@redhat.com>
36929 * opts.c: Include "spellcheck.h"
36930 (struct string_fragment): New struct.
36931 (struct edit_distance_traits<const string_fragment &>): New
36933 (get_closest_sanitizer_option): New function.
36934 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
36936 2017-01-11 Jakub Jelinek <jakub@redhat.com>
36938 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
36940 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
36941 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
36942 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
36943 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
36944 for initial die_offset if dwarf_split_debug_info.
36945 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
36946 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
36948 (output_skeleton_debug_sections): Formatting fix. Use
36949 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
36950 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
36952 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
36954 * config/arm/cortex-a53.md: Add bypasses for
36955 cortex_a53_r2f_cvt.
36956 (cortex_a53_r2f): Only use for transfers.
36957 (cortex_a53_f2r): Likewise.
36958 (cortex_a53_r2f_cvt): Add reservation for conversions.
36959 (cortex_a53_f2r_cvt): Likewise.
36961 2017-01-11 Tamar Christina <tamar.christina@arm.com>
36963 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
36964 to all inlined functions, change static to extern.
36966 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
36969 * config/arm/arm.c (legitimize_pic_address): Handle reference to
36971 (arm_assemble_integer): Likewise.
36973 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
36975 * config.gcc: Use new awk script to check CPU, FPU and architecture
36976 parameters for --with-... options.
36977 * config/arm/parsecpu.awk: New file
36978 * config/arm/arm-cpus.in: New file.
36979 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
36981 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
36983 * config/arm/t-arm: Update dependency rules.
36984 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
36985 of processing .def files.
36986 * config/arm/genopt.sh: Deleted.
36987 * config/arm/gentune.sh: Deleted.
36988 * config/arm/arm-cores.def: Deleted.
36989 * config/arm/arm-arches.def: Deleted.
36990 * config/arm/arm-fpus.def: Deleted.
36991 * config/arm/arm-tune.md: Regenerated.
36992 * config/arm/arm-tables.opt: Regenerated.
36993 * config/arm/arm-cpu.h: New generated file.
36994 * config/arm/arm-cpu-data.h: New generated file.
36995 * config/arm/arm-cpu-cdata.h: New generated file.
36997 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
37000 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
37002 (input_varpool_node): Unpack dynamically_initialized bit.
37004 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
37006 PR rtl-optimization/79032
37007 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
37008 the alignment of the adjusted memory reference against that of MODE,
37009 instead of the alignment of the original memory reference.
37011 2017-01-11 Martin Jambor <mjambor@suse.cz>
37013 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
37015 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
37016 decorated functions.
37018 2017-01-11 Richard Biener <rguenther@suse.de>
37020 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
37021 set range/nonnull info for PHI results. Do not set it on
37022 stmts marked for removal.
37024 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
37026 * expr.c (store_field): In the bitfield case, fetch the return value
37027 from the registers before applying a single big-endian adjustment.
37028 Always do a final load for a BLKmode value not larger than a word.
37030 2017-01-10 David Malcolm <dmalcolm@redhat.com>
37033 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37034 that we correctly handle column numbers greater than
37035 LINE_MAP_MAX_COLUMN_NUMBER.
37037 2017-01-10 Martin Sebor <msebor@redhat.com>
37039 PR middle-end/78245
37040 * gimple-ssa-sprintf.c (get_destination_size): Call
37041 {init,fini}object_sizes.
37042 * tree-object-size.c (addr_object_size): Adjust.
37043 (pass_through_call): Adjust.
37044 (pass_object_sizes::execute): Adjust.
37045 * tree-object-size.h (fini_object_sizes): Declare.
37047 2017-01-10 Martin Sebor <msebor@redhat.com>
37049 PR tree-optimization/78775
37050 * builtins.c (get_size_range): Move...
37051 * calls.c: ...to here.
37052 (alloc_max_size): Accept zero argument.
37053 (operand_signed_p): Remove.
37054 (maybe_warn_alloc_args_overflow): Call get_size_range.
37055 * calls.h (get_size_range): Declare.
37057 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
37059 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
37060 from TI's devices.csv file as of September 2016.
37061 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
37063 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
37065 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
37066 * doc/invoke.texi: Likewise.
37067 * doc/md.texi: Likewise.
37068 * doc/objc.texi: Likewise.
37070 2017-01-10 Joshua Conner <joshconner@google.com>
37072 * config/arm/fuchsia-elf.h: New file.
37073 * config/fuchsia.h: New file.
37074 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
37075 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
37077 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
37079 2016-01-10 Richard Biener <rguenther@suse.de>
37081 PR tree-optimization/79034
37082 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
37083 Propagate out degenerate PHIs in the joiner.
37085 2017-01-10 Martin Liska <mliska@suse.cz>
37087 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
37088 (sort_congruence_classes_by_decl_uid): Likewise.
37089 (sort_congruence_class_groups_by_decl_uid): Likewise.
37090 (sem_item_optimizer::merge_classes): Sort class, groups in these
37091 classes and members in the groups by DECL_UID of declarations.
37092 This would make merge operations stable.
37094 2017-01-10 Martin Liska <mliska@suse.cz>
37096 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
37097 usage of m_classes_vec.
37098 (sem_item_optimizer::~sem_item_optimizer): Likewise.
37099 (sem_item_optimizer::get_group_by_hash): Likewise.
37100 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
37101 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
37102 (sem_item_optimizer::verify_classes): Likewise.
37103 (sem_item_optimizer::process_cong_reduction): Likewise.
37104 (sem_item_optimizer::dump_cong_classes): Likewise.
37105 (sem_item_optimizer::merge_classes): Likewise.
37106 * ipa-icf.h (congruence_class_hash): Rename from
37107 congruence_class_group_hash. Remove declaration of m_classes_vec.
37109 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
37111 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
37112 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
37113 * config.gcc: Add avx512vpopcntdqintrin.h.
37114 * config/i386/avx512vpopcntdqintrin.h: New.
37115 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
37116 * config/i386/i386-builtin-types.def: Add new types.
37117 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
37118 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
37119 __builtin_ia32_vpopcountq_v8di_mask): New.
37120 * config/i386/i386-c.c (ix86_target_macros_internal): Define
37121 __AVX512VPOPCNTDQ__.
37122 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
37123 (PTA_AVX512VPOPCNTDQ): Define.
37124 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
37125 TARGET_AVX512VPOPCNTDQ_P): Define.
37126 * config/i386/i386.opt: Add mavx512vpopcntdq.
37127 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
37128 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
37130 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37132 PR middle-end/77484
37133 * predict.def (PRED_CALL): Set to 67.
37135 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
37137 * expr.c (store_field): In the bitfield case, if the value comes from
37138 a function call and is of an aggregate type returned in registers, do
37139 not modify the field mode; extract the value in all cases if the mode
37140 is BLKmode and the size is not larger than a word.
37142 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
37145 * config/i386/cpuid.h: Fix undefined behavior.
37147 2017-01-04 Jeff Law <law@redhat.com>
37149 PR tree-optimization/79007
37150 PR tree-optimization/67955
37151 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
37152 conservative for pt.null when flag_non_call_exceptions is on.
37154 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37156 PR translation/79019
37157 PR translation/79020
37158 * params.def (PARAM_INLINE_MIN_SPEEDUP,
37159 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
37160 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
37162 * config/avr/avr.opt (maccumulate-args): Likewise.
37163 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
37164 * common.opt (freport-bug): Likewise.
37165 * cif-code.def (CIF_FINAL_ERROR): Likewise.
37166 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
37167 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
37168 translatable string.
37169 * config/i386/i386.c (function_value_32): Likewise.
37170 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
37171 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
37173 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
37174 * common/config/msp430/msp430-common.c (msp430_handle_option):
37176 * symtab.c (symtab_node::verify_base): Likewise.
37177 * opts.c (set_debug_level): Likewise.
37178 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
37179 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
37180 missing whitespace to translatable strings.
37181 * config/avr/avr.md (bswapsi2): Fix typo in comment.
37182 * config/sh/superh.h: Likewise.
37183 * config/i386/xopintrin.h: Likewise.
37184 * config/i386/znver1.md: Likewise.
37185 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
37186 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
37187 * double-int.h (struct double_int): Likewise.
37188 * double-int.c (div_and_round_double): Likewise.
37189 * wide-int.cc: Likewise.
37190 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
37191 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
37192 * cfgcleanup.c (crossjumps_occured): Renamed to ...
37193 (crossjumps_occurred): ... this.
37194 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
37197 PR tree-optimization/78899
37198 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
37199 returning bool return struct loop *, NULL for failure and the new
37201 (versionable_outer_loop_p): Don't version outer loop if it has
37202 dont_vectorized bit set.
37203 (tree_if_conversion): When versioning outer loop, ensure
37204 tree_if_conversion is performed also on the inner loop of the
37205 non-vectorizable outer loop copy.
37206 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
37207 LOOP_VECTORIZED in inner loop of the scalar outer loop and
37208 prevent vectorization of it.
37209 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
37210 the outer loop vectorization of the non-scalar version is attempted
37211 before vectorization of the inner loop in scalar version. If
37212 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
37213 vectorization of its inner loop.
37214 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
37215 has 2 inner loops, rename also on edges from bb whose single pred
37216 is outer_loop->header. Fix typo in function comment.
37218 2017-01-09 Martin Sebor <msebor@redhat.com>
37221 * asan.c (asan_emit_stack_protection): Increase local buffer size
37222 to avoid snprintf truncation warning.
37224 2017-01-09 Andrew Pinski <apinski@cavium.com>
37226 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
37227 to reference thunderx2t99 for the tuning structure
37228 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
37230 (thunderx2t99_extra_costs): This.
37231 * config/aarch64/aarch64-tune.md: Regenerate.
37232 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
37233 (vulcan_addrcost_table): This.
37234 (vulcan_regmove_cost): Rename to ...
37235 (thunderx2t99_regmove_cost): This.
37236 (vulcan_vector_cost): Rename to ...
37237 (thunderx2t99_vector_cost): this.
37238 (vulcan_branch_cost): Rename to ...
37239 (thunderx2t99_branch_cost): This.
37240 (vulcan_tunings): Rename to ...
37241 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
37242 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
37244 2017-01-09 Martin Jambor <mjambor@suse.cz>
37248 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
37249 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
37250 (propagate_vr_accross_jump_function): Use the above function for all
37251 value range computations for pass-through jump functions and type
37252 converasion from explicit value range values.
37253 (ipcp_propagate_stage): Do not attempt to deduce types of formal
37254 parameters from TYPE_ARG_TYPES.
37255 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
37256 (ipa_write_node_info): Stream type of the actual argument.
37257 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
37259 2017-01-09 Martin Liska <mliska@suse.cz>
37262 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
37263 (lookup_compiler): Do not show error message with have_E.
37265 2017-01-09 Jakub Jelinek <jakub@redhat.com>
37267 PR tree-optimization/78938
37268 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
37269 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
37270 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
37271 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
37274 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37276 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
37279 2017-01-09 Richard Biener <rguenther@suse.de>
37281 PR tree-optimization/78997
37282 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
37283 name condition properly.
37285 2017-01-09 Richard Biener <rguenther@suse.de>
37288 * dwarf2out.c (is_cxx): New overload with context.
37289 (is_naming_typedef_decl): Use it.
37291 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37293 * invoke.texi (Option Summary): Correct spacing in option lists
37294 and add line breaks to fix over-long lines.
37296 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37298 PR middle-end/17660
37300 * extend.texi (Common Variable Attributes): Add xref to GCC
37301 Internals manual to explain mode attribute keywords.
37303 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
37306 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
37307 and Preprocessor Options.
37308 (Options for Linking): Document -pthread here....
37309 (RS/6000 and PowerPC Options): ...not here.
37310 (Solaris 2 Options): ...or here.
37311 * doc/cppopts.texi: Document -pthread.
37313 2017-01-08 Martin Sebor <msebor@redhat.com>
37315 PR middle-end/77708
37316 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
37317 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
37318 New member functions.
37319 (format_directive): Used them.
37321 (pass_sprintf_length::handle_gimple_call): Same.
37322 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
37323 to avoid truncation for any argument.
37324 (extract_affine_mul): Same.
37325 * tree.c (get_file_function_name): Same.
37327 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37329 PR middle-end/77484
37330 * predict.def (PRED_INDIR_CALL): Set to 86.
37332 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37334 PR preprocessor/54124
37335 * doc/cppopts.texi: Reformat -d subtable to list the full name
37336 of the options. Add cross-reference to the docs for the general
37337 compiler -d options.
37338 * doc/invoke.texi (Developer Options): Add cross-reference to the
37339 preprocessor-specific -d option documentation.
37341 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37343 PR preprocessor/13498
37344 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
37345 redudant material, and reflect new command-line options.
37346 (System Headers): Likewise.
37348 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
37350 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
37351 -isystem, and -idirafter. Copy-edit.
37352 * doc/cppopts.texi: Copy-edit. Remove contradiction about
37353 default for -ftrack-macro-expansion. Delete obsolete and
37354 badly-formatted implementation details about -fdebug-cpp output.
37355 * doc/cppwarnopts.texi: Copy-edit.
37357 2017-01-07 David Malcolm <dmalcolm@redhat.com>
37360 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
37361 that the transition from a max line width >= 1<<10 to narrower
37362 lines works correctly.
37364 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
37366 * doc/options.texi (PerFunction): New.
37367 * opt-functions.awk (switch_flags): Map both Optimization and
37368 PerFunction to CL_OPTIMIZATION.
37369 * opth-gen.awk: Test for PerFunction flag along with
37371 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
37372 it only when the latter is present. Skip those that don't in
37373 the hash function generator.
37374 * common.opt (fvar-tracking): Mark as PerFunction instead of
37376 (fvar-tracking-assignments): Likewise.
37377 (fvar-tracking-assignments-toggle): Likewise.
37378 (fvar-tracking-uninit): Likewise.
37380 2017-01-07 Jakub Jelinek <jakub@redhat.com>
37382 PR translation/79018
37383 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
37386 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
37389 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
37390 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
37391 TARGET_LONG_JUMP_TABLE_OFFSETS.
37392 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
37393 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
37394 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
37395 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
37396 * config/m68k/m68k.md (tablejump expander): Likewise.
37397 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
37398 TARGET_LONG_JUMP_TABLE_OFFSETS.
37399 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
37400 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
37402 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
37403 David Holsgrove <david.holsgrove@xilinx.com>
37405 * common/config/microblaze/microblaze-common.c
37406 (TARGET_EXCEPT_UNWIND_INFO): Remove.
37407 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
37409 * config/microblaze/microblaze.c (microblaze_must_save_register)
37410 (microblaze_expand_epilogue, microblaze_return_addr): Handle
37412 (microblaze_eh_return): New function.
37413 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
37414 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
37415 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
37416 * config/microblaze/microblaze.md (eh_return): New pattern.
37418 2017-01-06 Jakub Jelinek <jakub@redhat.com>
37420 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
37421 GCC_DIAGNOSTIC_STRINGIFY): Define.
37423 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
37425 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37427 * config/arm/arm.md (<mcrr>): New.
37429 * config/arm/arm.c (arm_arch5te): New.
37430 (arm_option_override): Set arm_arch5te.
37431 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
37433 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
37434 (arm_mcrr_qualifiers): ... this. New.
37435 (MRRC_QUALIFIERS): Define to...
37436 (arm_mrrc_qualifiers): ... this. New.
37437 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
37439 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
37440 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
37441 (MRRCI, mrrc, MRRC): New.
37442 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
37443 VUNSPEC_MRRC2): New.
37445 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37447 * config/arm/arm.md (<mcr>): New.
37449 * config/arm/arm.c (arm_coproc_builtin_available): Add
37450 support for mcr, mrc, mcr2 and mrc2.
37451 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
37452 (arm_mcr_qualifiers): ... this. New.
37453 (MRC_QUALIFIERS): Define to ...
37454 (arm_mrc_qualifiers): ... this. New.
37455 (MCR_QUALIFIERS): Define to ...
37456 (arm_mcr_qualifiers): ... this. New.
37457 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
37459 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
37460 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
37461 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
37462 VUNSPEC_MRC2): New.
37464 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37466 * config/arm/arm.md (*ldc): New.
37470 * config/arm/arm.c (arm_coproc_builtin_available): Add
37471 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
37472 (arm_coproc_ldc_stc_legitimate_address): New.
37473 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
37474 'qualifier_const_pointer'.
37475 (LDC_QUALIFIERS): Define to...
37476 (arm_ldc_qualifiers): ... this. New.
37477 (STC_QUALIFIERS): Define to...
37478 (arm_stc_qualifiers): ... this. New.
37479 * config/arm/arm-protos.h
37480 (arm_coproc_ldc_stc_legitimate_address): New.
37481 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
37482 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
37483 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
37484 stc2, stcl, stc2l): New.
37485 * config/arm/constraints.md (Uz): New.
37486 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
37487 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
37488 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
37489 VUNSPEC_STC2L): New.
37491 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37493 * config/arm/arm.md (<cdp>): New.
37494 * config/arm/arm.c (neon_const_bounds): Rename this ...
37495 (arm_const_bounds): ... this.
37496 (arm_coproc_builtin_available): New.
37497 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
37498 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
37499 (CDP_QUALIFIERS): Define to...
37500 (arm_cdp_qualifiers): ... this. New.
37502 (arm_expand_builtin_args): Add case for 6 arguments.
37503 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
37504 (arm_const_bounds): ... this.
37505 (arm_coproc_builtin_available): New.
37506 * config/arm/arm_acle.h (__arm_cdp): New.
37508 * config/arm/arm_acle_builtins.def (cdp): New.
37510 * config/arm/iterators.md (CDPI,CDP,cdp): New.
37511 * config/arm/neon.md: Rename all 'neon_const_bounds' to
37512 'arm_const_bounds'.
37513 * config/arm/types.md (coproc): New.
37514 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
37515 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
37516 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
37517 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
37519 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37521 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
37522 (UBINOP_QUALIFIERS): New.
37524 (acle_builtin_data): New. Change comment.
37525 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
37526 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
37527 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
37528 arm_acle_builtins.def.
37529 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
37530 (arm_init_acle_builtins): New.
37531 (CRC32_BUILTIN): Remove.
37532 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
37533 crc32cb, crc32ch and crc32cw.
37534 (arm_init_crc32_builtins): Remove.
37535 (arm_init_builtins): Use arm_init_acle_builtins rather
37536 than arm_init_crc32_builtins.
37537 (arm_expand_acle_builtin): New.
37538 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
37539 * config/arm/arm_acle_builtins.def: New.
37541 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
37543 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
37544 (arm_builtin_datum): ... this.
37545 (arm_init_neon_builtin): Rename to ...
37546 (arm_init_builtin): ... this. Add a new parameters PREFIX
37547 and USE_SIG_IN_NAME.
37548 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
37549 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
37550 'arm_builtin_datum'.
37551 (arm_init_vfp_builtins): Likewise.
37552 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
37553 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
37554 (arm_expand_neon_args): Rename to ...
37555 (arm_expand_builtin_args): ... this. Rename builtin_arg
37556 enum values and differentiate between ARG_BUILTIN_MEMORY
37557 and ARG_BUILTIN_NEON_MEMORY.
37558 (arm_expand_neon_builtin_1): Rename to ...
37559 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
37560 values, arm_expand_builtin_args and add bool parameter NEON.
37561 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
37562 (arm_expand_vfp_builtin): Likewise.
37563 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
37565 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
37567 PR middle-end/77484
37568 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
37569 * predict.c (tree_estimate_probability_bb): Reverse direction of
37570 polymorphic call predictor.
37572 2017-01-06 David Malcolm <dmalcolm@redhat.com>
37574 * passes.c (execute_one_pass): Split out pass-skipping logic into...
37575 (determine_pass_name_match): ...this new function and...
37576 (should_skip_pass_p): ...this new function.
37578 2017-01-06 Nathan Sidwell <nathan@acm.org>
37580 * ipa-visibility.c (function_and_variable_visibility): Reformat
37581 comments and long lines. Remove extrneous if.
37582 * symtab.c (symtab_node::make_decl_local): Fix code format.
37583 (symtab_node::set_section_for_node): Fix comment typo.
37585 2017-01-06 Martin Liska <mliska@suse.cz>
37588 * lra-constraints.c: Rename invariant to lra_invariant.
37589 * predict.c (set_even_probabilities): Initialize e to NULL.
37591 2017-01-05 Martin Sebor <msebor@redhat.com>
37593 PR tree-optimization/78910
37594 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
37595 (format_integer): Correct off-by-one error in the handling
37596 of precision with negative numbers in signed conversions..
37598 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
37600 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
37602 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37604 PR tree-optimization/71016
37605 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
37606 factor_out_conditional_conversion. Formatting fix.
37607 (factor_out_conditional_conversion): Add cond_stmt argument.
37608 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
37609 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
37612 2017-01-05 David Malcolm <dmalcolm@redhat.com>
37614 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
37615 read-rtl-function.o, and selftest-rtl.o.
37616 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
37617 (selftest::aarch64_test_loading_full_dump): New function.
37618 (selftest::aarch64_run_selftests): New function.
37619 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
37620 selftest::aarch64_run_selftests.
37621 * config/i386/i386.c
37622 (selftest::ix86_test_loading_dump_fragment_1): New function.
37623 (selftest::ix86_test_loading_call_insn): New function.
37624 (selftest::ix86_test_loading_full_dump): New function.
37625 (selftest::ix86_test_loading_unspec): New function.
37626 (selftest::ix86_run_selftests): Call the new functions.
37627 * emit-rtl.c (maybe_set_max_label_num): New function.
37628 * emit-rtl.h (maybe_set_max_label_num): New decl.
37629 * function.c (instantiate_decls): Guard call to
37630 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
37631 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
37633 * gensupport.c (gen_reader::gen_reader): Pass "false"
37634 for new "compact" param of rtx_reader.
37635 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
37636 rather than an empty string for NULL strings.
37637 * read-md.c: Potentially include config.h rather than bconfig.h.
37638 Wrap include of errors.h with #ifdef GENERATOR_FILE.
37639 (have_error): New global, copied from errors.c.
37640 (md_reader::read_name): Rename to...
37641 (md_reader::read_name_1): ...this, adding "out_loc" param,
37642 and converting "missing name or number" to returning false, rather
37644 (md_reader::read_name): Reimplement in terms of read_name_1.
37645 (md_reader::read_name_or_nil): New function.
37646 (md_reader::read_string): Handle "(nil)" by returning NULL.
37647 (md_reader::md_reader): Add new param "compact".
37648 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
37649 (md_reader::read_file): New method.
37650 * read-md.h (md_reader::md_reader): Add new param "compact".
37651 (md_reader::read_file): New method.
37652 (md_reader::is_compact): New accessor.
37653 (md_reader::read_name): Convert return type from void to file_location.
37654 (md_reader::read_name_or_nil): New decl.
37655 (md_reader::read_name_1): New decl.
37656 (md_reader::m_compact): New field.
37657 (noop_reader::noop_reader): Pass "false" for new "compact" param
37659 (rtx_reader::rtx_reader): Add new "compact" param.
37660 (rtx_reader::read_rtx_operand): Make virtual and convert return
37661 type from void to rtx.
37662 (rtx_reader::read_until): New decl.
37663 (rtx_reader::handle_any_trailing_information): New virtual function.
37664 (rtx_reader::postprocess): New virtual function.
37665 (rtx_reader::finalize_string): New virtual function.
37666 (rtx_reader::m_in_call_function_usage): New field.
37667 (rtx_reader::m_reuse_rtx_by_id): New field.
37668 * read-rtl-function.c: New file.
37669 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
37670 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
37671 (selftest::verify_three_block_rtl_cfg): New decl.
37672 * read-rtl-function.h: New file.
37673 * read-rtl.c: Potentially include config.h rather than bconfig.h.
37674 For host, include function.h, memmodel.h, and emit-rtl.h.
37675 (one_time_initialization): New function.
37676 (struct compact_insn_name): New struct.
37677 (compact_insn_names): New array.
37678 (find_code): Handle insn codes in compact dumps.
37679 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
37680 (bind_subst_iter_and_attr): Likewise.
37681 (add_condition_to_string): Likewise.
37682 (add_condition_to_rtx): Likewise.
37683 (apply_attribute_uses): Likewise.
37684 (add_current_iterators): Likewise.
37685 (apply_iterators): Likewise.
37686 (initialize_iterators): Guard usage of apply_subst_iterator with
37687 #ifdef GENERATOR_FILE.
37688 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
37689 (md_reader::read_mapping): Likewise.
37690 (add_define_attr_for_define_subst): Likewise.
37691 (add_define_subst_attr): Likewise.
37692 (read_subst_mapping): Likewise.
37693 (check_code_iterator): Likewise.
37694 (rtx_reader::read_rtx): Likewise. Move one-time initialization
37696 (one_time_initialization): New function.
37697 (rtx_reader::read_until): New method.
37698 (read_flags): New function.
37699 (parse_reg_note_name): New function.
37700 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
37701 Handle reuse_rtx ids.
37702 Wrap iterator lookup within #ifdef GENERATOR_FILE.
37703 Add parsing support for RTL dumps, mirroring the special-cases in
37704 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
37705 values, and calling handle_any_trailing_information.
37706 (rtx_reader::read_rtx_operand): Convert return type from void
37707 to rtx, returning return_rtx. Handle case 'e'. Call
37708 finalize_string on XSTR and XTMPL fields.
37709 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
37710 "(nil)" values were omitted. Call the postprocess vfunc on the
37712 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
37713 class ctor. Initialize m_in_call_function_usage. Call
37714 one_time_initialization.
37715 * rtl-tests.c (selftest::test_uncond_jump): Call
37716 set_new_first_and_last_insn.
37717 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
37718 * selftest-rtl.c: New file.
37719 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
37720 (selftest::get_insn_by_uid): New decl.
37721 * selftest-run-tests.c (selftest::run_tests): Call
37722 read_rtl_function_c_tests.
37723 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
37724 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
37727 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
37729 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
37730 operands in a special way. Assert that pos+len <= mode precision.
37732 2017-01-05 Jakub Jelinek <jakub@redhat.com>
37734 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
37735 3 argument Alias with unlimited for the negative form.
37736 (fno-vect-cost-model): Removed.
37738 2017-01-05 Martin Liska <mliska@suse.cz>
37740 * hsa-gen.c (gen_hsa_divmod): New function.
37741 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
37743 2017-01-05 Martin Liska <mliska@suse.cz>
37746 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
37749 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37751 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
37752 small constant length operands.
37754 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37756 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
37757 between loop iterations.
37759 2017-01-05 Martin Liska <mliska@suse.cz>
37762 * gimplify.c (gimplify_decl_expr): Compare to
37763 asan_poisoned_variables instread of checking flags.
37764 (gimplify_target_expr): Likewise.
37765 (gimplify_expr): Likewise.
37766 (gimplify_function_tree): Conditionally initialize
37767 asan_poisoned_variables.
37769 2017-01-04 Jeff Law <law@redhat.com>
37771 PR tree-optimizatin/78812
37772 * rtl.h (contains_mem_rtx_p): Prototype.
37773 * ifcvt.c (containts_mem_rtx_p): Move from here to...
37774 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
37775 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
37776 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
37777 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
37779 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37781 * input.c (assert_char_at_range): Default-initialize actual_range.
37783 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37785 * df-scan.c (df_ref_create_structure): Make regno unsigned,
37786 to match the caller.
37788 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37790 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
37791 insns after final jump in test to emit dummy move.
37793 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37795 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
37796 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
37798 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
37800 * multiple_target.c (create_dispatcher_calls): Init e_next.
37801 * tree-ssa-loop-split.c (split_loop): Init border.
37802 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
37805 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
37810 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
37811 (altivec_register_operand): Do not return true if the operand
37812 contains a SUBREG mixing SImode and SFmode.
37813 (vsx_register_operand): Likewise.
37814 (vsx_reg_sfsubreg_ok): New predicate.
37815 (vfloat_operand): Do not return true if the operand contains a
37816 SUBREG mixing SImode and SFmode.
37817 (vint_operand): Likewise.
37818 (vlogical_operand): Likewise.
37819 (gpc_reg_operand): Likewise.
37820 (int_reg_operand): Likewise.
37821 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
37822 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
37823 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
37825 (rs6000_emit_move_si_sf_subreg): New helper function.
37826 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
37827 fixup SUBREGs involving SImode and SFmode.
37828 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
37829 numbers for the new peephole2 optimization.
37830 (peephole2 for SFmode unions): New peephole2 to optimize cases in
37831 the GLIBC math library that do AND/IOR/XOR operations on single
37832 precision floating point.
37833 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
37834 target macros to say whether we need to avoid SUBREGs mixing
37836 (TARGET_ALLOW_SF_SUBREG): Likewise.
37837 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
37838 (UNSPEC_SI_FROM_SF): Likewise.
37839 (iorxor): Change spacing.
37840 (and_ior_xor): New iterator for AND, IOR, and XOR.
37841 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
37842 (movdi_from_sf_zero_ext): Likewise.
37843 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
37844 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
37845 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
37846 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
37847 (fms<mode>4): Likewise.
37848 (fnma<mode>4): Likewise.
37849 (fnms<mode>4): Likewise.
37850 (nfma<mode>4): Likewise.
37851 (nfms<mode>4): Likewise.
37853 2017-01-04 Marek Polacek <polacek@redhat.com>
37856 * doc/invoke.texi: Document -Wpointer-compare.
37858 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37860 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
37863 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
37864 descriptions for -gdwarf-5 and emit them as uleb128 instead of
37867 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
37870 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
37871 documentation of the powerpc_popcntb_ok attribute.
37872 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
37873 code to issue warning messages if a requested CPU configuration is
37874 not supported by the binary (assembler and loader) toolchain.
37875 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
37876 made to define a built-in function that has been disabled.
37877 (paired_init_builtins): Add assertion to prevent ICE if attempt is
37878 made to define a built-in function that has been disabled.
37879 (altivec_init_builtins): Add comment explaining why definition
37880 of the DST built-in functions is not preceded by an assertion
37881 check. Add assertions to prevent ICE if attempts are made to
37882 define an altivec predicate or an abs* built-in function that has
37884 (htm_init_builtins): Add comment explaining why definition of the
37885 htm built-in functions is not preceded by an assertion check.
37887 2017-01-04 Jeff Law <law@redhat.com>
37889 PR tree-optimizatin/67955
37890 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
37891 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
37892 the points-to solution does not include pt_null. Use DECL_PT_UID
37895 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
37897 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
37898 Use gen_int_mode instead of gen_lopwart for const_int operands.
37900 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37902 PR tree-optimization/71563
37903 * match.pd: Simplify X << Y into X if Y is known to be 0 or
37904 out of range value - has low bits known to be zero.
37906 2017-01-04 Alan Modra <amodra@gmail.com>
37908 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
37909 * configure: Regenerate.
37910 * config.in: Regenerate.
37912 2017-01-04 Jakub Jelinek <jakub@redhat.com>
37915 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
37916 a substring of the message, but strcmp with the whole message. Ifdef
37917 ENABLE_NLS, translate the message first using dgettext.
37919 2017-01-03 Jeff Law <law@redhat.com>
37921 PR tree-optimizatin/78856
37922 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
37923 (mark_threaded_blocks): Remove code to truncate thread paths that
37924 cross multiple loop headers. Instead invalidate the cached loop
37925 iteration information and handle case of a thread path walking
37926 into an irreducible region.
37928 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
37931 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
37932 assertions. Add support for doing the signbit if the IEEE 128-bit
37933 floating point value is in a GPR.
37934 * config/rs6000/rs6000.md (Fsignbit): Delete.
37935 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
37936 Update the length attribute if the value is in a GPR.
37937 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
37938 the sign or zero extension instruction, since the value is always 0/1.
37939 (signbit<mode>2_dm2): Delete using <Fsignbit>.
37942 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
37943 extracting SImode to a GPR register so that we can generate a
37944 store, limit the vector to be in a traditional Altivec register
37945 for the vextuwrx instruction.
37947 2017-01-03 Ian Lance Taylor <iant@google.com>
37949 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
37951 2017-01-03 Martin Sebor <msebor@redhat.com>
37953 PR tree-optimization/78696
37954 * gimple-ssa-sprintf.c (format_floating): Correct handling of
37955 precision. Use MPFR for %f for greater fidelity. Correct handling
37957 (pass_sprintf_length::compute_format_length): Set width and precision
37958 specified by asrerisk to void_node for vararg functions.
37959 (try_substitute_return_value): Adjust dump output.
37961 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
37963 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
37965 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
37967 * doc/invoke.texi (SPARC options): Document -mlra as the default.
37968 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
37969 -mlra/-mno-lra was passed to the compiler.
37971 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
37973 PR rtl-optimization/65618
37974 * emit-rtl.c (try_split): Move initialization of "before" and
37975 "after" to just before the call to emit_insn_after_setloc.
37977 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
37979 * doc/md.texi (Standard Names): Remove reference to Java frontend.
37981 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
37983 * dwarf2out.c (gen_enumeration_type_die): When
37984 -gno-strict-dwarf, add a DW_AT_encoding attribute.
37986 2017-01-03 Jakub Jelinek <jakub@redhat.com>
37988 PR tree-optimization/78965
37989 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
37990 Change first argument from const call_info & to call_info &. For %n
37991 set info.nowrite to false.
37993 PR middle-end/78901
37994 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
37995 possibly throwing calls.
37997 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
37998 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
37999 and fns handling, rather than in a separate case SSA_NAME.
38001 2017-01-02 Jeff Law <law@redhat.com>
38003 * config/darwin-driver.c (darwin_driver_init): Const-correctness
38004 fixes for first_period and second_period variables.
38006 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
38009 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
38010 (*insvqi_1): New insn pattern.
38011 (*insvqi_1_mem_rex64): Ditto.
38012 (*insvqi_2): Ditto.
38013 (*insvqi_3): Rename from *insvqi.
38015 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
38017 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
38019 * doc/cfg.texi (Edges): Remove reference to Java.
38020 (Maintaining the CFG): Ditto.
38022 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38024 PR middle-end/77674
38025 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
38026 transparent aliases.
38028 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38030 PR middle-end/77484
38031 * predict.def (PRED_CALL): Update hitrate.
38032 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
38033 * predict.c (tree_estimate_probability_bb): Split CALL predictor
38034 into direct/indirect/polymorphic variants.
38036 2017-01-01 Jakub Jelinek <jakub@redhat.com>
38038 Update copyright years.
38040 * gcc.c (process_command): Update copyright notice dates.
38041 * gcov-dump.c (print_version): Ditto.
38042 * gcov.c (print_version): Ditto.
38043 * gcov-tool.c (print_version): Ditto.
38044 * gengtype.c (create_file): Ditto.
38045 * doc/cpp.texi: Bump @copying's copyright year.
38046 * doc/cppinternals.texi: Ditto.
38047 * doc/gcc.texi: Ditto.
38048 * doc/gccint.texi: Ditto.
38049 * doc/gcov.texi: Ditto.
38050 * doc/install.texi: Ditto.
38051 * doc/invoke.texi: Ditto.
38053 Copyright (C) 2017 Free Software Foundation, Inc.
38055 Copying and distribution of this file, with or without modification,
38056 are permitted in any medium without royalty provided the copyright
38057 notice and this notice are preserved.