1 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * optabs-query.h (convert_optab_p): New function, split out from...
6 (convert_optab_handler): ...here.
7 (widening_optab_handler): Delete.
8 (find_widening_optab_handler): Remove permit_non_widening parameter.
9 (find_widening_optab_handler_and_mode): Likewise. Provide an
10 override that operates on mode class wrappers.
11 * optabs-query.c (widening_optab_handler): Delete.
12 (find_widening_optab_handler_and_mode): Remove permit_non_widening
13 parameter. Assert that the two modes are the same class and that
14 the "from" mode is narrower than the "to" mode. Use
15 convert_optab_handler instead of widening_optab_handler.
16 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
17 instead of widening_optab_handler.
18 * expr.c (expand_expr_real_2): Update calls to
19 find_widening_optab_handler.
20 * optabs.c (expand_widen_pattern_expr): Likewise.
21 (expand_binop_directly): Take the insn_code as a parameter.
22 (expand_binop): Only call find_widening_optab_handler for
23 conversion optabs; use optab_handler otherwise. Update calls
24 to find_widening_optab_handler and expand_binop_directly.
25 Use convert_optab_handler instead of widening_optab_handler.
26 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
27 find_widening_optab_handler and use scalar_mode rather than
29 (convert_plusminus_to_widen): Likewise.
31 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
32 Alan Hayward <alan.hayward@arm.com>
33 David Sherwood <david.sherwood@arm.com>
35 * machmode.h (fixed_size_mode): New class.
36 * rtl.h (get_pool_mode): Return fixed_size_mode.
37 * gengtype.c (main): Add fixed_size_mode.
38 * target.def (get_raw_result_mode): Return a fixed_size_mode.
39 (get_raw_arg_mode): Likewise.
40 * doc/tm.texi: Regenerate.
41 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
42 * targhooks.c (default_get_reg_raw_mode): Likewise.
43 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
44 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
45 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
46 (msp430_get_raw_result_mode): Likewise.
47 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
48 * dbxout.c (dbxout_parms): Require fixed-size modes.
49 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
50 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
51 * omp-low.c (lower_oacc_reductions): Likewise.
52 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
53 (simplify_subreg): Update accordingly.
54 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
55 (force_const_mem): Update accordingly. Return NULL_RTX for modes
56 that aren't fixed-size.
57 (get_pool_mode): Return a fixed_size_mode.
58 (output_constant_pool_2): Take a fixed_size_mode.
60 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
61 Alan Hayward <alan.hayward@arm.com>
62 David Sherwood <david.sherwood@arm.com>
64 * doc/rtl.texi (vec_series): Document.
65 (const): Say that the operand can be a vec_series.
66 * rtl.def (VEC_SERIES): New rtx code.
67 * rtl.h (const_vec_series_p_1): Declare.
68 (const_vec_series_p): New function.
69 * emit-rtl.h (gen_const_vec_series): Declare.
70 (gen_vec_series): Likewise.
71 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
72 (gen_vec_series): Likewise.
73 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
74 * simplify-rtx.c (simplify_unary_operation): Handle negations
76 (simplify_binary_operation_series): New function.
77 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
78 (test_vector_ops_series): New function.
79 (test_vector_ops): Call it.
80 * config/powerpcspe/altivec.md (altivec_lvsl): Use
82 (altivec_lvsr): Likewise.
83 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
85 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
86 Alan Hayward <alan.hayward@arm.com>
87 David Sherwood <david.sherwood@arm.com>
89 * doc/rtl.texi (const): Update description of address constants.
90 Say that vector constants are allowed too.
91 * common.md (E, F): Use CONSTANT_P instead of checking for
93 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
94 checking for CONST_VECTOR.
95 * expmed.c (make_tree): Use build_vector_from_val for a CONST
97 * expr.c (expand_expr_real_2): Check for vector modes instead
98 of checking for CONST_VECTOR.
99 * rtl.h (const_vec_p): New function.
100 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
101 (unwrap_const_vec_duplicate): Handle them here too.
103 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
104 David Malcolm <dmalcolm@redhat.com>
105 Alan Hayward <alan.hayward@arm.com>
106 David Sherwood <david.sherwood@arm.com>
108 * rtl.h (vec_duplicate_p): New function.
109 * selftest-rtl.c (assert_rtx_eq_at): New function.
110 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
111 (assert_rtx_eq_at): Declare.
112 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
113 * selftest-run-tests.c (selftest::run_tests): Call it.
114 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
115 (simplify_unary_operation_1): Recursively handle vector duplicates.
116 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
118 (simplify_subreg): Handle subregs of vector duplicates.
119 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
120 (selftest::simplify_rtx_c_tests): New functions.
122 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
123 Alan Hayward <alan.hayward@arm.com>
124 David Sherwood <david.sherwood@arm.com>
126 * emit-rtl.h (gen_const_vec_duplicate): Declare.
127 (gen_vec_duplicate): Likewise.
128 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
130 (gen_const_vector): ...here.
131 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
132 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
133 whose elements are all equal.
134 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
135 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
136 (simplify_relational_operation): Likewise.
137 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
139 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
140 (aarch64_expand_vector_init): Likewise.
141 * config/arm/arm.c (neon_vdup_constant): Likewise.
142 (neon_expand_vector_init): Likewise.
143 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
144 (arm_block_set_unaligned_vect): Likewise.
145 (arm_block_set_aligned_vect): Likewise.
146 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
147 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
148 (expand_vec_perm_even_odd_pack): Likewise.
149 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
150 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
151 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
152 gen_const_vec_duplicate.
153 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
154 * config/mips/mips.c (mips_gen_const_int_vector): Use
155 gen_const_vec_duplicate.
156 (mips_expand_vector_init): Use CONST0_RTX.
157 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
158 (define_split): Use gen_const_vec_duplicate.
159 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
160 (define_split): Use gen_const_vec_duplicate.
161 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
162 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
163 * config/spu/spu.c (spu_const): Likewise.
165 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
166 Alan Hayward <alan.hayward@arm.com>
167 David Sherwood <david.sherwood@arm.com>
169 * combine.c (can_change_dest_mode): Reject changes in
170 REGMODE_NATURAL_SIZE.
172 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
174 * configure.ac (--enable-libssp): New.
175 (gcc_cv_libc_provides_ssp): Check for explicit setting before
176 trying to determine target-specific default. Adjust indentation.
177 * configure: Regenerated.
178 * doc/install.texi (Configuration): Expand --disable-libssp
181 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
183 config/i386/i386.c (ix86_expand_epilogue): Correct stack
186 2017-10-31 Martin Jambor <mjambor@suse.cz>
189 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
191 2017-10-31 David Malcolm <dmalcolm@redhat.com>
193 * auto-profile.c (autofdo_source_profile::read): Use
194 UNKNOWN_LOCATION rather than 0.
195 * diagnostic-core.h (warning_at_rich_loc): Rename to...
196 (warning_at): ...this overload.
197 (warning_at_rich_loc_n): Rename to...
198 (warning_n): ...this overload.
199 (error_at_rich_loc): Rename to...
200 (error_at): ...this overload.
201 (pedwarn_at_rich_loc): Rename to...
202 (pedwarn): ...this overload.
203 (permerror_at_rich_loc): Rename to...
204 (permerror): ...this overload.
205 (inform_at_rich_loc): Rename to...
206 (inform): ...this overload.
207 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
208 (diagnostic_n_impl_richloc): Rename to...
209 (diagnostic_n_impl): ...this rich_location *-based decl.
210 (inform_at_rich_loc): Rename to...
211 (inform): ...this, and add an assertion.
212 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
213 (warning_at_rich_loc): Rename to...
214 (warning_at): ...this, and add an assertion.
215 (warning_at_rich_loc_n): Rename to...
216 (warning_n): ...this, and add an assertion.
217 (warning_n): Update location_t-based implementation for removal of
218 location_t-based diagnostic_n_impl.
219 (pedwarn_at_rich_loc): Rename to...
220 (pedwarn): ...this, and add an assertion.
221 (permerror_at_rich_loc): Rename to...
222 (permerror): ...this, and add an assertion.
223 (error_n): Update for removal of location_t-based diagnostic_n_impl.
224 (error_at_rich_loc): Rename to...
225 (error_at): ...this, and add an assertion.
226 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
227 (driver::do_spec_on_infiles): Likewise.
228 * substring-locations.c (format_warning_va): Update for renaming
229 of inform_at_rich_loc.
231 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
233 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
234 _Float<N>X built-in functions so that the variant without the
235 "__builtin_" prefix is only enabled for the GNU C and Objective C
236 languages when they are in non-strict ANSI/ISO mode.
237 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
238 * target.def (floatn_builtin_p): Add a target hook to control
239 whether _Float<N> and _Float<N>X built-in functions without the
240 "__builtin_" prefix are enabled, and return true for C and
241 Objective C in the default hook. Include langhooks.h in
243 * targhooks.h (default_floatn_builtin_p): Likewise.
244 * targhooks.c (default_floatn_builtin_p): Likewise.
245 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
246 floatn_builtin_p target hook.
247 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
249 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
250 Eric Botcazou <ebotcazou@adacore.com>
252 PR rtl-optimization/81803
253 * lra-constraints.c (curr_insn_transform): Also reload the whole
254 register for a strict subreg no wider than a word if this is for
255 a WORD_REGISTER_OPERATIONS target.
257 2017-10-31 Jason Merrill <jason@redhat.com>
259 * gdbinit.in: Skip over inlines from timevar.h.
261 2017-10-31 Martin Liska <mliska@suse.cz>
263 * doc/gcov.texi: Document new option.
264 * gcov.c (print_usage): Likewise print it.
265 (process_args): Support the argument.
266 (format_count): New function.
267 (format_gcov): Use the function.
269 2017-10-31 Martin Liska <mliska@suse.cz>
271 * gcov.c (struct name_map): do not use typedef.
272 Define operator== and operator<.
273 (name_search): Remove.
275 (main): Do not allocate names.
276 (process_file): Add vertical space.
277 (generate_results): Use std::find.
278 (release_structures): Do not release memory.
279 (find_source): Use std::find.
281 2017-10-31 Martin Liska <mliska@suse.cz>
283 * gcov.c (struct line_info): Remove it's typedef.
284 (line_info::line_info): Add proper ctor.
285 (line_info::has_block): Do not use a typedef.
286 (struct source_info): Do not use typedef.
288 (get_cycles_count): Likewise.
289 (output_intermediate_file): Iterate via vector iterator.
290 (add_line_counts): Use std::vector methods.
291 (accumulate_line_counts): Likewise.
292 (output_lines): Likewise.
294 2017-10-31 Martin Liska <mliska@suse.cz>
296 * gcov.c (struct source_info): Remove typedef.
297 (source_info::source_info): Add proper ctor.
298 (accumulate_line_counts): Use struct, not it's typedef.
299 (output_gcov_file): Likewise.
300 (output_lines): Likewise.
301 (main): Do not allocate an array.
302 (output_intermediate_file): Use size of vector container.
303 (process_file): Resize the vector.
304 (generate_results): Do not preallocate, use newly added vector
306 (release_structures): Do not release sources.
307 (find_source): Use vector methods.
308 (add_line_counts): Do not use typedef.
310 2017-10-31 Martin Liska <mliska@suse.cz>
312 * doc/gcov.texi: Document that.
313 * gcov.c (add_line_counts): Mark lines with a non-executed
315 (output_line_beginning): Handle such lines.
316 (output_lines): Pass new argument.
317 (output_intermediate_file): Print it in intermediate format.
319 2017-10-31 Martin Liska <mliska@suse.cz>
321 * color-macros.h: New file.
322 * diagnostic-color.c: Factor out color related to macros to
324 * doc/gcov.texi: Document -k option.
325 * gcov.c (INCLUDE_STRING): Include string.h.
326 (print_usage): Add -k option.
327 (process_args): Parse it.
328 (pad_count_string): New function.
329 (output_line_beginning): Likewise.
330 (DEFAULT_LINE_START): New macro.
331 (output_lines): Support color output.
333 2017-10-31 Martin Liska <mliska@suse.cz>
335 PR gcov-profile/82633
336 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
337 their interaction with GCOV infrastructure.
338 * configure.ac: Add -fkeep-{inline,static}-functions to
340 * configure: Regenerate.
342 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
345 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
347 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
350 * config/rs6000/rs6000.md (allocate_stack): Force update interval
351 into a register if it does not fit into an immediate offset field.
353 2017-10-31 Olivier Hainque <hainque@adacore.com>
355 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
357 2017-10-31 Julia Koval <julia.koval@intel.com>
359 * config.gcc: Add gfniintrin.h.
360 * config/i386/gfniintrin.h: New.
361 * config/i386/i386-builtin-types.def
362 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
363 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
364 __builtin_ia32_vgf2p8affineinvqb_v32qi,
365 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
366 __builtin_ia32_vgf2p8affineinvqb_v16qi,
367 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
368 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
369 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
370 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
371 V64QI_FTYPE_V64QI_V64QI_INT): New types.
372 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
373 * config/i386/immintrin.h: Include gfniintrin.h.
374 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
376 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
378 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
380 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
382 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
383 (arm_ashldi3_1bit): Remove pattern.
384 (ashrdi3): Remove shift by 1 expansion.
385 (arm_ashrdi3_1bit): Remove pattern.
386 (lshrdi3): Remove shift by 1 expansion.
387 (arm_lshrdi3_1bit): Remove pattern.
388 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
389 cost of ashldi3 by 1.
390 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
391 (<shift>di3_neon): Likewise.
393 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
395 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
396 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
397 and (*aarch64_simd_mov<VQ:mode>).
398 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
400 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
401 attributes to match pattern alternative.
403 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
405 * config.gcc (powerpc*-*-*): Add emmintrin.h.
406 * config/rs6000/emmintrin.h: New file.
407 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
409 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
411 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
412 * (movdi_vfp_cortexa8): Remove pattern.
414 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
416 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
418 (Specific, alpha*-dec-osf5.1): Remove.
419 (Specific, mips-sgi-irix5): Remove.
420 (Specific, mips-sgi-irix6): Remove.
422 2017-10-30 Jakub Jelinek <jakub@redhat.com>
425 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
426 arguments to clear_bit_region_be.
428 2017-10-30 Jim Wilson <wilson@tuliptree.org>
430 * gimplify.c: Include memmodel.h.
432 2017-10-30 Martin Jambor <mjambor@suse.cz>
434 * omp-grid.c (grid_attempt_target_gridification): Also insert a
435 condition whether loop should be executed at all.
437 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
439 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
440 gimple folding of vec_madd() intrinsics.
441 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
442 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
443 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
445 2017-10-30 Richard Biener <rguenther@suse.de>
447 PR tree-optimization/82762
449 2017-10-23 Richard Biener <rguenther@suse.de>
451 PR tree-optimization/82129
453 2017-08-01 Richard Biener <rguenther@suse.de>
455 PR tree-optimization/81181
456 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
457 (compute_antic): ... end of iteration here.
459 2017-10-30 Joseph Myers <joseph@codesourcery.com>
461 * doc/invoke.texi (C Dialect Options): Document -std=c17,
462 -std=iso9899:2017 and -std=gnu17.
463 * doc/standards.texi (C Language): Document C17 support.
464 * doc/cpp.texi (Overview): Mention -std=c17.
465 (Standard Predefined Macros): Document C11 and C17 values of
466 __STDC_VERSION__. Do not refer to C99 support as incomplete.
467 * doc/extend.texi (Inline): Do not list individual options for
468 standards newer than C99.
469 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
471 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
474 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
476 * asan.c (asan_finish_file): Align asan globals array by shadow
479 2017-10-30 Jakub Jelinek <jakub@redhat.com>
482 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
483 (struct store_immediate_info): Add bitregion_start and bitregion_end
485 (store_immediate_info::store_immediate_info): Add brs and bre
486 arguments and initialize bitregion_{start,end} from those.
487 (struct merged_store_group): Add bitregion_start, bitregion_end,
488 align_base and mask fields. Drop unnecessary struct keyword from
489 struct store_immediate_info. Add do_merge method.
490 (clear_bit_region_be): Use memset instead of loop storing zeros.
491 (merged_store_group::do_merge): New method.
492 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
493 stores as long as the surrounding bitregions have no gaps.
494 (merged_store_group::merge_overlapping): Use do_merge.
495 (merged_store_group::apply_stores): Test that bitregion_{start,end}
496 is byte aligned, rather than requiring that start and width are
497 byte aligned. Drop unnecessary struct keyword from
498 struct store_immediate_info. Allocate and populate also mask array.
499 Make start of the arrays relative to bitregion_start rather than
500 start and size them according to bitregion_{end,start} difference.
501 (struct imm_store_chain_info): Drop unnecessary struct keyword from
502 struct store_immediate_info.
503 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
504 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
505 struct keyword from struct store_immediate_info.
506 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
507 between stores as long as the surrounding bitregions have no gaps.
509 (struct split_store): Add orig non-static data member.
510 (split_store::split_store): Initialize orig to false.
511 (find_constituent_stmts): Return store_immediate_info *, non-NULL
512 if there is exactly a single original stmt. Change stmts argument
513 to pointer from reference, if NULL, don't push anything to it. Add
514 first argument, use it to optimize skipping over orig stmts that
515 are known to be before bitpos already. Simplify.
516 (split_group): Return unsigned int count how many stores are or
517 would be needed rather than a bool. Add allow_unaligned argument.
518 Change split_stores argument from reference to pointer, if NULL,
519 only do a dry run computing how many stores would be produced.
520 Rewritten algorithm to use both alignment and misalign if
521 !allow_unaligned and handle bitfield stores with gaps.
522 (imm_store_chain_info::output_merged_store): Set start_byte_pos
523 from bitregion_start instead of start. Compute allow_unaligned
524 here, if true, do 2 split_group dry runs to compute which one
525 produces fewer stores and prefer aligned if equal. Punt if
526 new count is bigger or equal than original before emitting any
527 statements, rather than during that. Remove no longer needed
528 new_ssa_names tracking. Replace num_stmts with
529 split_stores.length (). Use 32-bit stack allocated entries
530 in split_stores auto_vec. Try to reuse original store lhs/rhs1
531 if possible. Handle bitfields with gaps.
532 (pass_store_merging::execute): Ignore bitsize == 0 stores.
533 Compute bitregion_{start,end} for the stores and construct
534 store_immediate_info with that. Formatting fixes.
536 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
539 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
540 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
542 2017-10-29 Jim Wilson <wilson@tuliptree.org>
544 * gimplify.c: Include tm_p.h.
546 * common.opt (gcoff): Re-add as ignored option.
547 (gcoff1, gcoff2, gcoff3): Likewise.
549 * Makefile.in (OBJS): Delete sdbout.o.
550 (GTFILES): Delete $(srcdir)/sdbout.c.
551 * debug.h: Delete sdb_debug_hooks.
552 * final.c: Delete sdbout.h include.
553 (final_scan_insn): Delete SDB_DEBUG check.
554 (rest_of_clean_state): Likewise.
555 * output.h: Delete sdb_begin_function_line.
558 * toplev.c: Delete sdbout.h include.
559 (process_options): Delete SDB_DEBUG check.
560 * tree-core.h (tree_type_common): Delete pointer field of
562 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
564 * tree.h (TYPE_SYMTAB_POINTER): Delete.
565 (TYPE_SYMTAB_IS_POINTER): Delete.
566 (TYPE_SYMTAB_IS_DIE): Renumber.
567 * xcoffout.c: Refer to former sdbout.c file.
568 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
570 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
571 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
572 (Debugging Options): Delete -gcoff.
573 (-gstabs): Delete SDB reference.
575 (-gcoff@var{level}): Delete.
576 * doc/passes.texi (Debugging information output): Delete SDB and
578 * doc/tm.texi: Regenerate.
579 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
580 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
582 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
583 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
584 (SDB_DEBUGGING_INFO): Delete.
585 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
586 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
587 * target.def (output_source_filename): Delete COFF reference.
589 * common.opt (gcoff): Delete.
590 (gxcoff+): Update Negative chain.
591 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
593 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
594 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
595 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
597 (expand_function_start): Change sdb reference to past tense.
598 (expand_function_end): Change sdb reference to past tense.
599 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
600 * opts.c (debug_type_names): Delete coff entry.
601 (common_handle_option): Delete OPT_gcoff case.
602 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
604 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
605 * config/cris/cris.h: Delete SDB reference in comment.
606 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
607 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
608 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
609 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
611 (ix86_expand_prologue): Likewise.
612 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
613 * config/ia64/ia64.h: Likewise.
614 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
615 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
617 * config/mmix/mmix.h: Likewise.
618 * config/nds32/nds32.c: Likewise.
619 * config/stormy/storym16.h: Likewise.
620 * config/visium/visium.h: Likewise.
621 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
623 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
625 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
626 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
627 FRAME_POINTER_REGNUM point at high end of local var area.
629 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
631 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
632 Move comment around. Do not reset best_edge for a copiable
633 destination if the copy would cause a partition change.
634 (better_edge_p): Remove redundant check.
636 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
638 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
640 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
642 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
643 for math functions that have _Float<N> and _Float<N>X variants.
644 (mathfn_built_in_2): Add support for math functions that have
645 _Float<N> and _Float<N>X variants.
646 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
647 (expand_builtin_mathfn_ternary): Add support for fma with
648 _Float<N> and _Float<N>X variants.
649 (expand_builtin): Likewise.
650 (fold_builtin_3): Likewise.
651 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
652 create math function _Float<N> and _Float<N>X variants as external
654 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
655 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
656 the __builtin_ prefix and if not strict ansi, without the prefix.
657 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
658 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
659 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
660 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
661 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
662 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
663 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
664 function signatures for fma _Float<N> and _Float<N>X variants.
665 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
666 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
667 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
668 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
669 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
670 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
671 * gencfn-macros.c (print_case_cfn): Add support for math functions
672 that have _Float<N> and _Float<N>X variants.
673 (print_define_operator_list): Likewise.
674 (fltfn_suffixes): Likewise.
676 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
677 for math functions that have _Float<N> and _Float<N>X variants.
678 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
679 and _Float<N>X variants.
680 (COPYSIGN): Likewise.
683 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
684 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
686 (integer_valued_read_call_p): Likewise.
687 * fold-const-call.c (fold_const_call_ss): Likewise.
688 (fold_const_call_sss): Add support for copysign, fmin, and fmax
689 _Float<N> and _Float<N>X variants.
690 (fold_const_call_ssss): Add support for fma _Float<N> and
692 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
693 support for copysign and fma _Float<N> and _Float<N>X variants.
694 (backprop::process_builtin_call_use): Likewise.
695 * tree-call-cdce.c (can_test_argument_range); Add support for
696 sqrt _Float<N> and _Float<N>X variants.
697 (edom_only_function): Likewise.
698 (get_no_error_domain): Likewise.
699 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
700 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
701 copysign _Float<N> and _Float<N>X variants.
702 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
703 handled by machine independent code.
705 * doc/cpp.texi (Common Predefined Macros): Document defining
706 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
707 fma _Float<N> and _Float<N>X variants.
709 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
712 * config/i386/i386-modes.def (CCFPU): Remove definition.
713 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
714 (ix86_cc_modes_compatible): Ditto.
715 (ix86_expand_carry_flag_compare): Ditto.
716 (ix86_expand_int_movcc): Ditto.
717 (ix86_expand_int_addcc): Ditto.
718 (ix86_reverse_condition): Ditto.
719 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
720 Return true/false for unordered/ordered fp comparisons.
721 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
722 (ix86_prepare_fp_compare_args): Update for rename.
723 (ix86_expand_fp_compare): Update for rename. Generate unordered
724 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
725 (ix86_expand_sse_compare_and_jump): Ditto.
726 * config/i386/predicates.md (fcmov_comparison_operator):
727 Remove CCFPU mode handling.
728 (ix86_comparison_operator): Ditto.
729 (ix86_carry_flag_operator): Ditto.
730 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
731 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
732 (*cmpu<mode>_cc_i387): Ditto.
733 (FPCMP): Remove mode iterator.
734 (unord): Remove mode attribute.
735 (unord_subst): New define_subst transformation
736 (unord): New define_subst attribute.
738 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
739 (*cmpi<unord>xf_i387): Ditto.
740 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
741 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
742 using unord_subst transformation.
743 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
744 (round_saeonly): Also handle CCFP mode.
745 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
746 Remove UNSPEC_SAHF unspec handling.
748 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
750 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
752 2017-10-27 Jeff Law <law@redhat.com>
754 * gimple-ssa-sprintf.c: Include domwalk.h.
755 (class sprintf_dom_walker): New class, derived from dom_walker.
756 (sprintf_dom_walker::before_dom_children): New function.
757 (struct call_info): Moved into sprintf_dom_walker class
758 (compute_formath_length, handle_gimple_call): Likewise.
759 (sprintf_length::execute): Call the dominator walker rather
760 than walking the statements.
762 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
763 gimple statement locations.
764 (check_array_bounds): Corresponding changes. Get the statement's
765 location directly from wi->stmt.
767 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
770 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
772 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
774 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
775 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
777 2017-10-27 Jakub Jelinek <jakub@redhat.com>
780 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
781 * config/i386/i386.c (maybe_get_pool_constant): Removed.
782 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
783 maybe_get_pool_constant.
784 * config/i386/predicates.md (zero_extended_scalar_load_operand):
787 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
789 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
790 2.26 caveat. Update gas and gld versions.
791 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
794 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
796 * cgraph.h (set_malloc_flag): Declare.
797 * cgraph.c (set_malloc_flag_1): New function.
798 (set_malloc_flag): Likewise.
799 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
800 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
802 (read_ipa_call_summary): Add support for reading is_return_callee.
803 (write_ipa_call_summary): Stream is_return_callee.
804 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
805 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
806 ipa-prop.h, ipa-fnsummary.h.
807 (pure_const_names): Change to static.
808 (malloc_state_e): Define.
809 (malloc_state_names): Define.
810 (funct_state_d): Add field malloc_state.
811 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
812 (check_retval_uses): New function.
813 (malloc_candidate_p): Likewise.
814 (analyze_function): Add support for malloc attribute.
815 (pure_const_write_summary): Stream malloc_state.
816 (pure_const_read_summary): Add support for reading malloc_state.
817 (dump_malloc_lattice): New function.
818 (propagate_malloc): New function.
819 (warn_function_malloc): New function.
820 (ipa_pure_const::execute): Call propagate_malloc and
822 (pass_local_pure_const::execute): Add support for malloc attribute.
823 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
824 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
826 2017-10-27 Martin Liska <mliska@suse.cz>
828 PR gcov-profile/82457
829 * doc/invoke.texi: Document that one needs a non-strict ISO mode
830 for fork-like functions to be properly instrumented.
832 2017-10-27 Richard Biener <rguenther@suse.de>
835 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
836 info when we redirected EH.
838 2017-10-26 Michael Collison <michael.collison@arm.com>
840 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
842 (<optab>_trunchf<GPI:mode>2: New pattern.
843 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
844 * config/aarch64/iterators.md (wv): New mode attribute.
845 (vf, VF): New mode attributes.
846 (vgp, VGP): New mode attributes.
847 (s): Update attribute with SImode and DImode prefixes.
849 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
851 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
852 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
853 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
854 (nios2_option_overide): Initialize it. Don't allow R0-relative
856 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
857 (nios2_symbolic_constant_p): Likewise.
858 (nios2_legitimate_address_p): Likewise.
859 (nios2_r0rel_section_name_p): New.
860 (nios2_symbol_ref_in_r0rel_data_p): New.
861 (nios2_emit_move_sequence): Handle r0rel_constant_p.
862 (r0rel_constant_p): New.
863 (nios2_print_operand_address): Handle r0rel_constant_p.
864 (nios2_cdx_narrow_form_p): Likewise.
865 * config/nios2/nios2.opt (mr0rel-sec=): New option.
866 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
867 (Nios II Options): Document -mr0rel-sec.
869 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
871 * config/nios2/nios2.c: Include xregex.h.
872 (nios2_gprel_sec_regex): New.
873 (nios2_option_overide): Initialize it. Don't allow GP-relative
875 (nios2_small_section_name_p): Check for regex match.
876 * config/nios2/nios2.opt (mgprel-sec=): New option.
877 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
878 (Nios II Options): Document -mgprel-sec.
880 2017-10-26 Jim Wilson <wilson@tuliptree.org>
882 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
884 2017-10-26 Tom de Vries <tom@codesourcery.com>
886 PR tree-optimization/82707
887 * gimple.c (gimple_copy): Fix unsharing of
888 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
890 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
892 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
893 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
894 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
895 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
896 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
897 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
898 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
899 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
900 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
901 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
902 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
903 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
904 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
905 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
906 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
907 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
908 _mm512_mask_cmpunord_ps_mask): New intrinsics.
910 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
912 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
914 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
915 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
916 warning to rs6000.c. Remove the Undocumented flag, since it has
918 (-mabi=ibmlongdouble): Likewise.
919 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
920 already set, set the default format for long double.
921 (rs6000_debug_reg_global): Print whether long double is IBM or
923 (rs6000_option_override_internal): Rework setting long double
924 format. Only warn if the user is changing the long double default
925 and they did not use -Wno-psabi.
926 * doc/invoke.texi (PowerPC options): Update the documentation for
927 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
929 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
930 Alan Hayward <alan.hayward@arm.com>
931 David Sherwood <david.sherwood@arm.com>
933 * rtl.h (wider_subreg_mode): New function.
934 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
935 rather than an unsigned int *.
936 * ira-color.c (regno_max_ref_width): Replace with...
937 (regno_max_ref_mode): ...this new variable.
938 (coalesced_pseudo_reg_slot_compare): Update accordingly.
939 Use wider_subreg_mode.
940 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
941 rather than an unsigned int *.
942 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
943 (process_alt_operands): Likewise.
944 (invariant_p): Likewise.
945 * lra-spills.c (assign_mem_slot): Likewise.
946 (add_pseudo_to_slot): Likewise.
947 * lra.c (collect_non_operand_hard_regs): Likewise.
948 (add_regs_to_insn_regno_info): Likewise.
949 * reload1.c (regno_max_ref_width): Replace with...
950 (regno_max_ref_mode): ...this new variable.
951 (reload): Update accordingly. Update call to
952 ira_sort_regnos_for_alter_reg.
953 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
954 (init_eliminable_invariants): Update to use regno_max_ref_mode.
955 (scan_paradoxical_subregs): Likewise.
957 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
959 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
960 (aarch64_frame): Add emit_frame_chain boolean.
961 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
962 Move eh_return case to aarch64_layout_frame.
963 (aarch64_layout_frame): Initialize emit_frame_chain.
964 (aarch64_expand_prologue): Use emit_frame_chain.
966 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
968 * config/aarch64/aarch64.c (aarch64_layout_frame):
969 Ensure LR is always stored at the bottom of the callee-saves.
970 Remove rarely used frame layout which saves callee-saves at top of
971 frame, so the store of LR can be used as a valid probe in all cases.
973 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
975 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
976 Improve unaligned TImode/TFmode base/offset split.
978 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
979 Alan Hayward <alan.hayward@arm.com>
980 David Sherwood <david.sherwood@arm.com>
982 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
983 * combine.c (find_single_use_1): Likewise.
984 (expand_field_assignment): Likewise.
985 (move_deaths): Likewise.
986 * lra-constraints.c (simplify_operand_subreg): Likewise.
987 (curr_insn_transform): Likewise.
988 * lra.c (collect_non_operand_hard_regs): Likewise.
989 (add_regs_to_insn_regno_info): Likewise.
990 * rtlanal.c (reg_referenced_p): Likewise.
991 (covers_regno_no_parallel_p): Likewise.
993 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
995 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
996 Don't print any bits outside the precision of the value.
997 * wide-int.cc (test_printing): Add some new tests.
999 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1001 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
1002 supports -xbrace_comment option.
1003 * configure: Regenerate.
1004 * config.in: Regenerate.
1005 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
1006 (ASM_CPU_SPEC): Use it.
1008 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1010 * target.def (static_rtx_alignment): New hook.
1011 * targhooks.h (default_static_rtx_alignment): Declare.
1012 * targhooks.c (default_static_rtx_alignment): New function.
1013 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
1014 * doc/tm.texi: Regenerate.
1015 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
1016 instead of targetm.constant_alignment. Remove call to
1018 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1019 (cris_preferred_mininum_alignment): New function, split out from...
1020 (cris_constant_alignment): ...here.
1021 (cris_static_rtx_alignment): New function.
1022 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
1024 (ix86_constant_alignment): ...here.
1025 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1026 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1027 (mmix_static_rtx_alignment): New function.
1028 * config/spu/spu.c (spu_static_rtx_alignment): New function.
1029 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1031 2017-10-26 Tamar Christina <tamar.christina@arm.com>
1034 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
1035 Add flag_trapping_math and flag_fp_int_builtin_inexact.
1037 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
1039 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
1040 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
1041 mark as a sign-extending load.
1042 (local_pic_load_u): Define.
1044 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
1047 * fold-const.c (operand_equal_for_comparison_p): Also return true
1048 if ARG0 is a simple variant of ARG1 with narrower precision.
1049 (fold_ternary_loc): Always pass unstripped operands to the predicate.
1051 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
1053 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
1055 * i386.h (processor_costs): Add gather_static, gather_per_elt,
1056 scatter_static, scatter_per_elt.
1057 * x86-tune-costs.h: Add new cost entries.
1059 2017-10-25 Richard Biener <rguenther@suse.de>
1061 * tree-ssa-sccvn.h (vn_eliminate): Declare.
1062 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
1063 class pass_fre): Move to ...
1064 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
1065 class pass_fre): ... here and adjust for statistics.
1067 2017-10-25 Jakub Jelinek <jakub@redhat.com>
1070 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
1071 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
1072 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
1073 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
1076 2017-10-25 Richard Biener <rguenther@suse.de>
1078 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
1079 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
1080 eliminate_push_avail, eliminate_insert): Move inside...
1081 (class eliminate_dom_walker): ... this class in preparation
1083 (fini_eliminate): Remove by merging with ...
1084 (eliminate): ... this function. Adjust for class changes.
1085 (pass_pre::execute): Remove fini_eliminate call.
1086 (pass_fre::execute): Likewise.
1088 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1091 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
1092 (VPERMI2, VPERMI2I): New mode iterators.
1093 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
1094 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
1096 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
1097 mode iterator. Remove 3 old define_insn patterns.
1098 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
1099 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
1100 VPERMI2 mode iterator, remove the other two expanders.
1101 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
1102 to use VPERMI2 mode iterator, add another alternative for vpermi2*
1103 instructions, remove the other two patterns.
1104 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
1105 mode iterator, remove the other two patterns.
1106 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
1107 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
1108 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
1109 and adjust argument order accordingly.
1110 (ix86_expand_vec_perm): Adjust caller.
1111 (expand_vec_perm_1): Likewise.
1112 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
1113 (expand_vec_perm_vpermt2_vpshub2): ... this.
1114 (ix86_expand_vec_perm_const_1): Adjust caller.
1115 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
1118 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
1119 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
1120 (vec_shl_<mode>): Remove unused expander.
1121 (avx512bw_<shift_insn><mode>3): New define_insn.
1122 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
1123 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
1125 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
1128 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
1131 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1133 PR rtl-optimization/82396
1134 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
1135 (autopref_multipass_init): Simplify initialization.
1136 (autopref_rank_data): Simplify sort order.
1137 * gcc/sched-int.h (autopref_multipass_data_): Remove
1138 multi_mem_insn_p, min_offset and max_offset.
1140 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1143 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1144 Check special value of flag_omit_frame_pointer.
1145 (aarch64_can_eliminate): Likewise.
1146 (aarch64_override_options_after_change_1): Simplify handling of
1147 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
1149 2017-10-24 Richard Biener <rguenther@suse.de>
1151 PR tree-optimization/82697
1152 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
1153 zero for conditional load and unconditional store.
1155 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1157 * doc/install.texi: Document bootstrap-cet.
1159 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1162 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
1163 ENDBR instruction at function entrance if function is only
1166 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1169 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
1170 patterns to better describe from which operation the CF is computed.
1171 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
1172 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
1173 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
1174 is 0, use _0 suffixed expanders instead of emitting a comparison
1177 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
1179 * config/i386/i386.md(*movsf_internal, *movdf_internal):
1180 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
1182 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
1185 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
1186 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
1187 * loop-iv.c (iv_get_reaching_def): Likewise.
1188 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
1189 variable is promoted and the partition contains undefined values.
1191 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1193 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
1195 (nios2_address_cost): Define.
1196 (nios2_legitimize_address): Recognize (exp + constant) directly.
1197 (TARGET_ADDRESS_COST): Define.
1199 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1201 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
1202 (nios2_symbolic_memory_operand_p): Declare.
1203 (nios2_split_large_constant): Declare.
1204 (nios2_split_symbolic_memory_operand): Declare.
1205 * config/nios2/nios2.c: Adjust includes.
1206 (nios2_symbolic_constant_allowed): New.
1207 (nios2_symbolic_constant_p): New.
1208 (nios2_plus_symbolic_constant_p): New.
1209 (nios2_valid_addr_expr_p): Recognize addresses involving
1211 (nios2_legitimate_address_p): Likewise, also LO_SUM.
1212 (nios2_symbolic_memory_operand_p): New.
1213 (nios2_large_constant_p): New.
1214 (nios2_split_large_constant): New.
1215 (nios2_split_plus_large_constant): New.
1216 (nios2_split_symbolic_memory_operand): New.
1217 (nios2_legitimize_address): Code refactoring. Handle addresses
1218 involving symbolic constants.
1219 (nios2_emit_move_sequence): Likewise.
1220 (nios2_print_operand): Improve error output.
1221 (nios2_print_operand_address): Handle LO_SUM.
1222 (nios2_cdx_narrow_form_p): Likewise.
1223 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
1224 operands involving symbolic constants.
1225 (movhi_internal, movsi_internal): Likewise.
1226 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
1227 (extendhisi2, extendqi<mode>2): Likewise.
1229 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1231 * tree-pass.h (PROP_rtl_split_insns): Define.
1232 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
1234 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1236 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
1238 2017-10-23 Jakub Jelinek <jakub@redhat.com>
1241 * target.def (const_not_ok_for_debug_p): Default to
1242 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
1243 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
1244 * targhooks.c (default_const_not_ok_for_debug_p): New function.
1245 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
1246 which targetm.const_not_ok_for_debug_p returned true.
1247 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
1249 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
1251 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
1252 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
1253 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
1255 (ix86_const_not_ok_for_debug_p): New function.
1256 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
1257 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
1259 2017-10-23 David Malcolm <dmalcolm@redhat.com>
1262 * system.h: Conditionally include "unique-ptr.h" if
1263 INCLUDE_UNIQUE_PTR is defined.
1264 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
1265 of defining INCLUDE_UNIQUE_PTR before including "system.h".
1267 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
1269 * config/rl78/rl78.md: New define_expand "subdi3".
1271 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
1274 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
1275 DF_REF_INSN if DF_REF_INSN_INFO is false.
1277 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
1279 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
1280 xmm_move instead of sse_move.
1281 (sse_store_index): New function.
1282 (ix86_register_move_cost): Be more sensible about mismatch stall;
1283 model AVX moves correctly; make difference between sse->integer and
1285 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
1286 moves; make difference between SSE and AVX.
1287 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
1288 and zmm_move. Increase size of sse load and store tables;
1289 add unaligned load and store tables; add ssemmx_to_integer.
1290 * x86-tune-costs.h: Update all entries according to real
1291 move latencies from Agner Fog's manual and chip documentation.
1293 2017-10-23 Jakub Jelinek <jakub@redhat.com>
1296 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
1297 * config/i386/constraints.md (Wf): New constraint.
1298 * config/i386/i386.md (UNSPEC_SBB): New unspec.
1299 (cmp<dwi>_doubleword): Removed.
1300 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
1301 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
1302 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
1303 expand with cmp<dwi>_doubleword. For LTU and GEU use
1304 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
1306 * common.opt (gcolumn-info): Enable by default.
1307 * doc/invoke.texi (gcolumn-info): Document new default.
1309 2017-10-23 Richard Biener <rguenther@suse.de>
1311 PR tree-optimization/82672
1312 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
1313 Fold the stmt if we propagated into it.
1315 2017-10-23 Richard Biener <rguenther@suse.de>
1317 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
1318 (bitmap_remove_expr_from_set): ... this. All callers call this
1319 for non-constant values.
1320 (bitmap_set_subtract): Rename to...
1321 (bitmap_set_subtract_expressions): ... this. Adjust and
1323 (bitmap_set_contains_value): Remove superfluous check.
1324 (bitmap_set_replace_value): Inline into single caller ...
1325 (bitmap_value_replace_in_set): ... here and simplify.
1326 (dependent_clean): Merge into ...
1327 (clean): ... this using an overload. Adjust.
1328 (prune_clobbered_mems): Adjust.
1329 (compute_antic_aux): Likewise.
1330 (compute_partial_antic_aux): Likewise.
1332 2017-10-23 Richard Biener <rguenther@suse.de>
1334 PR tree-optimization/82129
1336 2017-08-01 Richard Biener <rguenther@suse.de>
1338 PR tree-optimization/81181
1339 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1340 (compute_antic): ... end of iteration here.
1342 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1344 * target.def (starting_frame_offset): New hook.
1345 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
1346 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
1347 * doc/tm.texi.in: Regenerate.
1348 * hooks.h (hook_hwi_void_0): Declare.
1349 * hooks.c (hook_hwi_void_0): New function.
1350 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
1351 STARTING_FRAME_OFFSET.
1352 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
1353 * reload1.c (reload): Likewise.
1354 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
1355 instead of STARTING_FRAME_OFFSET.
1356 * function.c (try_fit_stack_local): Likewise.
1357 (assign_stack_local_1): Likewise
1358 (instantiate_virtual_regs): Likewise.
1359 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
1360 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
1361 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
1362 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
1363 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
1364 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
1365 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
1366 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
1367 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
1368 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
1369 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
1370 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
1371 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
1372 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
1373 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
1374 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
1375 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
1376 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
1377 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
1378 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
1379 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
1380 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
1381 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
1382 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
1383 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
1384 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
1385 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
1386 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
1387 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
1388 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
1389 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
1390 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
1391 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
1392 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
1393 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
1394 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
1395 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
1396 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
1397 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
1398 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
1399 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
1400 * config/avr/avr.c (avr_starting_frame_offset): Make static and
1401 return a HOST_WIDE_INT.
1402 (avr_builtin_setjmp_frame_value): Use it instead of
1403 STARTING_FRAME_OFFSET.
1404 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1405 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
1406 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
1408 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1409 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
1410 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
1411 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1412 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
1413 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
1414 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1415 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
1416 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
1417 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1418 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
1419 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
1421 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1422 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
1423 * config/mips/mips.c (mips_compute_frame_info): Refer to
1424 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
1425 (mips_starting_frame_offset): New function.
1426 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1427 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
1428 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
1429 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
1430 and return a HOST_WIDE_INT.
1431 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1432 (mmix_initial_elimination_offset): Refer to
1433 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
1434 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
1435 * config/pa/pa.c (pa_starting_frame_offset): New function.
1436 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
1437 (pa_expand_prologue): Likewise.
1438 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1439 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
1440 !FRAME_GROWS_DOWNWARD handling to...
1441 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1442 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
1443 !FRAME_GROWS_DOWNWARD handling to...
1444 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1445 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
1446 !FRAME_GROWS_DOWNWARD handling to...
1447 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1448 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
1450 (rs6000_starting_frame_offset): New function.
1451 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
1452 !FRAME_GROWS_DOWNWARD handling to...
1453 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1454 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
1455 !FRAME_GROWS_DOWNWARD handling to...
1456 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1457 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
1458 !FRAME_GROWS_DOWNWARD handling to...
1459 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1460 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
1461 (rs6000_starting_frame_offset): New function.
1462 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
1463 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
1464 * config/vax/vax.c (vax_starting_frame_offset): New function.
1465 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
1466 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1467 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
1468 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
1469 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1470 * system.h (STARTING_FRAME_OFFSET): Poison.
1472 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1474 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
1475 SCALAR_TYPE_MODE instead of TYPE_MODE.
1477 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1478 Alan Hayward <alan.hayward@arm.com>
1479 David Sherwood <david.sherwood@arm.com>
1481 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
1483 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1484 Alan Hayward <alan.hayward@arm.com>
1485 David Sherwood <david.sherwood@arm.com>
1487 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
1489 2017-10-23 Richard Biener <rguenther@suse.de>
1491 PR tree-optimization/82129
1492 * tree-ssa-pre.c (bitmap_set_and): Remove.
1493 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
1494 canonicalizing expressions in the set to those with lowest
1495 ID rather than taking that from the first edge.
1497 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1499 * combine.c (rtx_equal_for_field_assignment_p): Use
1500 byte_lowpart_offset.
1502 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1503 Alan Hayward <alan.hayward@arm.com>
1504 David Sherwood <david.sherwood@arm.com>
1506 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
1507 to a SUBREG_PROMOTED_VAR.
1509 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1510 Alan Hayward <alan.hayward@arm.com>
1511 David Sherwood <david.sherwood@arm.com>
1513 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
1514 (expand_debug_source_expr): Likewise.
1515 * combine.c (combine_simplify_rtx): Likewise.
1516 * cse.c (fold_rtx): Likewise.
1517 * optabs.c (expand_float): Likewise.
1518 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1519 (simplify_binary_operation_1): Likewise.
1521 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1522 Alan Hayward <alan.hayward@arm.com>
1523 David Sherwood <david.sherwood@arm.com>
1525 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
1526 (record_promoted_value): Likewise.
1527 * expr.c (expand_expr_real_2): Likewise.
1528 * ree.c (update_reg_equal_equiv_notes): Likewise.
1529 (combine_set_extension): Likewise.
1530 * rtlanal.c (low_bitmask_len): Likewise.
1531 * simplify-rtx.c (neg_const_int): Likewise.
1532 (simplify_binary_operation_1): Likewise.
1534 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1535 Alan Hayward <alan.hayward@arm.com>
1536 David Sherwood <david.sherwood@arm.com>
1538 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
1539 * regcprop.c (maybe_mode_change): Likewise.
1540 * reload1.c (alter_reg): Likewise.
1542 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1544 * inchash.h (inchash::hash::add_wide_int): New function.
1545 * lto-streamer-out.c (hash_tree): Use it.
1547 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1549 * inchash.h (inchash::hash::add_wide_int): Rename to...
1550 (inchash::hash::add_hwi): ...this.
1551 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
1552 (polymorphic_call_target_hasher::hash): Likewise.
1553 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
1554 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
1555 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
1556 * lto-streamer-out.c (hash_tree): Likewise.
1557 * optc-save-gen.awk: Likewise.
1558 * tree.c (add_expr): Likewise.
1560 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1563 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
1564 for ordered inequality comparisons even with TARGET_IEEE_FP.
1566 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1569 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
1570 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1571 Expand with cmp<dwi>_doubleword.
1573 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1575 * extend.texi: Add x86 specific to 'nocf_check' attribute.
1576 List CET intrinsics.
1577 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
1578 specific to -fcf-protection option.
1580 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1582 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
1583 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
1584 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
1585 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
1586 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
1587 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
1588 (extra_objs): Add cet.o for Linux/x86 targets.
1589 (tmake_file): Add i386/t-cet for Linux/x86 targets.
1590 * config/i386/cet.c: New file.
1591 * config/i386/cetintrin.h: Likewise.
1592 * config/i386/t-cet: Likewise.
1593 * config/i386/cpuid.h (bit_SHSTK): New.
1594 (bit_IBT): Likewise.
1595 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
1596 pass IBT and SHSTK bits.
1597 * config/i386/i386-builtin-types.def
1598 (VOID_FTYPE_UNSIGNED_PVOID): New.
1599 (VOID_FTYPE_UINT64_PVOID): Likewise.
1600 * config/i386/i386-builtin.def: Add CET intrinsics.
1601 * config/i386/i386-c.c (ix86_target_macros_internal): Add
1602 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
1603 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
1604 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
1606 * config/i386/i386.c (rest_of_insert_endbranch): New.
1607 (pass_data_insert_endbranch): Likewise.
1608 (pass_insert_endbranch): Likewise.
1609 (make_pass_insert_endbranch): Likewise.
1610 (ix86_notrack_prefixed_insn_p): Likewise.
1611 (ix86_target_string): Add -mibt, -mshstk flags.
1612 (ix86_option_override_internal): Add flag_cf_protection
1614 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
1615 (ix86_print_operand): Add 'notrack' prefix output.
1616 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
1617 (ix86_expand_builtin): Expand CET intrinsics.
1618 (x86_output_mi_thunk): Add 'endbranch' instruction.
1619 * config/i386/i386.h (TARGET_IBT): New.
1620 (TARGET_IBT_P): Likewise.
1621 (TARGET_SHSTK): Likewise.
1622 (TARGET_SHSTK_P): Likewise.
1623 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
1624 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
1625 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
1626 (builtin_setjmp_setup): New pattern.
1627 (builtin_longjmp): Likewise.
1628 (rdssp<mode>): Likewise.
1629 (incssp<mode>): Likewise.
1630 (saveprevssp): Likewise.
1631 (rstorssp): Likewise.
1632 (wrss<mode>): Likewise.
1633 (wruss<mode>): Likewise.
1634 (setssbsy): Likewise.
1635 (clrssbsy): Likewise.
1636 (nop_endbr): Likewise.
1637 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
1639 * config/i386/immintrin.h: Include <cetintrin.h>.
1640 * config/i386/linux-common.h
1641 (file_end_indicate_exec_stack_and_cet): New prototype.
1642 (TARGET_ASM_FILE_END): New.
1644 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
1646 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
1647 latencies instead of having separate table; make difference between
1648 integer and float costs.
1649 * i386.h (processor_costs): Remove scalar_stmt_cost,
1650 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
1651 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
1653 * x86-tune-costs.h: Remove entries which has been removed in
1654 procesor_costs from all tables; make cond_taken_branch_cost
1655 and cond_not_taken_branch_cost COST_N_INSNS based.
1657 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
1659 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
1661 2017-10-20 Jakub Jelinek <jakub@redhat.com>
1663 * config/i386/i386.md (isa): Remove fma_avx512f.
1664 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
1665 <avx512>_fmadd_<mode>_mask3<round_name>,
1666 <avx512>_fmsub_<mode>_mask<round_name>,
1667 <avx512>_fmsub_<mode>_mask3<round_name>,
1668 <avx512>_fnmadd_<mode>_mask<round_name>,
1669 <avx512>_fnmadd_<mode>_mask3<round_name>,
1670 <avx512>_fnmsub_<mode>_mask<round_name>,
1671 <avx512>_fnmsub_<mode>_mask3<round_name>,
1672 <avx512>_fmaddsub_<mode>_mask<round_name>,
1673 <avx512>_fmaddsub_<mode>_mask3<round_name>,
1674 <avx512>_fmsubadd_<mode>_mask<round_name>,
1675 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
1676 (*vec_widen_umult_even_v16si<mask_name>,
1677 *vec_widen_smult_even_v16si<mask_name>): Likewise.
1678 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
1680 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1682 * extend.texi: Add 'nocf_check' documentation.
1683 * gimple.texi: Add second parameter to
1684 gimple_build_call_from_tree.
1685 * invoke.texi: Add -fcf-protection documentation.
1686 * rtl.texi: Add REG_CALL_NOTRACK documenation.
1688 2017-10-20 Richard Biener <rguenther@suse.de>
1690 PR tree-optimization/82473
1691 * tree-vect-loop.c (vectorizable_reduction): Properly get at
1692 the largest input type.
1694 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1696 * c-attribs.c (handle_nocf_check_attribute): New function.
1697 (c_common_attribute_table): Add 'nocf_check' handling.
1698 * gimple-parser.c: Add second argument NULL to
1699 gimple_build_call_from_tree.
1700 * attrib.c (comp_type_attributes): Check nocf_check attribute.
1701 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
1703 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
1704 * common.opt: Add fcf-protection flag.
1705 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
1706 * flag-types.h: Add enum cf_protection_level.
1707 * gimple.c (gimple_build_call_from_tree): Add second parameter.
1708 Add 'nocf_check' attribute propagation to gimple call.
1709 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
1710 (gimple_build_call_from_tree): Update prototype.
1711 (gimple_call_nocf_check_p): New function.
1712 (gimple_call_set_nocf_check): Likewise.
1713 * gimplify.c: Add second argument to gimple_build_call_from_tree.
1714 * ipa-icf.c: Add nocf_check attribute in statement hash.
1715 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
1716 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
1717 * toplev.c (process_options): Add flag_cf_protection handling.
1719 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1721 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
1723 2017-10-20 Richard Biener <rguenther@suse.de>
1725 PR tree-optimization/82603
1726 * tree-if-conv.c (predicate_mem_writes): Make sure to only
1727 remove false predicated stores.
1729 2017-10-20 Richard Biener <rguenther@suse.de>
1731 * graphite-isl-ast-to-gimple.c
1732 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
1733 Remove return value and simplify, dump copied stmt after lhs
1735 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
1736 Reduce dump verbosity.
1737 (gsi_insert_earliest): Likewise.
1738 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
1739 * graphite.c (print_global_statistics): Adjust dumping.
1740 (print_graphite_scop_statistics): Likewise.
1741 (print_graphite_statistics): Do not dump loops here.
1742 (graphite_transform_loops): But here.
1744 2017-10-20 Nicolas Roche <roche@adacore.com>
1746 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
1747 * configure: Regenerate.
1749 2017-10-20 Jakub Jelinek <jakub@redhat.com>
1752 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
1753 functions when optimizing replace GIMPLE_RETURN stmts with
1754 calls to __builtin_unreachable ().
1757 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
1758 for -fsanitize=thread link of executables.
1759 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
1760 link of executables.
1763 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
1765 (<shift_insn><mode>3<mask_name>): Change the last of the 3
1766 define_insns for logical vector shifts to use VI248_AVX512BW
1767 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
1768 condition, useless isa and prefix attributes. Change the first
1769 2 of these define_insns to ...
1770 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
1771 define_insn for avx512vl.
1772 (<shift_insn><mode>3): ... and this, new define_insn without
1773 masking for non-avx512vl.
1776 * config/i386/sse.md (*andnot<mode>3,
1777 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
1778 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
1779 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
1780 not applied use empty suffix even for TARGET_AVX512VL.
1781 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
1782 is applied, supply evex,evex or evex,evex,evex instead of just
1785 2017-10-20 Julia Koval <julia.koval@intel.com>
1787 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
1788 (OPTION_MASK_ISA_GFNI_UNSET): New.
1789 (ix86_handle_option): Handle OPT_mgfni.
1790 * config/i386/cpuid.h (bit_GFNI): New.
1791 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
1792 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
1793 * config/i386/i386.c (ix86_target_string): Add -mgfni.
1794 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
1795 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
1796 * config/i386/i386.opt: Add mgfni.
1798 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
1800 * config/msp430/msp430.c (msp430_option_override): Disable
1801 -fdelete-null-pointer-checks.
1802 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
1804 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1806 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
1807 of x87 and SSE instructions.
1809 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1811 * asan.c (create_cond_insert_point): Do not update edge count.
1812 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
1813 (afdo_propagate_circuit): Likewise.
1814 (afdo_calculate_branch_prob): Likewise.
1815 (afdo_annotate_cfg): Likewise.
1816 * basic-block.h (struct edge_def): Remove count.
1817 (edge_def::count): New accessor.
1818 * bb-reorder.c (rotate_loop): Update.
1819 (find_traces_1_round): Update.
1820 (connect_traces): Update.
1821 (sanitize_hot_paths): Update.
1822 * cfg.c (unchecked_make_edge): Update.
1823 (make_single_succ_edge): Update.
1824 (check_bb_profile): Update.
1825 (dump_edge_info): Update.
1826 (update_bb_profile_for_threading): Update.
1827 (scale_bbs_frequencies_int): Update.
1828 (scale_bbs_frequencies_gcov_type): Update.
1829 (scale_bbs_frequencies_profile_count): Update.
1830 (scale_bbs_frequencies): Update.
1831 * cfganal.c (connect_infinite_loops_to_exit): Update.
1832 * cfgbuild.c (compute_outgoing_frequencies): Update.
1833 (find_many_sub_basic_blocks): Update.
1834 * cfgcleanup.c (try_forward_edges): Update.
1835 (try_crossjump_to_edge): Update
1836 * cfgexpand.c (expand_gimple_cond): Update
1837 (expand_gimple_tailcall): Update
1838 (construct_exit_block): Update
1839 * cfghooks.c (verify_flow_info): Update
1840 (redirect_edge_succ_nodup): Update
1841 (split_edge): Update
1842 (make_forwarder_block): Update
1843 (duplicate_block): Update
1844 (account_profile_record): Update
1845 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
1846 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
1847 * cfgloopmanip.c (scale_loop_profile): Update.
1849 (lv_adjust_loop_entry_edge): Update.
1850 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
1851 (force_nonfallthru_and_redirect): Update.
1852 (purge_dead_edges): Update.
1853 (rtl_flow_call_edges_add): Update.
1854 * cgraphunit.c (init_lowered_empty_function): Update.
1855 (cgraph_node::expand_thunk): Update.
1856 * gimple-pretty-print.c (dump_probability): Update.
1857 (dump_edge_probability): Update.
1858 * gimple-ssa-isolate-paths.c (isolate_path): Update.
1859 * haifa-sched.c (sched_create_recovery_edges): Update.
1860 * hsa-gen.c (convert_switch_statements): Update.
1861 * ifcvt.c (dead_or_predicable): Update.
1862 * ipa-inline-transform.c (inline_transform): Update.
1863 * ipa-split.c (split_function): Update.
1864 * ipa-utils.c (ipa_merge_profiles): Update.
1865 * loop-doloop.c (add_test): Update.
1866 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
1867 * lto-streamer-in.c (input_cfg): Update.
1868 (input_function): Update.
1869 * lto-streamer-out.c (output_cfg): Update.
1870 * modulo-sched.c (sms_schedule): Update.
1871 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
1872 * predict.c (maybe_hot_edge_p): Update.
1873 (unlikely_executed_edge_p): Update.
1874 (probably_never_executed_edge_p): Update.
1875 (dump_prediction): Update.
1876 (drop_profile): Update.
1877 (propagate_unlikely_bbs_forward): Update.
1878 (determine_unlikely_bbs): Update.
1879 (force_edge_cold): Update.
1880 * profile.c (compute_branch_probabilities): Update.
1881 * reg-stack.c (better_edge): Update.
1882 * shrink-wrap.c (handle_simple_exit): Update.
1883 * tracer.c (better_p): Update.
1884 * trans-mem.c (expand_transaction): Update.
1885 (split_bb_make_tm_edge): Update.
1886 * tree-call-cdce.c: Update.
1887 * tree-cfg.c (gimple_find_sub_bbs): Update.
1888 (gimple_split_edge): Update.
1889 (gimple_duplicate_sese_region): Update.
1890 (gimple_duplicate_sese_tail): Update.
1891 (gimple_flow_call_edges_add): Update.
1892 (insert_cond_bb): Update.
1893 (execute_fixup_cfg): Update.
1894 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
1895 * tree-complex.c (expand_complex_div_wide): Update.
1896 * tree-eh.c (lower_resx): Update.
1897 (unsplit_eh): Update.
1898 (cleanup_empty_eh_move_lp): Update.
1899 * tree-inline.c (copy_edges_for_bb): Update.
1900 (freqs_to_counts): Update.
1901 (copy_cfg_body): Update.
1902 * tree-ssa-dce.c (remove_dead_stmt): Update.
1903 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
1904 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
1905 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
1906 (unloop_loops): Update.
1907 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
1908 * tree-ssa-loop-split.c (connect_loops): Update.
1909 (split_loop): Update.
1910 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
1911 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
1912 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
1913 * tree-ssa-reassoc.c (branch_fixup): Update.
1914 * tree-ssa-tail-merge.c (replace_block_by): Update.
1915 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
1916 (compute_path_counts): Update.
1917 (update_profile): Update.
1918 (recompute_probabilities): Update.
1919 (update_joiner_offpath_counts): Update.
1920 (estimated_freqs_path): Update.
1921 (freqs_to_counts_path): Update.
1922 (clear_counts_path): Update.
1923 (ssa_fix_duplicate_block_edges): Update.
1924 (duplicate_thread_path): Update.
1925 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
1926 (case_bit_test_cmp): Update.
1927 (collect_switch_conv_info): Update.
1928 (gen_inbound_check): Update.
1929 (do_jump_if_equal): Update.
1930 (emit_cmp_and_jump_insns): Update.
1931 * tree-tailcall.c (decrease_profile): Update.
1932 (eliminate_tail_call): Update.
1933 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
1934 (vect_do_peeling): Update.
1935 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
1936 * ubsan.c (ubsan_expand_null_ifn): Update.
1937 (ubsan_expand_ptr_ifn): Update.
1938 * value-prof.c (gimple_divmod_fixed_value): Update.
1939 (gimple_mod_pow2): Update.
1940 (gimple_mod_subtract): Update.
1941 (gimple_ic): Update.
1942 (gimple_stringop_fixed_value): Update.
1944 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1947 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
1949 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
1951 PR rtl-optimization/82395
1952 * ira-color.c (allocno_priority_compare_func): Fix comparison step
1953 based on non_spilled_static_chain_regno_p.
1955 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1957 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
1958 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
1959 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
1961 2017-10-19 Martin Sebor <msebor@redhat.com>
1963 PR tree-optimization/82596
1964 * tree.c (array_at_struct_end_p): Handle STRING_CST.
1966 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1968 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
1969 (get_mem_refs_of_builtin_call): Likewise.
1970 * builtins.c (expand_builtin_apply): Adjust call to
1971 allocate_dynamic_stack_space.
1972 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
1973 the third argument to allocate_dynamic_stack_space, otherwise -1.
1974 (expand_builtin): Deal with all alloca variants.
1975 (is_inexpensive_builtin): Likewise.
1976 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
1977 * calls.c (special_function_p): Deal with all alloca variants.
1978 (initialize_argument_information): Adjust call to
1979 allocate_dynamic_stack_space.
1980 (expand_call): Likewise.
1981 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
1982 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
1983 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
1984 use it for the stack usage computation.
1985 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
1986 * function.c (gimplify_parameters): Call build_alloca_call_expr.
1987 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
1988 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
1989 (in_loop_p): Remove first argument and useless check.
1990 (pass_walloca::execute): Remove useless test and adjust call to above.
1991 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
1992 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
1993 (gimplify_call_expr): Deal with all alloca variants.
1994 * hsa-gen.c (gen_hsa_alloca): Likewise.
1995 (gen_hsa_insns_for_call): Likewise.
1996 * ipa-pure-const.c (special_builtin_state): Likewise.
1997 * tree-chkp.c (chkp_build_returned_bound): Likewise.
1998 * tree-object-size.c (alloc_object_size): Likewise.
1999 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2000 (call_may_clobber_ref_p_1): Likewise.
2001 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
2002 (ccp_fold_stmt): Likewise.
2003 (optimize_stack_restore): Likewise.
2004 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
2005 (mark_all_reaching_defs_necessary_1): Likewise.
2006 (propagate_necessity): Likewise.
2007 (eliminate_unnecessary_stmts): Likewise.
2008 * tree.c (build_common_builtin_nodes): Build
2009 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
2010 (build_alloca_call_expr): New function.
2011 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
2012 (CASE_BUILT_IN_ALLOCA): Likewise.
2013 (build_alloca_call_expr): Declare.
2014 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
2016 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2019 * dwarf2out.c (new_die_raw): New static inline function.
2020 (new_die): Use it to create the DIE.
2021 (add_AT_external_die_ref): Likewise.
2022 (clone_die): Likewise.
2023 (clone_as_declaration): Likewise.
2024 (dwarf2out_vms_debug_main_pointer): Likewise.
2025 (base_type_die): Likewise. Remove early return for corner cases.
2026 Do not call add_pubtype on the DIE here.
2027 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
2028 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
2029 typedefs for base types with DW_AT_endianity. Make sure a DIE with
2030 native order exists for base types, attach the DIE manually and call
2031 add_pubtype on it. Do not equate a reverse order DIE to the type.
2033 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
2035 * config/arm/arm.c (align_ok_ldrd_strd): New function.
2036 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
2038 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
2040 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2042 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
2043 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
2044 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
2045 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
2046 * opts.c (sanitizer_opts): Add builtin.
2047 * ubsan.c (instrument_builtin): New function.
2048 (pass_ubsan::execute): Call it.
2049 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
2050 * doc/invoke.texi: Document -fsanitize=builtin.
2052 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
2053 builtins, store max (log2 (align), 0) into uchar field instead of
2054 align into uptr field.
2055 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
2056 store uchar 0 field instead of uptr 0 field.
2057 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
2058 instead of passing one address of struct with 2 locations pass
2059 two addresses of structs with 1 location each.
2060 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
2061 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
2062 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
2063 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
2064 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
2065 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
2066 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
2067 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
2069 2017-10-19 Martin Liska <mliska@suse.cz>
2072 * file-find.c (remove_prefix): Remove.
2073 * file-find.h (remove_prefix): Likewise.
2074 * gcc-ar.c: Remove smartness of lookup.
2076 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
2078 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
2079 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
2080 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
2082 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2085 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
2086 (setcc + and to xor + setcc): New peephole2.
2088 2017-10-19 Tom de Vries <tom@codesourcery.com>
2090 * doc/sourcebuild.texi (Test Directives, Variants of
2091 dg-require-support): Add dg-require-stack-size.
2093 2017-10-19 Martin Liska <mliska@suse.cz>
2096 * gimplify.c (gimplify_decl_expr): Do not instrument variables
2097 that have a large alignment.
2098 (gimplify_target_expr): Likewise.
2100 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
2102 PR rtl-optimization/82602
2103 * ira.c (rtx_moveable_p): Return false for volatile asm.
2105 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
2108 * config/i386/i386-modes.def (CCGZ): New CC mode.
2109 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
2110 * config/i386/predicates.md (ix86_comparison_operator):
2112 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2113 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
2114 with double-word subtraction.
2115 (put_condition_code): Handle CCGZmode.
2117 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
2119 * wide-int.cc (debug (const wide_int &)): New.
2120 (debug (const wide_int *)): New.
2121 (debug (const widest_int &)): New.
2122 (debug (const widest_int *)): New.
2124 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
2127 * lra-constraints.c (curr_insn_transform): Use non-input operand
2128 instead of output one for matched reload.
2130 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2132 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
2133 (tree-ssa-loop-ivopts.h): New header file.
2134 (struct builtin_info): New fields.
2135 (classify_builtin_1): Compute and record base and offset parts for
2136 memset builtin partition by calling strip_offset.
2137 (offset_cmp, fuse_memset_builtins): New functions.
2138 (finalize_partitions): Fuse adjacent memset partitions by calling
2140 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
2141 Expose the interface.
2142 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
2144 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2146 PR tree-optimization/82574
2147 * tree-loop-distribution.c (find_single_drs): New parameter. Check
2148 that data reference must be executed exactly once per iteration
2149 against the outermost loop in nest.
2150 (classify_partition): Update call to above function.
2152 2017-10-18 Richard Biener <rguenther@suse.de>
2154 PR tree-optimization/82591
2155 * graphite.c (graphite_transform_loops): Move code gen message
2157 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2158 Here. Handle scop_to_isl_ast failing.
2159 (scop_to_isl_ast): Limit the number of ISL operations.
2161 2017-10-18 Richard Biener <rguenther@suse.de>
2163 * graphite-isl-ast-to-gimple.c
2164 (translate_isl_ast_to_gimple::set_rename): Simplify.
2165 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
2166 (graphite_copy_stmts_from_block): ... here.
2167 (copy_bb_and_scalar_dependences): Simplify.
2168 (add_parameters_to_ivs_params): Canonicalize.
2169 (generate_entry_out_of_ssa_copies): Simplify.
2170 * graphite-sese-to-poly.c (extract_affine_name): Simplify
2171 by passing in ISL dimension.
2172 (parameter_index_in_region_1): Rename to ...
2173 (parameter_index_in_region): ... this.
2174 (extract_affine): Adjust assert, pass down parameter index.
2175 (add_param_constraints): Use range-info when available.
2176 (build_scop_context): Adjust.
2177 * sese.c (new_sese_info): Adjust.
2178 (free_sese_info): Likewise.
2179 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
2180 Remove unused typedefs.
2181 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
2183 2017-10-18 Martin Liska <mliska@suse.cz>
2185 * combine.c (simplify_compare_const): Add gcc_fallthrough.
2187 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2189 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
2190 (s390_sched_init): Do not reset s390_sched_state if we entered the
2191 current basic block via a fallthru edge and all others are unlikely.
2193 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2195 * config/s390/s390.c (NUM_SIDES): New variable.
2196 (LONGRUNNING_THRESHOLD): New variable.
2197 (LATENCY_FACTOR): New variable.
2198 (s390_sched_score): Decrease score for long-running instructions on
2200 (s390_sched_variable_issue): Perform bookkeeping for long-running
2203 2017-10-18 Richard Biener <rguenther@suse.de>
2205 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
2206 Simplify with removal of the parameter rename map.
2207 (set_rename): Likewise.
2208 (should_copy_to_new_region): Likewise.
2209 (graphite_copy_stmts_from_block): Likewise.
2210 (copy_bb_and_scalar_dependences): Remove initialization of
2211 unused copied_bb_map.
2213 (copy_internal_parameters): Likewise.
2214 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
2215 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
2216 Use INTEGRAL_TYPE_P.
2217 (parameter_index_in_region_1): Rename to ...
2218 (assign_parameter_index_in_region): ... this. Assert we have
2219 a parameter we handle.
2220 (scan_tree_for_params): Adjust.
2221 * sese.h (parameter_rename_map_t): Remove.
2222 (struct sese_info_t): Remove unused parameter_rename_map and
2223 copied_bb_map members.
2224 * sese.c (new_sese_info): Adjust.
2225 (free_sese_info): Likewise.
2227 2017-10-18 Martin Liska <mliska@suse.cz>
2230 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
2231 on an abnormal edge.
2233 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
2235 * doc/invoke.texi (ffunction-sections and fdata-sections):
2238 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
2240 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
2241 the use statement can throw internally.
2243 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
2245 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
2246 any RTX present on the RHS of a SET.
2247 * compare-elim.c (try_eliminate_compare): Restore comment.
2249 2017-10-17 Jakub Jelinek <jakub@redhat.com>
2251 * langhooks.h (struct lang_hooks): Document that tree_size langhook
2252 may be also called on tcc_type nodes.
2253 * langhooks.c (lhd_tree_size): Likewise.
2255 2017-10-17 David Malcolm <dmalcolm@redhat.com>
2257 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
2258 format_warning_at_substring.
2259 (maybe_warn): Convert source_range * param to a location_t. Pass
2260 UNKNOWN_LOCATION rather than NULL to fmtwarn.
2261 (format_directive): Remove code to extract source_ranges and
2262 source_range * in favor of just a location_t.
2263 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
2265 * substring-locations.c (format_warning_va): Convert
2266 source_range * param to a location_t.
2267 (format_warning_at_substring): Likewise.
2268 * substring-locations.h (format_warning_va): Likewise.
2269 (format_warning_at_substring): Likewise.
2271 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
2273 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
2275 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
2276 and scatter/gather ops.
2278 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
2279 vec_gather_load and vec_scatter_store.
2280 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
2281 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
2282 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
2283 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
2284 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
2285 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
2287 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
2289 * reg-stack.c (compare_for_stack_reg): Add bool argument.
2290 Detect FTST instruction and handle its register pops. Only pop
2291 second operand if can_pop_second_op is true.
2292 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
2293 set can_pop_second_op to false in the compare_for_stack_reg call.
2295 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
2296 output_fp_compare for stack register operands.
2297 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
2298 instructions here. Do not emit stack register pops here. Assert
2299 that FCOMPP pops next to top stack register. Rewrite function.
2301 2017-10-17 Nathan Sidwell <nathan@acm.org>
2304 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
2305 use DECL_ASSEMBLER_NAME_RAW.
2308 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
2311 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
2312 Wilco Dijkstra <wilco.dijkstra@arm.com>
2314 * builtins.c (expand_builtin_update_setjmp_buf): Add a
2315 converstion to Pmode from the buf_addr.
2317 2017-10-17 Richard Biener <rguenther@suse.de>
2319 * graphite-dependences.c (scop_get_reads_and_writes): Change
2320 output parameters to references.
2322 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
2324 PR 71026/tree-optimization
2325 * fold-const.c (distribute_real_division): Removed.
2326 (fold_binary_loc): Remove calls to distribute_real_divison.
2328 2017-10-17 Richard Biener <rguenther@suse.de>
2330 * graphite-scop-detection.c
2331 (scop_detection::stmt_has_simple_data_refs_p): Always use
2332 the full nest as region.
2333 (try_generate_gimple_bb): Likewise.
2334 * sese.c (scalar_evolution_in_region): Simplify now that
2335 SCEV can handle instantiation in regions.
2336 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
2337 in the non-loop part of a function if requested.
2339 2017-10-17 Richard Biener <rguenther@suse.de>
2341 PR tree-optimization/82563
2342 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
2344 (graphite_regenerate_ast_isl): Call it.
2345 * graphite-scop-detection.c (build_scops): Remove entry edge split.
2347 2017-10-17 Jakub Jelinek <jakub@redhat.com>
2349 PR tree-optimization/82549
2350 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
2351 Formatting fixes. Instead of calling make_bit_field_ref with negative
2354 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
2356 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
2357 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
2358 _mm_maskz_reduce_ss): New.
2359 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
2360 __builtin_ia32_reducess_mask): Ditto..
2361 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
2362 * config/i386/sse.md (reduces<mode>): Renamed to ...
2363 (reduces<mode><mask_scalar_name>): ... this.
2364 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
2366 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
2367 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
2369 2017-10-16 David Malcolm <dmalcolm@redhat.com>
2371 * Makefile.in (OBJS): Add unique-ptr-tests.o.
2372 * selftest-run-tests.c (selftest::run_tests): Call
2373 selftest::unique_ptr_tests_cc_tests.
2374 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
2375 * unique-ptr-tests.cc: New file.
2377 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
2380 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2382 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2383 (make_hard_regno_born, make_hard_regno_dead): Update
2384 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
2386 2017-10-16 Jeff Law <law@redhat.com>
2388 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
2390 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
2392 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
2394 2017-10-16 Olivier Hainque <hainque@adacore.com>
2396 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
2397 with_cpu if we were configured for an e500v2 target cpu name.
2399 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
2401 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
2402 * doc/invoke.texi: Document +nodsp as a valid extension for
2405 2017-10-16 Martin Liska <mliska@suse.cz>
2407 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
2408 (test_set_range): Likewise.
2409 (test_range_functions): Rename to ...
2410 (test_bit_in_range): ... this.
2411 (sbitmap_c_tests): Add new test.
2413 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2415 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
2417 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
2418 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
2420 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2422 * config/aarch64/aarch64-builtins.c
2423 (aarch64_types_quadopu_lane_qualifiers): New.
2424 (TYPES_QUADOPU_LANE): New.
2425 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
2426 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
2427 (aarch64_<sur>dot_laneq<vsi2qi>): New.
2428 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
2429 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
2430 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
2431 (Vdottype, DOTPROD): New.
2432 (sur): Add SDOT and UDOT.
2434 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2436 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
2437 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
2438 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
2440 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
2441 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
2442 Enable TARGET_DOTPROD.
2443 (cortex-a75.cortex-a55): Likewise.
2444 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
2446 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2448 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
2449 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
2451 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
2453 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
2454 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
2455 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
2456 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
2457 * config/arm/types.md (neon_dot, neon_dot_q): New.
2458 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
2460 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2462 * config/arm/arm.h (TARGET_DOTPROD): New.
2463 * config/arm/arm.c (arm_arch_dotprod): New.
2464 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
2465 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
2466 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
2467 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
2468 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
2469 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
2470 * doc/invoke.texi (armv8.2-a): Document dotprod
2472 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
2474 * i386.c (ix86_vec_cost): New function.
2475 (ix86_rtx_costs): Handle vector operations better.
2476 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
2477 * x86-tune-costs.h: Add new costs to all tables.
2479 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
2481 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
2483 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
2484 divsd, sqrtss and sqrtsd
2485 * x86-tune-costs.h: Add new entries to all costs.
2486 (znver1_cost): Fix to match real instruction latencies.
2488 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2489 Michael Collison <michael.collison@arm.com>
2491 * compare-elim.c: Include emit-rtl.h.
2492 (can_merge_compare_into_arith): New function.
2493 (try_validate_parallel): Likewise.
2494 (try_merge_compare): Likewise.
2495 (try_eliminate_compare): Call the above when no previous clobber
2497 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
2500 2017-10-14 Jakub Jelinek <jakub@redhat.com>
2504 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
2505 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
2509 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
2510 to be any operand_equal_p operands. For & (B - 1) require
2511 B to be power of 2. Recognize
2512 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
2514 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2517 * optabs.c (expand_memory_blockage): Fix call of
2518 targetm.have_memory_blockage.
2520 2017-10-14 Jakub Jelinek <jakub@redhat.com>
2523 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
2524 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
2525 objects to extra_objs instead of overwriting it.
2527 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2529 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
2530 Use any_fp_register_operand as operand[3] predicate. Simplify
2531 equality test for operands[2] and operands[4] memory location.
2532 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
2533 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
2534 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
2535 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
2536 any_fp_register_operand as operand[1] predicate. Simplify
2537 equality test for operands[0] and operands[3] memory location.
2538 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
2539 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
2540 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
2542 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2544 * target-insns.def: Add memory_blockage.
2545 * optabs.c (expand_memory_blockage): New function.
2546 (expand_asm_memory_barrier): Rename ...
2547 (expand_asm_memory_blockage): ... to this.
2548 (expand_mem_thread_fence): Call expand_memory_blockage
2549 instead of expand_asm_memory_barrier.
2550 (expand_mem_singnal_fence): Ditto.
2551 (expand_atomic_load): Ditto.
2552 (expand_atomic_store): Ditto.
2553 * doc/md.texi (Standard Pattern Names For Generation):
2554 Document memory_blockage instruction pattern.
2556 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
2558 * config/rl78/rl78.c (rl78_emit_libcall): New function.
2559 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
2560 * config/rl78/rl78.md: New define_expand "adddi3".
2562 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2564 * cfghooks.c (verify_flow_info): Disable check that all probabilities
2567 2017-10-13 Jeff Law <law@redhat.com>
2569 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
2571 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2574 * internal-fn.c (expand_mul_overflow): If both operands have
2575 the same highpart of -1 or 0 and the topmost bit of lowpart
2576 is different, overflow is if res <= 0 rather than res < 0.
2578 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
2580 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
2581 TARGET_P9_VECTOR code for unaligned_load case.
2583 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2585 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
2587 2017-10-13 Nathan Sidwell <nathan@acm.org>
2589 * tree-core.h (tree_contains_struct): Make bool.
2590 * tree.c (tree_contains_struct): Likewise.
2591 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
2592 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
2593 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
2594 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
2595 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
2597 2017-10-13 Richard Biener <rguenther@suse.de>
2599 * graphite-isl-ast-to-gimple.c
2600 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
2601 parameters and dominance check.
2602 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
2603 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
2604 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
2605 Do not update SSA form here or do intermediate IL verification.
2606 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
2607 (graphite_initialize): Remove check on the number of loops in
2608 the function and inline into graphite_transform_loops.
2609 (graphite_finalize): Inline into graphite_transform_loops.
2610 (graphite_transform_loops): Perform SSA update and IL verification
2612 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
2614 2017-10-13 Richard Biener <rguenther@suse.de>
2616 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
2617 graphite_expression_type_precision): Avoid global constructor
2619 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
2620 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
2621 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
2622 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
2623 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
2624 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
2626 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
2629 * config/i386/i386.h (ix86_red_zone_size): New.
2630 * config/i386/i386.md (push peephole2s): Replace
2631 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
2633 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2634 Alan Hayward <alan.hayward@arm.com>
2635 David Sherwood <david.sherwood@arm.com>
2637 * combine.c (can_change_dest_mode): Reject changes in
2638 REGMODE_NATURAL_SIZE.
2640 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2641 Alan Hayward <alan.hayward@arm.com>
2642 David Sherwood <david.sherwood@arm.com>
2644 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
2645 (expand_debug_source_expr): Likewise.
2646 * combine.c (combine_simplify_rtx): Likewise.
2647 * cse.c (fold_rtx): Likewise.
2648 * fwprop.c (canonicalize_address): Likewise.
2649 * targhooks.c (default_shift_truncation_mask): Likewise.
2651 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2652 Alan Hayward <alan.hayward@arm.com>
2653 David Sherwood <david.sherwood@arm.com>
2655 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
2656 (widened_mode): Likewise.
2657 (expand_unop): Likewise.
2658 * ree.c (transform_ifelse): Likewise.
2659 (merge_def_and_ext): Likewise.
2660 (combine_reaching_defs): Likewise.
2661 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2663 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2664 Alan Hayward <alan.hayward@arm.com>
2665 David Sherwood <david.sherwood@arm.com>
2667 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
2668 * combine.c (gen_lowpart_for_combine): Likewise.
2669 * dwarf2out.c (rtl_for_decl_location): Likewise.
2670 * final.c (alter_subreg): Likewise.
2671 * rtlhooks.c (gen_lowpart_general): Likewise.
2672 (gen_lowpart_if_possible): Likewise.
2674 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2675 Alan Hayward <alan.hayward@arm.com>
2676 David Sherwood <david.sherwood@arm.com>
2678 * calls.c (expand_call): Use subreg_lowpart_offset.
2679 * cse.c (cse_insn): Likewise.
2680 * regcprop.c (copy_value): Likewise.
2681 (copyprop_hardreg_forward_1): Likewise.
2683 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2686 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
2687 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
2688 =Q constraints to +Q and into insn condition add check
2689 that operands[0] and operands[1] are equal.
2690 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
2691 =Q constraints to +Q and into insn condition add check
2692 that operands[0] is equal to either operands[1] or operands[2].
2695 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
2696 instead of handling MINUS_EXPR twice (once for each argument),
2697 canonicalize operand order and handle just once, use rtype where
2698 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
2701 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
2702 any values of __C while still being pattern recognizable as a simple
2705 2017-10-13 Richard Biener <rguenther@suse.de>
2707 PR tree-optimization/82451
2709 2017-10-02 Richard Biener <rguenther@suse.de>
2711 PR tree-optimization/82355
2712 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2713 a mapping for the enclosing loop but avoid generating one for
2715 (copy_bb_and_scalar_dependences): Remove premature codegen
2716 error on PHIs in blocks duplicated into multiple places.
2717 * graphite-scop-detection.c
2718 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2719 in the region use it as loop and nest to analyze the DR in.
2720 (try_generate_gimple_bb): Likewise.
2721 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2722 (add_loop_constraints): For blocks in a loop not in the region
2723 create a dimension with a single iteration.
2724 * sese.h (gbb_loop_at_index): Remove assert.
2726 * cfgloop.c (loop_preheader_edge): For the loop tree root
2727 return the single successor of the entry block.
2728 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2729 Reset the SCEV hashtable and niters.
2730 * graphite-scop-detection.c
2731 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
2732 assert that we only have POLYNOMIAL_CHREC that vary in loops
2733 contained in the region.
2734 (scop_detection::graphite_can_represent_expr): Adjust.
2735 (scop_detection::stmt_has_simple_data_refs_p): For loops
2736 not in the region set loop to NULL. The nest is now the
2737 entry edge to the region.
2738 (try_generate_gimple_bb): Likewise.
2739 * sese.c (scalar_evolution_in_region): Adjust for
2740 instantiate_scev change.
2741 * tree-data-ref.h (graphite_find_data_references_in_stmt):
2742 Make nest parameter the edge into the region.
2743 (create_data_ref): Likewise.
2744 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
2745 entry edge into a region and adjust instantiate_scev calls.
2746 (create_data_ref): Likewise.
2747 (graphite_find_data_references_in_stmt): Likewise.
2748 (find_data_references_in_stmt): Pass the loop preheader edge
2749 from the nest argument.
2750 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
2751 parameter the edge into the region.
2752 (instantiate_parameters): Use the loop preheader edge as entry.
2753 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
2755 (get_instantiated_value_entry): Make instantiate_below parameter
2756 the edge into the region.
2757 (instantiate_scev_name): Likewise. Adjust dominance checks,
2758 when we cannot use loop-based instantiation instantiate by
2759 walking use-def chains.
2760 (instantiate_scev_poly): Adjust.
2761 (instantiate_scev_binary): Likewise.
2762 (instantiate_scev_convert): Likewise.
2763 (instantiate_scev_not): Likewise.
2764 (instantiate_array_ref): Remove.
2765 (instantiate_scev_3): Likewise.
2766 (instantiate_scev_2): Likewise.
2767 (instantiate_scev_1): Likewise.
2768 (instantiate_scev_r): Do not blindly handle N-operand trees.
2769 Do not instantiate array-refs. Handle all constants and invariants.
2770 (instantiate_scev): Make instantiate_below parameter
2771 the edge into the region.
2772 (resolve_mixers): Use the loop preheader edge for the region
2773 parameter to instantiate_scev_r.
2774 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
2776 2017-10-13 Richard Biener <rguenther@suse.de>
2778 PR tree-optimization/82525
2779 * graphite-isl-ast-to-gimple.c
2780 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
2782 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
2783 Fail code generation when we cannot represent the isl integer.
2784 (binary_op_to_tree): Elide modulo operations that are no-ops
2785 in the type we code generate. Remove now superfluous code
2788 2017-10-13 Richard Biener <rguenther@suse.de>
2790 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
2791 (scop_detection::harmful_loop_in_region): Remove premature
2792 IV type restriction.
2793 (scop_detection::graphite_can_represent_scev): We can handle
2794 pointer IVs just fine.
2796 2017-10-13 Alan Modra <amodra@gmail.com>
2798 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
2799 "Clobbers and Scratch Registers". Add paragraph on
2800 alternative to clobbers for scratch registers and OpenBLAS
2803 2017-10-13 Alan Modra <amodra@gmail.com>
2805 * doc/extend.texi (Clobbers): Correct vax example. Delete old
2806 example of a memory input for a string of known length. Move
2807 commentary out of table. Add a number of new examples
2808 covering array memory inputs.
2810 2017-10-12 Martin Liska <mliska@suse.cz>
2812 PR tree-optimization/82493
2813 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
2814 (test_range_functions): New function.
2815 (sbitmap_c_tests): Likewise.
2816 * selftest-run-tests.c (selftest::run_tests): Run new tests.
2817 * selftest.h (sbitmap_c_tests): New function.
2819 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
2821 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
2823 * config/rs6000/amo.h: Fix spacing issue.
2825 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2828 * config/i386/i386.md (*ashl<mode>3_mask_1,
2829 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
2830 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
2833 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
2835 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
2836 (profile_probability): Set max_probability
2837 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
2839 * profile-count.c (profile_probability::differs_from_p): Do not
2840 rely on max_probaiblity == 10000
2842 2017-10-12 Jeff Law <law@redhat.com>
2844 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
2847 2017-10-12 Martin Sebor <msebor@redhat.com>
2851 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
2852 (handle_alias_pairs): Call it.
2853 * common.opt (-Wattribute-alias): New option.
2854 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
2855 * doc/invoke.texi (-Wattribute-alias): Document.
2857 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
2860 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2862 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2864 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2865 (make_hard_regno_born, make_hard_regno_dead): Update
2866 bb_killed_pseudos and bb_gen_pseudos.
2868 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
2870 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
2872 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
2874 * config/alpha/alpha.c (alpha_split_conditional_move):
2875 Use std::swap instead of manually swapping.
2876 (alpha_stdarg_optimize_hook): Ditto.
2877 (alpha_canonicalize_comparison): Ditto.
2879 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2881 * tree-loop-distribution.c (struct builtin_info): New struct.
2882 (struct partition): Refactor fields into struct builtin_info.
2883 (partition_free): Free struct builtin_info.
2884 (build_size_arg_loc, build_addr_arg_loc): Delete.
2885 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
2886 information from struct builtin_info.
2887 (find_single_drs): New function refactored from classify_partition.
2888 Also moved builtin validity checks to this function.
2889 (compute_access_range, alloc_builtin): New functions.
2890 (classify_builtin_st, classify_builtin_ldst): New functions.
2891 (classify_partition): Refactor code into functions find_single_drs,
2892 classify_builtin_st and classify_builtin_ldst.
2893 (distribute_loop): Don't do runtime alias check when distributing
2895 (find_seed_stmts_for_distribution): New function.
2896 (pass_loop_distribution::execute): Refactor code finding seed
2897 stmts into above function. Support distribution for the innermost
2898 two-level loop nest. Adjust dump information.
2900 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2902 * tree-loop-distribution.c: Adjust the general comment.
2903 (NUM_PARTITION_THRESHOLD): New macro.
2904 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
2905 (classify_partition): Skip builtin pattern of loop nest's inner loop.
2906 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
2907 in call to build_partition_graph.
2908 (finalize_partitions): New parameter. Make loop distribution more
2909 conservative by fusing more partitions.
2910 (distribute_loop): Don't do runtime alias check in case of loop nest
2912 (find_seed_stmts_for_distribution): New function.
2913 (prepare_perfect_loop_nest): New function.
2914 (pass_loop_distribution::execute): Refactor code finding seed stmts
2915 and loop nest into above functions. Support loop nest distribution.
2916 Adjust dump information accordingly.
2918 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2920 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
2921 and set PTYPE_SEQUENTIAL for merged partition.
2923 2017-10-12 Richard Biener <rguenther@suse.de>
2925 PR tree-optimization/69728
2927 2017-09-19 Richard Biener <rguenther@suse.de>
2929 PR tree-optimization/69728
2930 * graphite-sese-to-poly.c (schedule_error): New global.
2931 (add_loop_schedule): Handle empty domain by failing the
2933 (build_original_schedule): Handle schedule_error.
2935 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
2936 domain by returning an unchanged schedule.
2938 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2940 * genrecog.c (validate_pattern): For VEC_SELECT verify that
2941 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
2943 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
2945 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
2948 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2951 * expr.c (store_field): Don't optimize away bitsize == 0 store
2952 from CALL_EXPR with addressable return type.
2954 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2956 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
2957 * config/rs6000/rs6000.md (sel): Delete mode attribute.
2958 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
2959 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
2960 TARGET_ISEL instead of TARGET_ISEL<sel>.
2962 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
2964 * config/rs6000/rs6000.c
2965 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
2967 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2969 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
2971 (reg_or_cint_operand): Fix comment.
2972 (reg_or_zero_operand): New predicate.
2973 * config/rs6000/rs6000-protos.h (output_isel): Delete.
2974 * config/rs6000/rs6000.c (output_isel): Delete.
2975 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
2976 instead of reg_or_cint_operand. Output instruction directly (not via
2978 (isel_unsigned_<mode>): Ditto.
2979 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
2980 gpc_reg_operand. Add an instruction alternative for this. Output
2981 instruction directly.
2982 (*isel_reversed_unsigned_<mode>): Ditto.
2984 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2986 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
2987 (TARGET_CANONICALIZE_COMPARISON): Define.
2989 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
2992 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
2993 Check whether the dest is REG before adding REG_EQUIV note.
2995 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2998 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3000 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3001 (make_hard_regno_born, make_hard_regno_dead): Update
3002 bb_killed_pseudos and bb_gen_pseudos.
3004 2017-10-11 Nathan Sidwell <nathan@acm.org>
3006 * incpath.h (enum incpath_kind): Name enum, prefix values.
3007 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
3008 * incpath.c (heads, tails): Use INC_MAX.
3009 (add_env_var_paths, add_standard_paths): Use incpath_kind.
3010 (merge_include_chains, split_quote_chain,
3011 register_include_chains): Update incpath_kind names.
3012 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
3013 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
3015 (add_framework_path, darwin_register_objc_includes): Likewise.
3016 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
3018 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3020 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
3021 Do not use float_operator operator predicate.
3022 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
3023 * config/i386/predicates.md (float_operator): Remove predicate.
3025 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3027 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
3028 (*jccxf_i387): Ditto.
3029 (*jcc<mode>_i387): Ditto.
3030 (*jccu<mode>_i387): Ditto.
3031 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
3032 (*jcc_*_i387 splitters): Remove.
3033 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
3034 * config/i386/i386.c (ix86_split_fp_branch): Remove.
3035 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
3038 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3040 * profile-count.h (slow_safe_scale_64bit): New function.
3041 (safe_scale_64bit): New inline.
3042 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
3043 * profile-count.c: Include wide-int.h
3044 (slow_safe_scale_64bit): New.
3046 2017-10-11 Nathan Sidwell <nathan@acm.org>
3048 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
3049 HAS_DECL_ASSEMBLER_NAME_P.
3050 * gimple-expr.c (gimple_decl_printable_name: Check
3051 HAS_DECL_ASSEMBLER_NAME_P too.
3052 * ipa-utils.h (type_in_anonymous_namespace_p): Check
3053 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
3054 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
3055 * passes.c (rest_of_decl_compilation): Check
3056 HAS_DECL_ASSEMBLER_NAME_P too.
3057 * recog.c (verify_changes): Likewise.
3058 * tree-pretty-print.c (dump_decl_name): Likewise.
3059 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
3061 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
3062 (SET_DECL_ASSEMBLER_NAME): Use it.
3063 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
3064 (COPY_DECL_ASSEMBLER_NAME): Likewise.
3065 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
3067 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3069 * config.gcc (i386, x86_64): Add extra objects.
3070 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
3071 (ix86_min_insn_size): Declare.
3072 (ix86_issue_rate): Declare.
3073 (ix86_adjust_cost): Declare.
3074 (ia32_multipass_dfa_lookahead): Declare.
3075 (ix86_macro_fusion_p): Declare.
3076 (ix86_macro_fusion_pair_p): Declare.
3077 (ix86_bd_has_dispatch): Declare.
3078 (ix86_bd_do_dispatch): Declare.
3079 (ix86_core2i7_init_hooks): Declare.
3080 (ix86_atom_sched_reorder): Declare.
3081 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
3082 (COSTS_N_BYTES): Move to x86-tune-costs.h.
3083 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
3084 (rip_relative_addr_p): Rename to ...
3085 (ix86_rip_relative_addr_p): ... this one; export.
3086 (memory_address_length): Update.
3087 (ix86_issue_rate): Move to x86-tune-sched.c.
3088 (ix86_flags_dependent): Move to x86-tune-sched.c.
3089 (ix86_agi_dependent): Move to x86-tune-sched.c.
3090 (exact_dependency_1): Move to x86-tune-sched.c.
3091 (exact_store_load_dependency): Move to x86-tune-sched.c.
3092 (ix86_adjust_cost): Move to x86-tune-sched.c.
3093 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
3094 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
3095 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
3096 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
3097 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
3098 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
3099 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
3100 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
3101 (min_insn_size): Rename to ...
3102 (ix86_min_insn_size): ... this one; export.
3103 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
3104 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
3105 (core2i7_first_cycle_multipass_backtrack): Move to
3106 x86-tune-sched-core.c.
3107 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
3108 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
3109 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
3110 (ix86_avoid_jump_mispredicts): Update.
3111 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
3112 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
3113 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
3114 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
3115 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
3116 (MAX_INSN): Move to ix86-tune-sched-bd.c.
3117 (MAX_IMM): Move to ix86-tune-sched-bd.c.
3118 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
3119 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
3120 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
3121 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
3122 (MAX_STORE): Move to ix86-tune-sched-bd.c.
3123 (BIG): Move to ix86-tune-sched-bd.c.
3124 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
3125 (enum insn_path): Move to ix86-tune-sched-bd.c.
3126 (get_mem_group): Move to ix86-tune-sched-bd.c.
3127 (is_cmp): Move to ix86-tune-sched-bd.c.
3128 (dispatch_violation): Move to ix86-tune-sched-bd.c.
3129 (is_branch): Move to ix86-tune-sched-bd.c.
3130 (is_prefetch): Move to ix86-tune-sched-bd.c.
3131 (init_window): Move to ix86-tune-sched-bd.c.
3132 (allocate_window): Move to ix86-tune-sched-bd.c.
3133 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
3134 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
3135 (process_end_window): Move to ix86-tune-sched-bd.c.
3136 (allocate_next_window): Move to ix86-tune-sched-bd.c.
3137 (find_constant): Move to ix86-tune-sched-bd.c.
3138 (get_num_immediates): Move to ix86-tune-sched-bd.c.
3139 (has_immediate): Move to ix86-tune-sched-bd.c.
3140 (get_insn_path): Move to ix86-tune-sched-bd.c.
3141 (get_insn_group): Move to ix86-tune-sched-bd.c.
3142 (count_num_restricted): Move to ix86-tune-sched-bd.c.
3143 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
3144 (add_insn_window): Move to ix86-tune-sched-bd.c.
3145 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
3146 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
3147 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
3148 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
3149 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
3150 (do_dispatch): Move to ix86-tune-sched-bd.c.
3151 (has_dispatch): Move to ix86-tune-sched-bd.c.
3152 * config/i386/t-i386: Add new object files.
3153 * config/i386/x86-tune-costs.h: New file.
3154 * config/i386/x86-tune-sched-atom.c: New file.
3155 * config/i386/x86-tune-sched-bd.c: New file.
3156 * config/i386/x86-tune-sched-core.c: New file.
3157 * config/i386/x86-tune-sched.c: New file.
3159 2017-10-11 Liu Hao <lh_mouse@126.com>
3161 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
3162 the generic version below instead.
3163 (should_colorize): Recognize Windows consoles as terminals
3165 * pretty-print.c [__MINGW32__] (write_all): New function.
3166 [__MINGW32__] (find_esc_head): Likewise.
3167 [__MINGW32__] (find_esc_terminator): Likewise.
3168 [__MINGW32__] (eat_esc_sequence): Likewise.
3169 [__MINGW32__] (mingw_ansi_fputs): New function that handles
3171 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
3174 2017-10-11 Richard Biener <rguenther@suse.de>
3176 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
3177 Properly call analyze_scalar_evolution with the loop of the stmt.
3179 2017-10-11 Richard Biener <rguenther@suse.de>
3181 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
3182 * tree-core.h (tree_base): Add chrec_var union member.
3183 * tree.h (CHREC_VAR): Remove.
3184 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
3185 * tree-chrec.h (build_polynomial_chrec): Adjust.
3186 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
3187 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
3189 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
3191 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
3192 * match.pd: ... here.
3193 ((T) X == (T) Y): Relax condition.
3195 2017-10-11 Bin Cheng <bin.cheng@arm.com>
3197 PR tree-optimization/82472
3198 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
3200 (break_alias_scc_partitions): Update postorder number.
3202 2017-10-11 Martin Liska <mliska@suse.cz>
3205 * opts.c (parse_no_sanitize_attribute): Do not use error_value
3207 * opts.h (parse_no_sanitize_attribute): Remove last argument.
3209 2017-10-11 Martin Liska <mliska@suse.cz>
3211 * print-rtl.c (print_insn): Move declaration of idbuf
3212 to same scope as name.
3214 2017-10-11 Martin Liska <mliska@suse.cz>
3219 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3222 2017-10-11 Martin Liska <mliska@suse.cz>
3225 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3228 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
3230 * config.gcc (default_gnu_indirect_function): Default to yes for
3231 arm*-*-linux* with glibc.
3233 2017-10-11 Richard Biener <rguenther@suse.de>
3235 * tree-scalar-evolution.c (get_scalar_evolution): Handle
3236 default-defs and types we do not want to analyze.
3237 (interpret_loop_phi): Replace unreachable code with an assert.
3238 (compute_scalar_evolution_in_loop): Remove and inline ...
3239 (analyze_scalar_evolution_1): ... here, replacing condition with
3240 what makes the intent clearer. Remove handling of cases
3241 get_scalar_evolution now handles.
3243 2017-10-10 Jim Wilson <wilson@tuliptree.org>
3245 PR rtl-optimization/81434
3246 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
3247 comment for main loop. In sched_group_found if, also add checks for
3248 pass and min_cost_group.
3250 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
3252 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
3253 (rs6000_insn_cost): New function.
3254 * config/rs6000/rs6000.md (cost): New attribute.
3256 2017-10-10 Jakub Jelinek <jakub@redhat.com>
3257 H.J. Lu <hongjiu.lu@intel.com>
3261 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
3262 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
3263 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
3264 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
3265 to OPTION_MASK_ISA_AVX512VL - builtins that have both
3266 OPTION_MASK_ISA_MMX and some other bit set require both
3267 mmx and the ISAs without the mmx bit.
3268 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
3269 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
3270 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
3271 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
3272 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
3273 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
3274 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
3275 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
3276 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
3277 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
3278 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
3279 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
3280 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
3281 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
3282 Add OPTION_MASK_ISA_MMX.
3284 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
3286 * config.gcc (armv7*-*-freebsd*): New target.
3287 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
3289 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
3291 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
3292 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
3295 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
3297 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
3298 a template parameter.
3299 (WIDE_INT_REF_FOR): Update accordingly.
3300 * tree.h (wi::int_traits <const_tree>): Delete.
3301 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
3302 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
3303 (wi::tree_to_wide_ref): New typedef.
3304 (wi::to_wide): New function.
3305 * calls.c (get_size_range): Use wi::to_wide when operating on
3307 * cgraph.c (cgraph_node::create_thunk): Likewise.
3308 * config/i386/i386.c (ix86_data_alignment): Likewise.
3309 (ix86_local_alignment): Likewise.
3310 * dbxout.c (stabstr_O): Likewise.
3311 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
3312 * expr.c (const_vector_from_tree): Likewise.
3313 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
3314 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
3315 (fold_negate_expr_1, int_const_binop_1, const_binop)
3316 (fold_convert_const_int_from_real, optimize_bit_field_compare)
3317 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
3318 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
3319 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
3320 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
3321 (fold_not_const, round_up_loc): Likewise.
3322 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
3323 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
3324 (alloca_call_type): Likewise.
3325 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
3326 * godump.c (go_output_typedef): Likewise.
3327 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
3328 * internal-fn.c (get_min_precision): Likewise.
3329 * ipa-cp.c (ipcp_store_vr_results): Likewise.
3330 * ipa-polymorphic-call.c
3331 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
3332 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
3333 (ipa_modify_call_arguments): Likewise.
3334 * match.pd: Likewise.
3335 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
3336 * print-tree.c (print_node_brief, print_node): Likewise.
3337 * stmt.c (expand_case): Likewise.
3338 * stor-layout.c (layout_type): Likewise.
3339 * tree-affine.c (tree_to_aff_combination): Likewise.
3340 * tree-cfg.c (group_case_labels_stmt): Likewise.
3341 * tree-data-ref.c (dr_analyze_indices): Likewise.
3342 (prune_runtime_alias_test_list): Likewise.
3343 * tree-dump.c (dequeue_and_dump): Likewise.
3344 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
3345 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
3346 * tree-pretty-print.c (dump_generic_node): Likewise.
3347 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
3348 (simple_iv_with_niters): Likewise.
3349 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
3350 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
3351 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
3352 * tree-ssa-loop-niter.c (split_to_var_and_offset)
3353 (refine_value_range_using_guard, number_of_iterations_ne_max)
3354 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
3355 (get_cst_init_from_scev, record_nonwrapping_iv)
3356 (scev_var_range_cant_overflow): Likewise.
3357 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
3358 * tree-ssa-pre.c (compute_avail): Likewise.
3359 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
3360 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
3361 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
3362 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
3363 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
3364 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
3365 (dump_case_nodes, try_switch_expansion): Likewise.
3366 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
3367 (vect_do_peeling): Likewise.
3368 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
3369 * tree-vect-stmts.c (vectorizable_load): Likewise.
3370 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
3371 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
3372 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
3373 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
3374 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
3375 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
3376 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
3377 (evrp_dom_walker::before_dom_children): Likewise.
3378 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
3379 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
3380 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
3381 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
3382 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
3383 (get_range_pos_neg): Likewise.
3384 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
3385 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
3386 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
3387 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
3388 * config/avr/avr.c (avr_fold_builtin): Likewise.
3389 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
3390 * config/msp430/msp430.c (msp430_attr): Likewise.
3391 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
3392 * config/powerpcspe/powerpcspe-c.c
3393 (altivec_resolve_overloaded_builtin): Likewise.
3394 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
3395 (rs6000_expand_ternop_builtin): Likewise.
3396 * config/rs6000/rs6000-c.c
3397 (altivec_resolve_overloaded_builtin): Likewise.
3398 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
3399 (rs6000_expand_ternop_builtin): Likewise.
3400 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
3402 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3404 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
3405 when copying loop nest with only one inner loop.
3407 2017-10-10 Richard Biener <rguenther@suse.de>
3409 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
3410 blocks if SCEV is active.
3411 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
3413 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
3414 (scev_initialize): Assert we are not yet initialized.
3416 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3418 * tree-loop-distribution.c (generate_loops_for_partition): Remove
3419 inner loop's exit stmt by making it always exit the loop, otherwise
3420 we would generate an infinite empty loop.
3422 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3424 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
3425 renaming variables in new preheader if it's deleted.
3427 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3429 * tree-loop-distribution.c (struct partition): Remove unused field
3430 loops of the structure.
3431 (partition_alloc, partition_free): Ditto.
3432 (build_rdg_partition_for_vertex): Ditto.
3434 2017-10-09 Jeff Law <law@redhat.com>
3436 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
3437 return type to match prototype and documentation.
3439 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3441 * config/rs6000/rs6000.c (processor_costs): Move to ...
3442 * config/rs6000/rs6000.h: ... here.
3443 (rs6000_cost): Declare.
3445 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
3447 * except.c (setjmp_fn): New global variable.
3448 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
3449 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
3450 if DONT_USE_BUILTIN_SETJMP is defined.
3452 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3454 * target.def (insn_cost): New hook.
3455 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
3456 * doc/tm.texi: Regenerate.
3457 * rtlanal.c (insn_cost): Use the new hook.
3459 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3461 * combine.c (combine_validate_cost): Compute the new insn_cost,
3462 not just pattern_cost.
3463 (try_combine): Adjust comment.
3465 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3467 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
3469 * combine.c (uid_insn_cost): Adjust comment.
3470 (combine_validate_cost): Adjust comment. Use pattern_cost instead
3472 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
3473 * dse.c (find_shift_sequence): Ditto.
3474 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
3475 (bb_valid_for_noce_process_p): Use pattern_cost.
3476 * rtl.h (insn_rtx_cost): Delete.
3477 (pattern_cost): New prototype.
3478 (insn_cost): New prototype.
3479 * rtlanal.c (insn_rtx_cost): Rename to...
3480 (pattern_cost): ... this.
3483 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
3485 * config/i386/i386.md (*jcc_2): Remove insn pattern.
3486 (*jcc<mode>_0_r_i387): Ditto.
3487 (*jccxf_r_i387): Ditto.
3488 (*jcc<mode>_r_i387): Ditto.
3489 (*jccu<mode>_r_i387): Ditto.
3490 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
3491 (*jcc): Rename from *jcc_1.
3493 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3495 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
3496 deferred rescans after the lvx/stvx recombination pre-pass.
3498 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
3500 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
3501 memory operation instruction support.
3502 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
3503 (rs6000-ibm-aix[789]*): Likewise.
3504 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
3505 Document new functions.
3507 2017-10-09 Richard Biener <rguenther@suse.de>
3509 PR tree-optimization/82397
3510 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
3511 equality only for semantically equal trees.
3513 2017-10-09 Richard Biener <rguenther@suse.de>
3515 PR tree-optimization/82449
3516 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
3517 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
3518 allow constant addresses.
3519 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
3522 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3524 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
3527 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3530 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
3533 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3536 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
3538 2017-10-09 Jakub Jelinek <jakub@redhat.com>
3541 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
3542 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
3544 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
3546 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
3547 (WI_BINARY_PREDICATE_RESULT): Likewise.
3548 (wi::binary_traits::operator_result): New type.
3549 (wi::binary_traits::predicate_result): Likewise.
3550 (generic_wide_int::operator~, unary generic_wide_int::operator-)
3551 (generic_wide_int::operator==, generic_wide_int::operator!=)
3552 (generic_wide_int::operator&, generic_wide_int::and_not)
3553 (generic_wide_int::operator|, generic_wide_int::or_not)
3554 (generic_wide_int::operator^, generic_wide_int::operator+
3555 (binary generic_wide_int::operator-, generic_wide_int::operator*):
3557 (operator~, unary operator-, operator==, operator!=, operator&)
3558 (operator|, operator^, operator+, binary operator-, operator*): New
3560 * expr.c (get_inner_reference): Use wi::bit_and_not.
3561 * fold-const.c (fold_binary_loc): Likewise.
3562 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
3563 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
3564 (bit_value_binop): Likewise.
3565 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
3566 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
3567 (extract_range_from_binary_expr_1): Likewise.
3568 (masked_increment): Likewise.
3569 (simplify_bit_ops_using_ranges): Likewise.
3571 2017-10-09 Martin Jambor <mjambor@suse.cz>
3574 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
3575 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
3576 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
3578 (hsa_fixup_mov_insn_type): New function.
3579 (hsa_op_with_type::get_in_type): Use it.
3580 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
3581 immediates in an assert.
3582 (hsa_op_with_type::extend_int_to_32bit): New method.
3583 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
3584 types. Convert to dest type if necessary.
3585 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
3586 (reg_for_gimple_ssa): Pass false as min32int to
3587 hsa_type_for_scalar_tree_type.
3588 (gen_hsa_addr): Fixup type when creating addresable temporary.
3589 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
3590 (gen_hsa_unary_operation): Extend operands and convert to dest type if
3591 necessary. Call hsa_fixup_mov_insn_type.
3592 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
3593 extend operands and convert to dest type if necessary.
3594 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
3595 to dest type if necessary.
3596 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
3597 if conversion nt necessary and size matches.
3598 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
3599 to dest type if necessary.
3600 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
3601 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
3603 (gen_hsa_clrsb): Likewise.
3604 (gen_hsa_ffs): Likewise.
3605 (gen_hsa_divmod): Extend operands and convert to dest type if
3607 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
3609 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3611 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
3612 Remove empty default arguments. Use a brace block as output
3614 (conditional return): Ditto.
3616 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
3617 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
3619 (group_ending_nop): Ditto.
3620 (doloop_end): Ditto.
3621 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
3622 (splitters for those): Ditto.
3624 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3626 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
3627 a conditional jump (and the compare for it) so that pc_rtx is the
3629 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
3630 for the deleted and renamed ctr<mode>_internal[234] patterns.
3631 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
3632 Delete second conditional return pattern.
3633 (ctr<mode>_internal2): Delete this second bdnz pattern.
3634 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
3635 (ctr<mode>_internal4): Delete this second bdz pattern.
3637 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3639 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
3640 (always_initialized_rtx_for_ssa_name_p): New predicate.
3641 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
3642 (finish_out_of_ssa): Free new field of SA.
3643 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
3644 * tree-ssa-coalesce.c: Include tree-ssa.h.
3645 (get_parm_default_def_partitions): Remove extern keyword.
3646 (get_undefined_value_partitions): New function.
3647 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
3648 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
3650 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
3652 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3654 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
3656 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3658 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
3659 for targets that preffer 128bit.
3661 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3663 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
3665 2017-10-08 Olivier Hainque <hainque@adacore.com>
3667 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
3668 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
3670 (thumb_set_return_address): Likewise.
3672 2017-10-08 Olivier Hainque <hainque@adacore.com>
3674 * common/config/arm/arm-common.c (arm_except_unwind_info):
3675 Handle DWARF2_UNWIND_INFO.
3677 2017-10-07 Michael Collison <michael.collison@arm.com>
3679 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
3682 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
3684 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
3685 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
3686 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
3687 defined, force the creation of a new block for a dispatch label.
3689 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
3691 * invoke.texi (Wsuggest-attribute=cold): Document.
3692 * common.opt (Wsuggest-attribute=cold): New
3693 * ipa-pure-const.c (warn_function_cold): New function.
3694 * predict.c (compute_function_frequency): Use it.
3695 * predict.h (warn_function_cold): Declare.
3697 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
3699 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
3702 2017-10-06 Martin Liska <mliska@suse.cz>
3704 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
3705 keyword for member functions.
3706 (struct sanopt_tree_couple): New struct.
3707 (struct sanopt_tree_couple_hash): New function.
3708 (struct sanopt_ctx): Add new hash_map.
3709 (has_dominating_ubsan_ptr_check): New function.
3710 (record_ubsan_ptr_check_stmt): Likewise.
3711 (maybe_optimize_ubsan_ptr_ifn): Likewise.
3712 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
3713 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
3715 2017-10-06 Sudakshina Das <sudi.das@arm.com>
3718 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
3719 aarch64_simd_valid_immediate on CONST_VECTORs.
3720 (aarch64_reg_or_bic_imm): Likewise.
3722 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
3724 PR rtl-optimization/82396
3725 * haifa-sched.c (ready_sort_real): Disable qsort checking.
3727 2017-10-06 Sebastian Pop <sebpop@gmail.com>
3729 * graphite-dependences.c (scop_get_reads): Move code to...
3730 (scop_get_must_writes): Move code to...
3731 (scop_get_may_writes): Move code to...
3732 (scop_get_reads_and_writes): ... here.
3733 (scop_get_dependences): Call scop_get_reads_and_writes.
3735 2017-10-06 Jakub Jelinek <jakub@redhat.com>
3737 PR tree-optimization/82434
3738 * fold-const.h (can_native_encode_type_p,
3739 can_native_encode_string_p): Remove.
3740 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
3741 don't encode anything, just return what would be otherwise returned.
3742 (native_encode_fixed, native_encode_complex, native_encode_vector):
3744 (native_encode_string): Likewise. Inline by hand
3745 can_native_encode_string_p.
3746 (can_native_encode_type_p): Remove.
3747 (can_native_encode_string_p): Remove.
3748 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
3749 STRING_CSTs using can_native_encode_string_p, test all
3750 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
3751 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
3752 argument from native_encode_expr.
3753 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
3754 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
3757 2017-10-06 Richard Biener <rguenther@suse.de>
3759 PR tree-optimization/82397
3760 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
3761 operand_equal_p but rely on data_ref_compare_tree for detecting
3763 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
3764 to match up with dr_group_sort_cmp.
3766 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3769 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
3771 * config/s390/s390-builtin-types.def: Regenerate.
3773 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3776 * config/s390/s390-builtin-types.def: Regenerate.
3777 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
3778 Change flag from B_VXE to B_VX.
3779 (s390_vec_min_dbl): Remove B_VXE flag.
3781 2017-10-06 Richard Biener <rguenther@suse.de>
3783 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
3784 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
3785 translate_isl_ast_to_gimple::is_valid_rename,
3786 translate_isl_ast_to_gimple::get_rename,
3787 translate_isl_ast_to_gimple::get_def_bb_for_const,
3788 translate_isl_ast_to_gimple::get_new_name,
3789 translate_isl_ast_to_gimple::collect_all_ssa_names,
3790 translate_isl_ast_to_gimple::copy_loop_phi_args,
3791 translate_isl_ast_to_gimple::collect_all_ssa_names,
3792 translate_isl_ast_to_gimple::copy_loop_phi_args,
3793 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
3794 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
3795 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
3796 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
3797 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
3798 translate_isl_ast_to_gimple::copy_cond_phi_args,
3799 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
3800 translate_isl_ast_to_gimple::edge_for_new_close_phis,
3801 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
3802 translate_isl_ast_to_gimple::rename_uses,
3803 translate_isl_ast_to_gimple::rename_all_uses): Remove.
3804 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
3805 (set_rename_for_each_def): Likewise.
3806 (graphite_copy_stmts_from_block): Handle debug stmt resetting
3807 here. Handle rewriting SCEV analyzable uses here.
3808 (copy_bb_and_scalar_dependences): Generate code for PHI
3810 (graphite_regenerate_ast_isl): Adjust.
3811 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
3812 (add_write, add_read): New functions.
3813 (build_cross_bb_scalars_def): Use it and simplify.
3814 (build_cross_bb_scalars_use): Likewise.
3815 (graphite_find_cross_bb_scalar_vars): Inline into...
3816 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
3817 simulating out-of-SSA. Compute liveout and add dependencies.
3818 (build_scops): Force an empty entry block.
3819 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
3821 (sese_build_liveouts): Declare.
3822 (sese_trivially_empty_bb_p): Likewise.
3823 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
3824 compute liveout and debug_liveout.
3825 (sese_bad_liveouts_use): Remove.
3826 (sese_reset_debug_liveouts_bb): Likewise.
3827 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
3828 (sese_build_liveouts): Build liveout and debug_liveout and store
3830 (new_sese_info): Adjust.
3831 (free_sese_info): Likewise.
3832 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
3833 do not build liveout here.
3834 (move_sese_in_condition): Adjust region entry.
3835 (scev_analyzable_p): Match up with chrec_apply requirements.
3836 (sese_trivially_empty_bb_p): New.
3837 * tree-into-ssa.c (get_reaching_def): Properly support generating
3838 default-defs for incremental rewrite of anonymous names.
3840 2017-10-06 Richard Biener <rguenther@suse.de>
3842 * graphite-sese-to-poly.c (extract_affine): For casts increasing
3843 precision do not perform modulo reduction.
3845 2017-10-06 Richard Biener <rguenther@suse.de>
3847 PR tree-optimization/82436
3848 * tree-vect-slp.c (vect_supported_load_permutation_p): More
3849 conservatively choose the vectorization factor when checking
3850 whether we can perform the required load permutation.
3851 (vect_transform_slp_perm_load): Assert when we may not fail.
3853 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
3855 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
3856 message for incompatible -msdata=* and -mcall-* options.
3858 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3860 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
3861 rate for post-reload scheduling.
3863 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3865 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
3867 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3869 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
3870 to improve monte carlo in scimark.
3872 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3874 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
3875 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
3876 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
3877 pentium4_cost, nocona_cost): Set reassociation width to 1.
3878 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
3879 width to 2 for fp operations and 1 otherwise.
3880 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
3882 (atom_cost): Set reassociation width to 2.
3883 (slm_cost, generic_cost): Set fp reassociation width
3884 to 2 and 1 otherwise.
3885 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
3886 (core_cost): Set fp reassociation width to 4 and vector to 2.
3887 (ix86_reassociation_width): Rewrite using cost table; special case
3888 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
3889 and TARGET_AVX128_OPTIMAL.
3890 * config/i386/i386.h (processor_costs): Add
3891 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
3892 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
3893 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
3894 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
3895 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
3896 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
3898 2017-10-05 Nathan Sidwell <nathan@acm.org>
3900 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
3902 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3904 * config/arm/arm.c (arm_test_fpu_data): New.
3905 (arm_run_selftests): Call arm_test_fpu_data.
3907 2017-10-04 Nathan Sidwell <nathan@acm.org>
3909 * toplev.c (toplev::main): Remove excess parens on pretty_printer
3911 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
3913 2017-10-04 Sudakshina Das <sudi.das@arm.com>
3915 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
3916 check type for aarch64_simd_valid_immediate.
3917 (aarch64_output_simd_mov_immediate): Update prototype.
3918 (aarch64_simd_valid_immediate): Update prototype.
3919 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
3920 support for ORR-immediate.
3921 (and<mode>3): modified pattern to add support for BIC-immediate.
3922 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
3923 now checks for valid immediate for BIC and ORR based on new enum
3925 (aarch64_output_simd_mov_immediate): Function now used to output
3926 BIC/ORR imm as well based on new enum argument.
3927 * config/aarch64/constraints.md (Do): New vector immediate constraint.
3929 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
3930 (aarch64_reg_or_bic_imm): Likewise.
3932 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3934 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
3935 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
3937 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3941 PR rtl-optimization/82396
3942 * haifa-sched.c (autopref_multipass_init): Simplify
3944 (autopref_rank_data): Simplify sort order.
3945 * sched-int.h (autopref_multipass_data_): Remove
3946 multi_mem_insn_p, min_offset and max_offset.
3948 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3950 * doc/sourcebuild.texi: Document vect_peeling_profitable.
3952 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3954 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
3957 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3959 * doc/sourcebuild.texi: Document vect_long_mult.
3961 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
3963 PR tree-optimization/82413
3964 * fold-const.c (build_range_check): Use widest_int when comparing
3965 the maximum ETYPE value with HIGH.
3967 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3969 PR rtl-optimization/82396
3970 * haifa-sched.c (autopref_multipass_init): Simplify
3972 (autopref_rank_data): Simplify sort order.
3973 * sched-int.h (autopref_multipass_data_): Remove
3974 multi_mem_insn_p, min_offset and max_offset.
3976 2017-10-04 Jakub Jelinek <jakub@redhat.com>
3978 PR tree-optimization/82381
3979 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
3980 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
3983 PR tree-optimization/82374
3984 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
3985 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
3986 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
3987 current_function_decl to the new decl.
3989 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
3991 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
3992 helper macro for IEEE float128 hardware built-in functions.
3993 (SQRTF128_ODD): Add built-in functions with the round-to-odd
3995 (TRUNCF128_ODD): Likewise.
3996 (ADDF128_ODD): Likewise.
3997 (SUBF128_ODD): Likewise.
3998 (MULF128_ODD): Likewise.
3999 (DIVF128_ODD): Likewise.
4000 (FMAF128_ODD): Likewise.
4001 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
4002 UNSPEC_TRUNC_ROUND_TO_ODD.
4003 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
4004 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
4005 floating point round to odd instructions.
4006 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
4007 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
4008 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
4009 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
4010 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
4011 (trunc<mode>sf2_hw): Change the truncate with round to odd
4012 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
4013 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
4014 to odd hardware instructions.
4015 (sub<mode>3_odd): Likewise.
4016 (mul<mode>3_odd): Likewise.
4017 (div<mode>3_odd): Likewise.
4018 (sqrt<mode>2_odd): Likewise.
4019 (fma<mode>4_odd): Likewise.
4020 (fms<mode>4_odd): Likewise.
4021 (nfma<mode>4_odd): Likewise.
4022 (nfms<mode>4_odd): Likewise.
4023 (trunc<mode>df2_odd): Change the truncate with round to odd
4024 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
4026 * doc/extend.texi (PowerPC built-in functions): Update documentation
4027 for existing IEEE float128-bit built-in functions. Add built-in
4028 functions that generate the IEEE 128-bit floating point round to
4031 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
4033 PR rtl-optimization/77729
4034 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
4035 to (X&(C1&~C2))|C2 transformations.
4037 2017-10-03 Martin Jambor <mjambor@suse.cz>
4039 PR tree-optimization/82363
4040 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
4041 mismatch, mark lacc written regardless of racc.
4043 2017-10-03 Jakub Jelinek <jakub@redhat.com>
4045 PR tree-optimization/82381
4046 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
4047 stmt_to_insert nor wheather SSA_NAMEs are default defs.
4048 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
4049 fallthrough into reassoc_stmt_dominates_stmt_p.
4052 * combine.c (combine_instructions): Don't combine in unreachable
4055 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4058 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
4059 function to not use the have_cpu variable. Do not set cpu_index,
4060 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
4062 (rs6000_valid_attribute_p): Remove duplicate initializations of
4063 old_optimize and func_optimize.
4064 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
4065 (rs6000_activate_target_options): Make global.
4066 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
4069 2017-10-02 Jakub Jelinek <jakub@redhat.com>
4071 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
4072 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
4073 Set *poffset to 0 and *psize and *pmax_size to -1 if
4074 *poffset + *psize overflows in HOST_WIDE_INT.
4076 PR tree-optimization/82387
4077 PR tree-optimization/82388
4078 PR tree-optimization/82389
4079 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
4080 instead of live_bytes non-NULL.
4082 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
4085 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
4088 2017-10-02 Richard Biener <rguenther@suse.de>
4090 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
4091 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
4092 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
4094 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4096 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
4097 requested precision matches the type's.
4098 * calls.c (alloc_max_size): Calculate the new candidate size as
4099 a widest_int and use wi::to_widest when comparing it with the
4100 current candidate size.
4101 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
4102 zero rather than integer_zero_node.
4103 * match.pd: Check for a no-op conversion before using wi::add
4104 rather than after. Use tree_to_uhwi when summing small shift
4105 counts into an unsigned int.
4107 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4108 Alan Hayward <alan.hayward@arm.com>
4109 David Sherwood <david.sherwood@arm.com>
4112 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
4113 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
4114 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
4115 POINTER_AND_FP_REGS.
4117 2017-10-02 Richard Biener <rguenther@suse.de>
4119 PR tree-optimization/82355
4120 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4121 a mapping for the enclosing loop but avoid generating one for
4123 (copy_bb_and_scalar_dependences): Remove premature codegen
4124 error on PHIs in blocks duplicated into multiple places.
4125 * graphite-scop-detection.c
4126 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4127 in the region use it as loop and nest to analyze the DR in.
4128 (try_generate_gimple_bb): Likewise.
4129 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4130 (add_loop_constraints): For blocks in a loop not in the region
4131 create a dimension with a single iteration.
4132 * sese.h (gbb_loop_at_index): Remove assert.
4134 2017-10-01 Kevin Buettner <kevinb@redhat.com>
4136 * omp-expand.c (adjust_context_scope): New function.
4137 (expand_parallel_call): Call adjust_context_scope.
4139 2017-10-01 Jeff Law <law@redhat.com>
4141 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
4142 dom_opt_dom_walker class with direct access to private members.
4143 Add comments. Call test_for_singularity.
4144 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
4145 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
4146 m_dummy_cond anymore.
4147 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
4149 (pass_dominator:execute): Build the dummy_cond here and pass it
4150 to the dom_opt_dom_walker ctor.
4151 (test_for_singularity): New function.
4153 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
4154 Maya Rashish <coypu@sdf.org>
4156 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
4157 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
4158 (alpha*-*-netbsd*) Use nbsd_tm_file.
4159 (arm*-*-netbsdelf*) Likewise.
4160 (i[34567]86-*-netbsdelf*) Likewise.
4161 (x86_64-*-netbsd*) Likewise.
4162 (mips*-*-netbsd*) Likewise.
4163 (powerpc-*-netbsd*) Likewise.
4164 (sh*-*-netbsd*) Likewise.
4165 (sparc-*-netbsdelf*) Likewise.
4166 (sparc64-*-netbsd*) Likewise.
4167 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
4169 (vax-*-netbsdelf*) Likewise.
4170 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
4171 (UINT_FAST8_TYPE) Likewise.
4172 (INT_FAST16_TYPE) Check CHAR_FAST16.
4173 (UINT_FAST16_TYPE) Likewise.
4175 2017-09-30 Jakub Jelinek <jakub@redhat.com>
4178 * config/i386/i386.md
4179 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
4180 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
4181 *divmodsi4_zext_2): New define_insn_and_split.
4182 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
4183 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
4184 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
4185 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
4186 New define_insn_and_split.
4187 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
4188 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
4189 operands[1] having DImode when mode is SImode.
4191 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
4192 always SImode for DIV and MOD in REG_EQUAL notes.
4194 2017-09-29 Yury Gribov <tetra2005@gmail.com>
4197 * match.pd: Fix handling of NaNs in pattern.
4199 2017-09-29 Jeff Law <law@redhat.com>
4201 * sbitmap.c (bitmap_bit_in_range_p): New function.
4202 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
4203 * tree-ssa-dse.c (live_bytes_read): New function.
4204 (dse_classify_store): Ignore reads of dead bytes.
4206 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
4207 typos and whitespace errors.
4208 * config/i386/predicates.md (address_no_seg_operand): Likewise.
4209 * config/s390/s390.c (s390_emit_prologue): Likewise.
4211 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4214 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
4215 with a symbol for LRA.
4217 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4219 PR rtl-optimization/82338
4220 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
4222 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
4224 * genmodes.c (calc_wider_mode): Suppress qsort macro.
4225 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
4226 (qsort_chk): Declare.
4227 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
4228 (qsort_chk): New function.
4230 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4232 PR tree-optimization/82337
4233 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
4234 phi definition if the PHI result appears in an abnormal PHI.
4235 (find_basis_for_base_expr): Don't record a basis if the LHS of the
4236 basis appears in an abnormal PHI.
4238 2017-09-29 Richard Biener <rguenther@suse.de>
4240 * graphite-isl-ast-to-gimple.c
4241 (translate_isl_ast_to_gimple::set_codegen_error): New function.
4242 (binary_op_to_tree): Use it.
4243 (get_rename_from_scev): Likewise.
4244 (copy_loop_phi_nodes): Likewise.
4245 (copy_bb_and_scalar_dependences): Likewise.
4246 (translate_pending_phi_nodes): Likewise.
4248 2017-09-29 Jakub Jelinek <jakub@redhat.com>
4251 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
4252 for movabsq $(i32 << shift), r64.
4254 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
4256 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
4257 index when encoding %esp as %rsp to avoid 0x67 prefix.
4259 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
4261 * config/i386/i386.md (*movsf_internal, *movdf_internal):
4262 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
4264 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
4266 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
4267 Extensions with more than 16 double VFP registers.
4268 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
4269 to_clear_mask and all code related to it. Replace the remaining
4270 entry by a sbitmap and adapt code accordingly.
4272 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
4274 * brig-builtins.def: Change pure attributes to const.
4276 2017-09-28 Joseph Myers <joseph@codesourcery.com>
4278 * config.gcc (default_gnu_indirect_function): Default to yes for
4279 sparc*-*-linux* with glibc.
4281 2017-09-28 Joseph Myers <joseph@codesourcery.com>
4283 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
4284 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
4285 when creating .init_array and .fini_array sections with priority
4288 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
4291 * config/aarch64/aarch64.c
4292 (aarch64_builtin_support_vector_misalignment): Always return false
4293 when misalignment is unknown.
4295 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
4297 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
4298 this function to return false if the definition used by the swap
4299 instruction is artificial, or if the memory address from which the
4300 constant value is loaded is not represented by a base address held
4301 in a register or if the base address register is a frame or stack
4302 pointer. Additionally, return false if the base address of the
4303 loaded constant is a SYMBOL_REF but is not considered to be a
4305 (replace_swapped_load_constant): New function.
4306 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
4307 loaded constant vector with a load of a swapped constant vector.
4309 2017-09-27 Carl Love <cel@us.ibm.com>
4311 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
4312 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
4313 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
4316 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
4318 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
4319 first, always call autopref_rank_data otherwise.
4321 2017-09-27 Richard Biener <rguenther@suse.de>
4323 * graphite-scop-detection.c (find_scop_parameters): Move
4324 loop bound handling ...
4325 (gather_bbs::before_dom_children): ... here, avoiding the need
4326 to build scop_info->loop_nest.
4327 (record_loop_in_sese): Remove.
4328 * sese.h (sese_info_t::loop_nest): Remove.
4329 * sese.c (new_sese_info): Do not allocate loop_nest.
4330 (free_sese_info): Do not free loop_nest.
4332 2017-09-27 Jakub Jelinek <jakub@redhat.com>
4335 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
4336 lhs from calls if the lhs has addressable type.
4338 2017-09-27 Richard Biener <rguenther@suse.de>
4340 * graphite.h (scop::max_alias_set): New member.
4341 * graphite-scop-detection.c: Remove references to non-existing
4342 --param in comments.
4343 (build_alias_sets): Record the maximum alias set used for drs.
4344 (build_scops): Support zero as unlimited for
4345 --param graphite-max-arrays-per-scop.
4346 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
4348 (build_poly_sr_1): ... here. Compute alias set based on the
4349 maximum alias set used for drs rather than
4350 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
4352 2017-09-27 Richard Biener <rguenther@suse.de>
4354 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
4355 --param loop-block-tile-size=0 to disable tiling.
4357 2017-09-27 Richard Biener <rguenther@suse.de>
4359 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
4360 (graphite-max-nb-scop-params): Document special value zero.
4361 * domwalk.h (dom_walker::STOP): New symbolical constant.
4362 (dom_walker::dom_walker): Add optional parameter for bb to
4364 (dom_walker::~dom_walker): Declare.
4365 (dom_walker::before_dom_children): Document STOP return value.
4366 (dom_walker::m_user_bb_to_rpo): New member.
4367 (dom_walker::m_bb_to_rpo): Likewise.
4368 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
4369 mapping here if not provided by the user.
4370 (dom_walker::~dom_walker): Free bb to RPO mapping if not
4371 provided by the user.
4372 (dom_walker::STOP): Define.
4373 (dom_walker::walk): Do not compute bb to RPO mapping here.
4374 Support STOP return value from before_dom_children to stop
4376 * graphite-optimize-isl.c (optimize_isl): If the schedule
4377 is the same still generate code if -fgraphite-identity
4378 or -floop-parallelize-all are given.
4379 * graphite-scop-detection.c: Include cfganal.h.
4380 (gather_bbs::gather_bbs): Get and pass through bb to RPO
4382 (gather_bbs::before_dom_children): Return STOP for BBs
4384 (build_scops): Compute bb to RPO mapping and pass it to
4385 the domwalk. Treat --param graphite-max-nb-scop-params=0
4386 as not limiting the number of params.
4387 * graphite.c (graphite_initialize): Remove limit on the
4388 number of basic-blocks in a function.
4389 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
4390 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
4391 default value of 10.
4393 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4395 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
4396 Adjust code to eliminate needing to do the shift right 32-bits
4397 operation after XSCVDPSPN.
4399 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4401 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
4402 ((X / Y) != 0 -> X >= Y): Likewise.
4404 2017-09-26 Carl Love <cel@us.ibm.com>
4406 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
4407 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
4408 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
4409 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
4410 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
4411 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
4412 definitions and overloading.
4413 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
4414 statement for P9V_BUILTIN_XST_LEN_R.
4415 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
4416 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
4417 define_expand and define_insn for the instructions and builtins.
4418 * doc/extend.texi: Update the built-in documentation file for the new
4420 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
4421 define_insn for the instructions
4423 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
4426 * gcc/config/netbsd-protos.h: New file.
4427 * gcc/config/netbsd.c: New file.
4428 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
4429 * gcc/config/t-netbsd: New file.
4430 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
4431 (tmake_file) Add t-netbsd.
4432 (extra_objs) Add netbsd.o.
4434 2017-09-26 Janus Weil <janus@gcc.gnu.org>
4438 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
4440 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4442 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
4443 sign extension from a vector register to a GPR by doing a 32-bit
4444 direct move and then an EXTSW.
4445 (extendsi<mode>2 splitter): Likewise.
4446 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
4447 right or vector extract after doing XSCVDPSPN. Use
4448 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
4450 (movdi_from_sf_zero_ext): Likewise.
4451 (reload_gpr_from_vsxsf): Likewise.
4452 (p8_mfvsrd_4_disf): Delete, no longer used.
4453 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
4454 then needing to move the SFmode to a GPR to use the XSCVDPSP
4455 instruction instead of FRSP and XSCVDPSPN.
4456 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
4457 it is adjacent to the other XSCVSPDP insns.
4458 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
4459 SFmode to be in traditional Altivec registers.
4460 (vsx_xscvdpspn): Eliminate useless alternative constraint.
4461 (vsx_xscvspdpn): Likewise.
4462 (vsx_xscvspdpn_scalar): Likewise.
4464 2017-09-26 Martin Jambor <mjambor@suse.cz>
4466 * tree-sra.c (compare_access_positions): Put integral types first,
4467 stabilize sorting of integral types, remove conditions putting
4468 non-full-precision integers last.
4469 (sort_and_splice_var_accesses): Disable scalarization if a
4470 non-integert would be represented by a non-full-precision integer.
4472 2017-09-26 Joseph Myers <joseph@codesourcery.com>
4474 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
4475 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
4476 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
4477 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
4478 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
4479 conditionals inside the function instead of around it. Call
4480 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
4481 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
4483 2017-09-26 Richard Biener <rguenther@suse.de>
4485 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
4487 (scop_detection::build_scop_breadth): ... this. Removed.
4488 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
4489 (scop_detection::harmful_stmt_in_bb): Likewise.
4490 (scop_detection::graphite_can_represent_stmt): Likewise.
4491 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
4492 (scop_detection::can_represent_loop): Remove recursion, fold in ...
4493 (scop_detection::can_represent_loop_1): ... this. Removed.
4494 (scop_detection::harmful_loop_in_region): Simplify after inlining
4495 the above and remove more quadraticness.
4496 (build_scops): Adjust.
4497 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
4500 2017-09-26 Jakub Jelinek <jakub@redhat.com>
4503 * config/i386/i386.c (ix86_print_operand_address_as): Only test
4504 REGNO (base) == SP_REG if base is a REG.
4507 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
4508 if it is different SSA_NAME.
4509 (optimize_range_tests_cmp_bitwise): New function.
4510 (optimize_range_tests): Call it.
4512 2017-09-26 Richard Biener <rguenther@suse.de>
4514 PR tree-optimization/82321
4515 * graphite.c (canonicalize_loop_closed_ssa): Properly check
4516 for the def being inside the loop.
4518 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4520 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
4522 * config/s390/s390-builtins.def: Fix constraint on op4.
4524 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4526 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
4527 independent expanders.
4528 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
4529 ("vec_ordered", "vec_unordered"): New expanders.
4531 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4533 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
4536 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4538 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
4539 vec_unpacks_lo_v16qi.
4540 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
4542 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4544 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
4545 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
4546 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
4548 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4550 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
4552 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
4554 ("*vec_slb<mode>"): New insn pattern.
4555 ("vec_shr_<mode>"): New expander.
4556 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
4557 and force the shift count operand to V16QImode.
4558 ("vec_srb<mode>"): Set shift count mode to V16QI.
4560 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4562 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
4563 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
4564 ("vec_widen_smult_hi_<mode>"): New expander definitions.
4566 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
4569 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
4571 2017-09-26 Richard Biener <rguenther@suse.de>
4573 PR tree-optimization/82320
4574 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
4577 2017-09-25 Jeff Law <law@redhat.com>
4579 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
4580 prototype for new argument.
4581 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
4582 mostly extracted from rs6000_emit_allocate_stack.
4583 (rs6000_emit_probe_stack_range_stack_clash): New function.
4584 (rs6000_emit_allocate_stack): Call
4585 rs6000_emit_probe_stack_range_stack_clash as needed.
4586 (rs6000_emit_probe_stack_range): Add additional argument
4587 to call to gen_probe_stack_range{si,di}.
4588 (output_probe_stack_range): New.
4589 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
4590 (output_probe_stack_range_stack_clash): New.
4591 (rs6000_emit_prologue): Emit notes into dump file as requested.
4592 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
4593 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
4594 Add additional operand and pass it to output_probe_stack_range.
4596 2017-09-25 Bin Cheng <bin.cheng@arm.com>
4598 PR tree-optimization/82163
4599 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
4600 (checking_verify_loop_closed_ssa): New parameter.
4601 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
4602 (check_loop_closed_ssa_stmt): Delete.
4603 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
4604 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
4605 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
4608 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
4610 * brig-builtins.def: Treat HSAIL barrier builtins as
4611 setjmp/longjump style functions.
4613 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
4615 * target.def (constant_alignment): New hook.
4616 * defaults.h (CONSTANT_ALIGNMENT): Delete.
4617 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
4618 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
4619 * doc/tm.texi: Regenerate.
4620 * targhooks.h (default_constant_alignment): Declare.
4621 (constant_alignment_word_strings): Likewise.
4622 * targhooks.c (default_constant_alignment): New function.
4623 (constant_alignment_word_strings): Likewise.
4624 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
4625 instead of CONSTANT_ALIGNMENT.
4626 * varasm.c (align_variable, get_variable_align, build_constant_desc)
4627 (force_const_mem): Likewise.
4628 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
4629 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
4630 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
4631 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4632 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
4634 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
4635 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4636 constant_alignment_word_strings.
4637 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
4638 (CONSTANT_ALIGNMENT): Likewise.
4639 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4640 (arm_constant_alignment): New function.
4641 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
4642 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4643 constant_alignment_word_strings.
4644 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
4645 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4646 constant_alignment_word_strings.
4647 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
4648 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4649 (cris_constant_alignment): New function.
4650 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
4651 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4652 (epiphany_constant_alignment): New function.
4653 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
4654 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4655 constant_alignment_word_strings.
4656 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
4657 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4658 constant_alignment_word_strings.
4659 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
4660 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4661 constant_alignment_word_strings.
4662 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
4663 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
4664 * config/i386/i386.c (ix86_constant_alignment): Make static.
4665 Use the same interface as the target hook.
4666 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4667 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
4668 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4669 constant_alignment_word_strings.
4670 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
4671 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
4672 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4673 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
4674 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4675 constant_alignment_word_strings.
4676 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
4677 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4678 constant_alignment_word_strings.
4679 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
4680 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4681 constant_alignment_word_strings.
4682 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
4683 * config/microblaze/microblaze.c (microblaze_constant_alignment):
4685 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4686 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
4687 * config/mips/mips.c (mips_constant_alignment): New function.
4688 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4689 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
4690 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
4691 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4692 (mmix_constant_alignment): Make static. Use the same interface
4694 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
4695 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4696 constant_alignment_word_strings.
4697 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
4698 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4699 constant_alignment_word_strings.
4700 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
4701 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4702 constant_alignment_word_strings.
4703 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
4704 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4705 (rs6000_constant_alignment): New function.
4706 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
4707 * config/riscv/riscv.c (riscv_constant_alignment): New function.
4708 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4709 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
4710 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4711 (rs6000_constant_alignment): New function.
4712 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
4713 * config/s390/s390.c (s390_constant_alignment): New function.
4714 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4715 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
4716 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4717 constant_alignment_word_strings.
4718 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
4719 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4720 (sparc_constant_alignment): New function.
4721 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
4722 * config/spu/spu.c (spu_constant_alignment): New function.
4723 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4724 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
4725 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4726 constant_alignment_word_strings.
4727 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
4728 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4729 constant_alignment_word_strings.
4730 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
4731 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4732 constant_alignment_word_strings.
4733 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
4734 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4735 (visium_constant_alignment): New function.
4736 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
4737 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4738 (xtensa_constant_alignment): New function.
4739 * system.h (CONSTANT_ALIGNMENT): Poison.
4741 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
4743 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4744 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
4745 (rs6000_builtin_valid_without_lhs): New helper function.
4746 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4747 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
4749 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
4751 * target.h (vec_perm_indices): Use unsigned short rather than
4753 (auto_vec_perm_indices): Likewise.
4754 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
4755 Use unsigned int rather than unsigned char.
4756 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
4758 2017-09-25 Richard Biener <rguenther@suse.de>
4760 * cfgloop.h (sort_sibling_loops): Declare.
4761 * cfgloop.c (sort_sibling_loops_cmp): New helper.
4762 (sort_sibling_loops): New function sorting the sibling loop list
4764 * graphite.c (graphite_transform_loops): Sort sibling loops.
4766 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
4768 * target.def (vec_perm_const_ok): Change sel parameter to
4770 * optabs-query.c (can_vec_perm_p): Update accordingly.
4771 * doc/tm.texi: Regenerate.
4772 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
4773 auto_vec_perm_indices and remove separate nelt field.
4774 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
4775 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
4776 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
4777 (aarch64_expand_vec_perm_const): Update accordingly.
4778 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
4779 to vec_perm_indices.
4780 * config/arm/arm.c (expand_vec_perm_d): Change perm to
4781 auto_vec_perm_indices and remove separate nelt field.
4782 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
4783 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
4784 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
4786 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
4787 to vec_perm_indices.
4788 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
4789 sel to vec_perm_indices.
4790 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
4791 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
4792 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
4794 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
4797 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
4800 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
4801 on the FUNCTION_DECL function context if it has a DIE that is a
4804 2017-09-25 Richard Biener <rguenther@suse.de>
4806 PR tree-optimization/82285
4807 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
4810 2017-09-25 Tom de Vries <tom@codesourcery.com>
4814 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
4817 2017-09-25 Richard Biener <rguenther@suse.de>
4819 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
4820 ISL errors other than isl_error_quota happen. Dump if the
4821 schedule is the same.
4822 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
4823 errors instead of aborting inside ISL.
4825 2017-09-25 Iain Sandoe <iain@codesourcery.com>
4828 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
4829 of libgcc_eh for m64.
4830 * config/i386/darwin64.h: Likewise.
4832 2017-09-25 Richard Biener <rguenther@suse.de>
4835 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
4836 attribute for incomplete types nor twice for complete ones.
4838 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
4841 * config/i386/i386.c (ix86_print_operand_address_as): Encode
4842 %esp as %rsp to avoid 0x67 prefix if there is no index or base
4845 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
4848 * config/i386/i386.opt (mprefer-avx256): Use
4849 ix86_target_flags variable.
4850 * config/i386/i386.c (ix86_target_string): Move
4851 -mprefer-avx256 to flag2_opts.
4853 2017-09-22 Jakub Jelinek <jakub@redhat.com>
4856 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
4857 and x != -1 | y != -1 into (x & y) != -1.
4859 2017-09-22 Steve Ellcey <sellcey@cavium.com>
4861 * config.gcc: Add new case statement to set
4862 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
4863 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
4864 s390x-*-linux* case statements. Added aarch64 to the list of
4865 supported architectures.
4867 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4869 PR tree-optimization/82289
4870 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
4871 STMT_VINFO_RELEVANT_P.
4873 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4874 Alan Hayward <alan.hayward@arm.com>
4875 David Sherwood <david.sherwood@arm.com>
4877 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
4878 for VR_RANGE only; don't allow VR_ANTI_RANGE.
4879 (extract_range_from_binary_expr_1): Don't call
4880 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
4882 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4883 Alan Hayward <alan.hayward@arm.com>
4884 David Sherwood <david.sherwood@arm.com>
4886 * target.def (preferred_vector_alignment): New hook.
4887 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
4889 * doc/tm.texi: Regenerate.
4890 * targhooks.h (default_preferred_vector_alignment): Declare.
4891 * targhooks.c (default_preferred_vector_alignment): New function.
4892 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
4894 (DR_TARGET_ALIGNMENT): New macro.
4895 (aligned_access_p): Update commentary.
4896 (vect_known_alignment_in_bytes): New function.
4897 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
4899 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
4900 Calculate the misalignment based on the target alignment rather than
4902 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
4903 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
4904 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
4905 the target alignment, rather than masking the element misalignment
4906 with the number of elements in a vector. Also use the target
4907 alignment when calculating the maximum number of peels.
4908 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
4909 instead of TYPE_ALIGN_UNIT.
4910 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
4911 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
4912 (vect_create_addr_base_for_vector_ref): Update call accordingly.
4913 (vect_create_data_ref_ptr): Likewise.
4914 (vect_setup_realignment): Realign by ANDing with
4915 -DR_TARGET_MISALIGNMENT.
4916 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
4917 the number of peels based on DR_TARGET_ALIGNMENT.
4918 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
4919 with the guaranteed alignment boundary when deciding whether
4921 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
4922 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
4923 (ensure_base_align): Remove stmt_info parameter. Get the
4924 target base alignment from DR_TARGET_ALIGNMENT.
4925 (vectorizable_store): Update call accordingly. Interpret
4926 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
4928 (vectorizable_load): Likewise.
4930 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4931 Alan Hayward <alan.hayward@arm.com>
4932 David Sherwood <david.sherwood@arm.com>
4934 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
4935 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
4936 (vect_enhance_data_refs_alignment): Likewise.
4938 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
4940 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4941 error. Only quit immediately if parsing is complete.
4942 (BEGIN): Initialize fatal_err and parse_done.
4943 (begin fpu, end fpu): Check number of arguments.
4944 (begin arch, end arch): Likewise.
4945 (begin cpu, end cpu): Likewise.
4946 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4947 (optalias): Likewise.
4949 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
4951 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4952 * config/arm/arm-isa.h: Delete. Move definitions to ...
4953 * arm-cpus.in: ... here. Use new feature and fgroup values.
4954 * config/arm/arm.c (arm_option_override): Use lower case for feature
4956 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4957 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4958 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4960 (print_isa_bits_for): New function.
4961 (gen_isa): New function.
4962 (gen_comm_data): Use print_isa_bits_for.
4963 (define feature): New keyword.
4964 (define fgroup): New keyword.
4965 * config/arm/t-arm (TM_H): Remove.
4966 (GTM_H): Add arm-isa.h.
4967 (arm-isa.h): Add rule to generate file.
4968 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4969 case for feature bit names.
4971 2017-09-22 Richard Biener <rguenther@suse.de>
4973 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
4975 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
4976 print of no dependency loops ...
4977 * graphite.c (graphite_transform_loops): ... here.
4978 (canonicalize_loop_closed_ssa_form): Work from inner to outer
4980 (same_close_phi_node, remove_duplicate_close_phi,
4981 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
4982 (canonicalize_loop_closed_ssa): ... here and simplify.
4983 * graphite-optimize-isl.c: Include tree-vectorizer.h.
4984 (optimize_isl): Use dump_printf_loc to tell when we stopped
4985 optimizing because of an ISL timeout.
4987 2017-09-22 Richard Biener <rguenther@suse.de>
4989 PR tree-optimization/82291
4990 * tree-if-conv.c (predicate_mem_writes): Make sure to
4991 remove writes in blocks predicated with false.
4993 2017-09-22 Richard Biener <rguenther@suse.de>
4995 * sese.c: Include cfganal.h.
4996 (if_region_set_false_region): Remove.
4997 (create_if_region_on_edge): Likewise.
4998 (move_sese_in_condition): Re-implement without destroying
5001 2017-09-22 Richard Biener <rguenther@suse.de>
5003 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
5004 Verify both BBs contain loop PHI nodes before dispatching to
5006 (graphite_regenerate_ast_isl): Do not recompute dominators,
5007 do not verify three times. Restructure for clarity.
5008 * graphite-scop-detection.c (same_close_phi_node,
5009 remove_duplicate_close_phi, make_close_phi_nodes_unique,
5010 defined_in_loop_p, canonicalize_loop_closed_ssa,
5011 canonicalize_loop_closed_ssa_form): Simplify, remove excess
5012 checking and SSA rewrite, move to ...
5013 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
5014 (graphite_initialize): Do not pass in ctx, do not reset the
5015 SCEV cache, compute only dominators.
5016 (graphite_transform_loops): Allocate ISL ctx after
5017 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
5018 Maintain post-dominators only around build_scops.
5019 * sese.c (if_region_set_false_region): Make static. Free
5020 and recompute dominators.
5021 (move_sese_in_condition): Assert we don't get called with
5022 post-dominators computed.
5023 * sese.h (if_region_set_false_region): Remove.
5025 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
5027 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
5028 mode attribute for TARGET_AVX512VL.
5030 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
5032 * config/i386/i386.opt (mprefer-avx256): New option.
5033 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
5035 (ix86_preferred_simd_mode): Return 256-bit AVX modes
5036 for TARGET_PREFER_AVX256.
5037 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
5039 2017-09-21 Jeff Law <law@redhat.com>
5041 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
5042 Fix dump output if the only stack space is for pushed registers.
5044 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5046 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
5049 2017-09-21 Martin Sebor <msebor@redhat.com>
5052 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
5053 code (in C++) or code that triggers warnings.
5055 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
5057 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
5059 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
5061 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
5062 * sched-rgn.c: Ditto.
5063 * sel-sched-ir.c: Ditto.
5065 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
5067 * toplev.h (set_random_seed): Adjust return type.
5068 * toplev.c (init_local_tick): Move eager initialization of random_seed
5069 to get_random_seed. Adjust comment.
5070 (init_random_seed): Inline to get_random_seed, delete.
5071 (get_random_seed): Initialize random_seed lazily.
5072 (set_random_seed): Do not return previous value.
5073 (print_switch_value): Do not call get_random_seed.
5075 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
5077 * cgraph.c (delete_function_version): New, broken out from...
5078 (cgraph_node::delete_function_version): ...here. Rename to
5079 cgraph_node::delete_function_version_by_decl. Update all uses.
5080 (cgraph_node::remove): Call delete_function_version.
5082 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5085 * tree-inline.c (expand_call_inline): Emit clobber stmts for
5086 VAR_DECLs to which addressable non-volatile parameters are mapped
5087 and for id->retvar after the return value assignment. Clear
5088 id->retval and id->retbnd after inlining.
5090 2017-09-21 Richard Biener <rguenther@suse.de>
5092 PR tree-optimization/82276
5093 PR tree-optimization/82244
5094 * tree-vrp.c (build_assert_expr_for): Set
5095 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
5097 (remove_range_assertions): Revert earlier change.
5099 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
5102 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
5104 2017-09-21 Richard Biener <rguenther@suse.de>
5106 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5107 Restore valid IL after code generation errors.
5108 * graphite.c (graphite_transform_loops): Diagnose code
5109 generation issues as MSG_MISSED_OPTIMIZATION and continue
5110 with processing SCOPs.
5112 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5113 Alan Hayward <alan.hayward@arm.com>
5114 David Sherwood <david.sherwood@arm.com>
5116 * calls.c (compute_argument_addresses): Use simplify_gen_binary
5117 rather than choosing between plus_constant and gen_rtx_<CODE>.
5118 * expr.c (emit_push_insn): Likewise.
5119 (expand_expr_real_2): Likewise.
5121 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5122 Alan Hayward <alan.hayward@arm.com>
5123 David Sherwood <david.sherwood@arm.com>
5125 * loop-unroll.c (split_iv): Call copy_rtx on the step.
5127 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5128 Alan Hayward <alan.hayward@arm.com>
5129 David Sherwood <david.sherwood@arm.com>
5131 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
5132 calling tree_to_uhwi.
5134 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5135 Alan Hayward <alan.hayward@arm.com>
5136 David Sherwood <david.sherwood@arm.com>
5138 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
5139 INTEGER_CST rather than a negative test for ADDR_EXPR.
5141 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5142 Alan Hayward <alan.hayward@arm.com>
5143 David Sherwood <david.sherwood@arm.com>
5145 * tree-vrp.c (extract_range_from_binary_expr_1): Check
5146 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
5148 2017-09-21 Richard Biener <rguenther@suse.de>
5150 PR tree-optimization/71351
5151 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
5152 graphite_create_new_loop_guard): Remove, fold remaining parts
5154 (translate_isl_ast_node_for): ... here and simplify.
5156 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5159 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
5160 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
5161 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
5162 alternative always use QI mode, for -Os imov (=R,R) alternative
5163 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
5166 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5167 Jeff Law <law@redhat.com>
5169 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
5170 (allocate_stack_space): New function, partially extracted from
5172 (s390_emit_prologue): Track offset to most recent stack probe.
5173 Code to allocate space moved into allocate_stack_space.
5174 Dump actions when no stack is allocated.
5175 (s390_prologue_plus_offset): New function.
5176 (s390_emit_stack_probe): Likewise.
5178 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
5180 * common.opt (Wa, Wl, Wp, g, gz=): Add
5182 (gno-column-info): Remove.
5183 (gcolumn-info): Drop RejectNegative.
5185 (gno-record-gcc-switches): Remove.
5186 (grecord-gcc-switches): Drop RejectNegative.
5187 (gno-split-dwarf): Remove.
5188 (gsplit-dwarf): Drop RejectNegative.
5189 (gno-strict-dwarf): Remove.
5190 (gstrict-dwarf): Drop RejectNegative.
5191 * config/darwin.opt (gfull, gused): Add RejectNegative.
5192 * dwarf2out.c (gen_producer_string): Drop
5193 gno-record-gcc-switches handler.
5194 * optc-gen.awk: Add g to prefixes with negative forms.
5195 * opts-common.c (remapping_prefix_p): New.
5196 (find_opt): Check it.
5197 (generate_canonical_option): Test g prefix.
5198 (option_map): Add -gno- mapping.
5199 (add_misspelling_candidates): Check remapping_prefix_p.
5201 2017-09-20 Jeff Law <law@redhat.com>
5203 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
5204 thinko in stack clash protection support.
5206 * explow.c (compute_stack_clash_protection_loop_data): Use
5207 CONST_INT_P instead of explicit test. Verify object is a
5208 CONST_INT_P before looking at INTVAL.
5209 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
5210 instead of explicit test.
5212 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
5215 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
5216 address instead of to r1 and r11.
5218 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
5220 * config.gcc: Support "knm".
5221 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
5222 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5224 * config/i386/i386.c (m_KNM): Define.
5225 (processor_target_table): Add "knm".
5227 (ix86_option_override_internal): Add "knm".
5228 (ix86_issue_rate): Add PROCESSOR_KNM.
5229 (ix86_adjust_cost): Ditto.
5230 (ia32_multipass_dfa_lookahead): Ditto.
5231 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
5232 (fold_builtin_cpu): Add M_INTEL_KNM.
5233 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
5234 (processor_type): Add PROCESSOR_KNM.
5235 * config/i386/x86-tune.def: Add m_KNM.
5236 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
5238 2017-09-20 Richard Biener <rguenther@suse.de>
5240 PR tree-optimization/80213
5241 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
5242 are allowed in empty BBs as well.
5243 (canonicalize_loop_closed_ssa): Also look for other complex
5245 (scop_detection::get_sese): Include the loop-closed PHI block
5247 (scop_detection::merge_sese): Remove code adding extra blocks.
5248 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
5249 (build_scops): Assert the final returned scop is invalid.
5251 2017-09-20 Richard Biener <rguenther@suse.de>
5253 PR tree-optimization/82264
5254 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
5256 (vn_phi_lookup): Likewise.
5257 (vn_phi_insert): Likewise.
5259 2017-09-20 Jakub Jelinek <jakub@redhat.com>
5261 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
5262 that fits into uhwi or shwi, add DW_AT_const_value regardless
5263 of early_dwarf without going through RTL, using add_AT_unsigned
5266 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
5267 (DEBUG_LTO_ABBREV_SECTION): Likewise.
5268 (DEBUG_LTO_MACINFO_SECTION): Likewise.
5269 (DEBUG_MACRO_SECTION): Likewise.
5270 (DEBUG_LTO_MACRO_SECTION): Likewise.
5271 (DEBUG_STR_DWO_SECTION): Likewise.
5272 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
5273 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
5274 (DEBUG_LTO_DWO_LINE_SECTION): Define.
5275 (DEBUG_LTO_LINE_STR_SECTION): Define.
5276 (init_sections_and_labels): Initialize debug_line_str_section
5277 variable. Initialize debug_loc_section for -gdwarf-5 to
5278 DEBUG_LOCLISTS_SECTION. Formatting fixes.
5280 2017-09-20 Richard Biener <rguenther@suse.de>
5282 * graphite-sese-to-poly.c (extract_affine): Properly handle
5283 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
5285 2017-09-20 Richard Biener <rguenther@suse.de>
5287 PR tree-optimization/81373
5288 * graphite-scop-detection.c (build_cross_bb_scalars_def):
5289 Force SESE live-out defs to be handled even if they are
5292 2017-09-19 Jeff Law <law@redhat.com>
5294 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
5295 nothing for stack adjustments with REG_STACK_CHECK.
5296 * sched-deps.c (parse_add_or_inc): Reject insns with
5297 REG_STACK_CHECK from dependency breaking.
5298 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
5299 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
5300 * reg-notes.def (STACK_CHECK): New note.
5302 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
5303 (ix86_expand_prologue): Dump stack clash info as needed.
5304 Call ix86_adjust_stack_and_probe_stack_clash as needed.
5306 * function.c (dump_stack_clash_frame_info): New function.
5307 * function.h (dump_stack_clash_frame_info): Prototype.
5308 (enum stack_clash_probes): New enum.
5310 * config/alpha/alpha.c (alpha_expand_prologue): Also check
5311 flag_stack_clash_protection.
5312 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
5313 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
5314 (arm_frame_pointer_required): Likewise.
5315 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
5316 (ia64_expand_prologue): Likewise.
5317 * config/mips/mips.c (mips_expand_prologue): Likewise.
5318 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
5319 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
5320 (sparc_flat_expand_prologue): Likewise.
5321 * config/spu/spu.c (spu_expand_prologue): Likewise.
5323 * explow.c: Include "params.h".
5324 (anti_adjust_stack_and_probe_stack_clash): New function.
5325 (get_stack_check_protect): Likewise.
5326 (compute_stack_clash_protection_loop_data): Likewise.
5327 (emit_stack_clash_protection_loop_start): Likewise.
5328 (emit_stack_clash_protection_loop_end): Likewise.
5329 (allocate_dynamic_stack_space): Use get_stack_check_protect.
5330 Use anti_adjust_stack_and_probe_stack_clash.
5331 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
5332 (emit_stack_clash_protection_loop_start): Likewise.
5333 (emit_stack_clash_protection_loop_end): Likewise.
5334 * rtl.h (get_stack_check_protect): Prototype.
5335 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
5336 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
5337 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
5339 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
5341 * doc/tm.texi: Rebuilt.
5342 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
5343 get_stack_check_protect.
5344 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
5345 * config/arm/arm.c (arm_expand_prologue): Likewise.
5346 (arm_frame_pointer_required): Likewise.
5347 * config/i386/i386.c (ix86_expand_prologue): Likewise.
5348 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
5349 * config/mips/mips.c (mips_expand_prologue): Likewise.
5350 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
5351 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
5352 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
5353 (sparc_flat_expand_prologue): Likewise.
5355 * common.opt (-fstack-clash-protection): New option.
5356 * flag-types.h (enum stack_check_type): Note difference between
5357 -fstack-check= and -fstack-clash-protection.
5358 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
5359 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
5360 * toplev.c (process_options): Issue warnings/errors for cases
5361 not handled with -fstack-clash-protection.
5362 * doc/invoke.texi (-fstack-clash-protection): Document new option.
5363 (-fstack-check): Note additional problem with -fstack-check=generic.
5364 Note that -fstack-check is primarily for Ada and refer users
5365 to -fstack-clash-protection for stack-clash-protection.
5366 Document new params for stack clash protection.
5368 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
5370 * config/i386/i386.c (ix86_split_long_move): Do not handle
5371 address used for LEA in a special way.
5373 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
5375 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
5377 2017-09-19 Martin Sebor <msebor@redhat.com>
5380 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
5381 of incompatible types.
5383 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
5385 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5386 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
5387 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5388 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
5390 2017-09-19 Richard Biener <rguenther@suse.de>
5392 PR tree-optimization/82244
5393 * tree-vrp.c (remove_range_assertions): Do not propagate
5394 a constant to abnormals but replace the assert with a copy.
5396 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
5398 PR rtl-optimization/57878
5399 PR rtl-optimization/68988
5400 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
5401 avoidance test involving non_reload_pseudos. Move frequency test
5402 below the general fragmentation avoidance test.
5404 2017-09-19 Richard Biener <rguenther@suse.de>
5406 PR tree-optimization/69728
5407 * graphite-sese-to-poly.c (schedule_error): New global.
5408 (add_loop_schedule): Handle empty domain by failing the
5410 (build_original_schedule): Handle schedule_error.
5412 2017-09-19 Richard Biener <rguenther@suse.de>
5414 * graphite-scop-detection.c (scop_detection::can_represent_loop):
5415 Do not iterate to sibling loops but only to siblings of inner
5418 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
5421 * config/m68k/m68k.md (moveq feeding equality comparison): Check
5422 that the registers are different.
5424 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
5426 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
5427 to processor_model and "amdfam17h" to arch_names_table.
5428 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
5430 2017-09-18 Jakub Jelinek <jakub@redhat.com>
5433 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
5435 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5436 Alan Hayward <alan.hayward@arm.com>
5437 David Sherwood <david.sherwood@arm.com>
5439 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
5441 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
5443 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
5444 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
5445 vect_schedule_slp_instance.
5446 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
5447 Update call to vect_slp_analyze_node_operations. Simplify return
5449 (vect_slp_analyze_bb_1): Update call accordingly.
5450 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
5451 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
5452 (vect_schedule_slp): Update call accordingly.
5454 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5455 Alan Hayward <alan.hayward@arm.com>
5456 David Sherwood <david.sherwood@arm.com>
5458 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
5459 with types that aren't in fact scalar.
5461 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5463 * tree-vect-slp.c (vect_record_max_nunits): New function,
5465 (vect_build_slp_tree_1): ...here.
5466 (vect_build_slp_tree_2): Call it for phis too.
5468 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5470 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
5471 to vect_get_vec_def_for_operand when getting the mask operand.
5473 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5474 Alan Hayward <alan.hayward@arm.com>
5475 David Sherwood <david.sherwood@arm.com>
5477 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
5480 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5481 Alan Hayward <alan.hayward@arm.com>
5482 David Sherwood <david.sherwood@arm.com>
5484 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
5485 calculation for vector booleans.
5487 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5488 Alan Hayward <alan.hayward@arm.com>
5489 David Sherwood <david.sherwood@arm.com>
5491 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
5493 (vect_transform_stmt): ...here.
5494 (vect_analyze_stmt): Use it instead of calling
5495 vectorizable_live_operation directly.
5497 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
5499 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
5500 non-SIMT targets in acc vector loops.
5502 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
5504 * configure.ac: Add arc and check if assembler supports gdwarf2.
5505 * configure: Regenerate.
5507 2017-09-18 Richard Biener <rguenther@suse.de>
5509 PR tree-optimization/82220
5510 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
5511 epilogue niters from the min_profitable_iters compute.
5513 2017-09-18 Jakub Jelinek <jakub@redhat.com>
5516 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
5517 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
5518 (ix86_init_pic_reg): Revert 2017-09-01 changes.
5520 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
5523 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
5524 switching to a new text section.
5526 2017-09-18 Richard Biener <rguenther@suse.de>
5528 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
5530 (build_alias_set): Reject aliases with no access function.
5532 2017-09-18 Richard Biener <rguenther@suse.de>
5534 PR tree-optimization/79622
5535 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
5537 (build_cross_bb_scalars_use): Likewise.
5539 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
5541 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
5543 2017-09-18 Alan Modra <amodra@gmail.com>
5546 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
5547 stack_pointer_rtx for count 0. Update comments. Break up
5548 large rtl expression.
5550 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
5552 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
5553 Increase to 20 bytes.
5554 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
5555 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
5556 or avx version of the stub.
5558 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
5561 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
5562 compute the minimum stack alignment. Also update preferred stack
5563 boundary for leaf functions.
5565 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
5567 PR tree-optimization/82228
5568 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
5571 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5573 * common/config/nds32/nds32-common.c
5574 (nds32_option_optimization_table): Refine formatting.
5575 (nds32_option_optimization_table): Use -fsched-pressure and
5576 -fomit-frame-pointer for specific optimization level.
5578 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5580 * config/nds32/nds32.c: Refine formatting and comments.
5581 * config/nds32/nds32.h: Likewise.
5582 * config/nds32/nds32.md: Likewise.
5583 * config/nds32/nds32-cost.c: Likewise.
5584 * config/nds32/nds32-isr.c: Likewise.
5585 * config/nds32/nds32-md-auxiliary.c: Likewise.
5586 * config/nds32/nds32-multiple.md: Likewise.
5587 * config/nds32/nds32-predicates.c: Likewise.
5589 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
5590 Jakub Jelinek <jakub@redhat.com>
5592 Add support for -std=c++2a.
5593 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
5595 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
5597 2017-09-15 Steve Ellcey <sellcey@cavium.com>
5600 * doc/extend.texi (Common Function Attributes): Add
5601 references to ARM, AArch64, and S/390 specific attributes.
5602 (Function Specific Option Pragmas): Add AArch64 and S/390
5603 to list of back ends that support the target pragma.
5605 2017-09-15 Nathan Sidwell <nathan@acm.org>
5607 * doc/standards.texi: Fix C++17 description. Update URLs for
5610 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
5612 * common.opt (Wcast-align=strict): New warning option.
5613 * doc/invoke.texi: Document -Wcast-align=strict.
5615 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
5617 * cgraph.h (cgraph_thunk_info): Add comments.
5618 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
5619 assert for VIRTUAL_* arguments stricter.
5621 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
5623 PR tree-optimization/71026
5624 * match.pd: Move RDIV patterns from fold-const.c
5625 * fold-const.c (distribute_real_division): Removed.
5626 (fold_binary_loc): Remove calls to distribute_real_divison.
5628 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5630 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
5631 c++1z and gnu++1z as deprecated. Change other references to
5632 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
5633 Change -Wc++1z-compat to -Wc++17-compat.
5634 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
5635 * dwarf2out.c (highest_c_language): Handle C++17.
5636 (gen_compile_unit_die): Likewise.
5638 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5640 PR rtl-optimization/82192
5641 * combine.c (make_extraction): Don't look through non-paradoxical
5642 SUBREGs or TRUNCATE if pos + len is or might be bigger than
5645 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5646 Alan Hayward <alan.hayward@arm.com>
5647 David Sherwood <david.sherwood@arm.com>
5649 * target.def (function_arg_offset): New hook.
5650 * targhooks.h (default_function_arg_offset): Declare.
5651 * targhooks.c (default_function_arg_offset): New function.
5652 * function.c (locate_and_pad_parm): Use
5653 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
5654 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
5655 (TARGET_FUNCTION_ARG_OFFSET): ...this.
5656 * doc/tm.texi: Regenerate.
5657 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
5658 * config/spu/spu.c (spu_function_arg_offset): New function.
5659 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
5660 * system.h (FUNCTION_ARG_OFFSET): Poison.
5662 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5663 Alan Hayard <alan.hayward@arm.com>
5664 David Sherwood <david.sherwood@arm.com>
5666 * target.def (truly_noop_truncation): New hook.
5667 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
5668 than TRULY_NOOP_TRUNCATION.
5669 * hooks.h (hook_bool_uint_uint_true): Declare.
5670 * hooks.c (hook_bool_uint_uint_true): New function.
5671 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
5672 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
5673 * doc/tm.texi: Regenerate.
5674 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
5675 rather than TRULY_NOOP_TRUNCATION in comments.
5676 (simplify_comparison): Likewise.
5677 (record_truncated_value): Likewise.
5678 * expmed.c (extract_bit_field_1): Likewise.
5679 (extract_split_bit_field): Likewise.
5680 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
5681 instead of TRULY_NOOP_TRUNCATION.
5682 * function.c (assign_parm_setup_block): Likewise.
5683 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
5684 * rtlhooks.c: Include target.h.
5685 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
5686 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
5687 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
5688 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
5689 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
5690 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
5691 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
5692 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
5693 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
5694 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
5695 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
5696 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
5697 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
5698 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
5699 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
5700 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
5701 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
5702 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
5703 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
5704 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
5705 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
5706 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
5707 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
5708 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
5709 * config/mips/mips.c (mips_truly_noop_truncation): New function.
5710 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5711 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5712 rather than TRULY_NOOP_TRUNCATION in comments.
5713 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
5714 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
5715 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
5716 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
5717 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
5718 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
5719 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
5720 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
5721 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
5722 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
5723 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
5724 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5725 rather than TRULY_NOOP_TRUNCATION in comments.
5726 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
5727 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
5728 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
5729 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
5730 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
5731 TRULY_NOOP_TRUNCATION condition.
5732 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
5733 (TRULY_NOOP_TRUNCATION): Delete.
5734 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
5735 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
5736 * config/spu/spu.c (spu_truly_noop_truncation): New function.
5737 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5738 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
5739 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
5740 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
5741 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5742 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5743 rather than TRULY_NOOP_TRUNCATION in comments.
5744 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
5745 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
5746 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
5747 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
5748 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
5749 * system.h (TRULY_NOOP_TRUNCATION): Poison.
5751 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
5754 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
5755 (*cmp_ior): Likewise.
5756 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
5757 (*ior_scc_scc_cmp): Likewise.
5758 (*and_scc_scc): Likewise.
5759 (*and_scc_scc_cmp): Likewise.
5761 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5762 Alan Hayard <alan.hayward@arm.com>
5763 David Sherwood <david.sherwood@arm.com>
5765 * target.def (can_change_mode_class): New hook.
5766 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
5767 (hard_regno_nregs): Likewise.
5768 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
5769 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
5770 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
5771 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
5772 (LOAD_EXTEND_OP): Update accordingly.
5773 * doc/tm.texi: Regenerate.
5774 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
5775 CANNOT_CHANGE_MODE_CLASS.
5776 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
5777 (REG_CAN_CHANGE_MODE_P): ...this new macro.
5778 * combine.c (simplify_set): Update accordingly.
5779 * emit-rtl.c (validate_subreg): Likewise.
5780 * recog.c (general_operand): Likewise.
5781 * regcprop.c (mode_change_ok): Likewise.
5782 * reload1.c (choose_reload_regs): Likewise.
5783 (inherit_piecemeal_p): Likewise.
5784 * rtlanal.c (simplify_subreg_regno): Likewise.
5785 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
5786 instead of CANNOT_CHANGE_MODE_CLASS.
5787 (reload_cse_simplify_operands): Likewise.
5788 * reload.c (push_reload): Use targetm.can_change_mode_class
5789 instead of CANNOT_CHANGE_MODE_CLASS.
5790 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
5791 REG_CANNOT_CHANGE_MODE_P.
5792 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5793 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
5794 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5795 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5796 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5797 (arm_can_change_mode_class): New function.
5798 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
5799 than CANNOT_CHANGE_MODE_CLASS in comments.
5800 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5801 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
5802 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
5803 (ix86_can_change_mode_class): ...this new function, inverting the
5804 sense of the return value.
5805 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5806 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5807 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5808 (ia64_can_change_mode_class): New function.
5809 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5810 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
5811 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
5812 (m32c_can_change_mode_class): ...this new function, inverting the
5813 sense of the return value.
5814 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5815 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5816 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
5817 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
5818 (mips_can_change_mode_class): ...this new function, inverting the
5819 sense of the return value.
5820 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5821 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5822 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5823 (msp430_can_change_mode_class): New function.
5824 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5825 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
5826 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5827 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5828 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5829 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
5830 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5831 (pa_cannot_change_mode_class): Replace with...
5832 (pa_can_change_mode_class): ...this new function, inverting the
5833 sense of the return value.
5834 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
5835 than CANNOT_CHANGE_MODE_CLASS in comments.
5836 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5837 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
5838 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5839 (pdp11_cannot_change_mode_class): Replace with...
5840 (pdp11_can_change_mode_class): ...this new function, inverting the
5841 sense of the return value.
5842 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5843 * config/powerpcspe/powerpcspe-protos.h
5844 (rs6000_cannot_change_mode_class_ptr): Delete.
5845 * config/powerpcspe/powerpcspe.c
5846 (rs6000_cannot_change_mode_class_ptr): Delete.
5847 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5848 (rs6000_option_override_internal): Assign to
5849 targetm.can_change_mode_class instead of
5850 rs6000_cannot_change_mode_class_ptr.
5851 (rs6000_cannot_change_mode_class): Replace with...
5852 (rs6000_can_change_mode_class): ...this new function, inverting the
5853 sense of the return value.
5854 (rs6000_debug_cannot_change_mode_class): Replace with...
5855 (rs6000_debug_can_change_mode_class): ...this new function.
5856 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5857 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
5858 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5859 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5860 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
5862 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
5863 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5864 (rs6000_option_override_internal): Assign to
5865 targetm.can_change_mode_class instead of
5866 rs6000_cannot_change_mode_class_ptr.
5867 (rs6000_cannot_change_mode_class): Replace with...
5868 (rs6000_can_change_mode_class): ...this new function, inverting the
5869 sense of the return value.
5870 (rs6000_debug_cannot_change_mode_class): Replace with...
5871 (rs6000_debug_can_change_mode_class): ...this new function.
5872 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5873 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
5874 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
5875 (s390_can_change_mode_class): ...this new function, inverting the
5876 sense of the return value.
5877 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5878 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5879 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
5880 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5881 (sh_cannot_change_mode_class): Replace with...
5882 (sh_can_change_mode_class): ...this new function, inverting the
5883 sense of the return value.
5884 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5885 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5886 (sparc_can_change_mode_class): New function.
5887 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5888 * config/spu/spu.c (spu_can_change_mode_class): New function.
5889 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5890 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5891 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5892 (visium_can_change_mode_class): New function.
5893 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
5895 2017-09-15 Richard Biener <rguenther@suse.de>
5897 PR tree-optimization/82217
5898 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
5899 but not undefined case.
5901 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5904 * postreload.c (reload_cse_simplify_operands): Skip
5905 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
5907 2017-09-15 Richard Biener <rguenther@suse.de>
5909 PR tree-optimization/68823
5910 * graphite-scop-detection.c (build_alias_set): If we have a
5911 possible dependence check whether we can handle them by just
5912 looking at the DRs DR_ACCESS_FNs.
5913 (build_scops): If build_alias_set fails, fail the SCOP.
5915 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5917 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
5918 to support float128 built-in functions that require the ISA 3.0
5920 (BU_FLOAT128_3_HW): Likewise.
5921 (SQRTF128): Add support for the IEEE 128-bit square root and fma
5923 (FMAF128): Likewise.
5925 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
5926 support for built-in functions that need the ISA 3.0 IEEE 128-bit
5927 floating point instructions.
5928 (rs6000_invalid_builtin): Likewise.
5929 (rs6000_builtin_mask_names): Likewise.
5930 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
5931 (RS6000_BTM_FLOAT128_HW): Likewise.
5932 (RS6000_BTM_COMMON): Likewise.
5933 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
5935 * doc/extend.texi (RS/6000 built-in functions): Document the
5936 IEEE 128-bit floating point square root and fused multiply-add
5939 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
5941 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
5942 reg (r2) isn't in the set of registers defined in the prologue.
5944 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5945 Alan Hayward <alan.hayward@arm.com>
5946 David Sherwood <david.sherwood@arm.com>
5948 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
5949 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
5950 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
5951 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
5952 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
5954 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
5955 max_vectorization_factor.
5956 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
5958 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5959 Alan Hayward <alan.hayward@arm.com>
5960 David Sherwood <david.sherwood@arm.com>
5962 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
5963 (vect_worthwhile_without_simd_p): Declare.
5964 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
5965 (vectorizable_reduction): Use it.
5966 * tree-vect-stmts.c (vectorizable_shift): Likewise.
5967 (vectorizable_operation): Likewise.
5969 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5970 Alan Hayward <alan.hayward@arm.com>
5971 David Sherwood <david.sherwood@arm.com>
5973 * tree-vectorizer.h (vect_get_num_copies): New function.
5974 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
5975 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5976 (vectorizable_induction): Likewise.
5977 (vectorizable_live_operation): Likewise.
5978 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5979 (vectorizable_bswap): Likewise.
5980 (vectorizable_call): Likewise.
5981 (vectorizable_conversion): Likewise.
5982 (vectorizable_assignment): Likewise.
5983 (vectorizable_shift): Likewise.
5984 (vectorizable_operation): Likewise.
5985 (vectorizable_store): Likewise.
5986 (vectorizable_load): Likewise.
5987 (vectorizable_condition): Likewise.
5988 (vectorizable_comparison): Likewise.
5989 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
5991 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5992 Alan Hayward <alan.hayward@arm.com>
5993 David Sherwood <david.sherwood@arm.com>
5995 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
5996 of vect_init_vector.
5998 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5999 Alan Hayward <alan.hayward@arm.com>
6000 David Sherwood <david.sherwood@arm.com>
6002 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
6003 an inline wrapper that provides a location.
6004 (gimple_build_vector): Likewise.
6005 * gimple-fold.c (gimple_build_vector_from_val): New function.
6006 (gimple_build_vector): Likewise.
6007 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
6008 functions to build the initial value. Always return a gimple value.
6009 (get_initial_defs_for_reduction): Likewise. Only compute
6011 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
6012 vect_init_vector on the results from get_initial_def(s)_for_reduction.
6013 (vectorizable_induction): Use gimple_build_vector rather than
6016 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6017 Alan Hayward <alan.hayward@arm.com>
6018 David Sherwood <david.sherwood@arm.com>
6020 * target.h (vec_perm_indices): New typedef.
6021 (auto_vec_perm_indices): Likewise.
6022 * optabs-query.h: Include target.h
6023 (can_vec_perm_p): Take a vec_perm_indices *.
6024 * optabs-query.c (can_vec_perm_p): Likewise.
6025 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
6026 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6027 * tree-vect-generic.c (lower_vec_perm): Likewise.
6028 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
6029 (vect_grouped_load_supported): Likewise.
6030 (vect_shift_permute_load_chain): Likewise.
6031 (vect_permute_store_chain): Use auto_vec_perm_indices.
6032 (vect_permute_load_chain): Likewise.
6033 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
6034 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
6035 Update uses of can_vec_perm_p.
6036 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
6037 mode with a number of elements. Take a vec_perm_indices *.
6038 (vect_create_epilog_for_reduction): Update accordingly.
6039 Use auto_vec_perm_indices.
6040 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
6041 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
6042 (vect_transform_slp_perm_load): Likewise.
6043 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
6044 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
6045 (vect_gen_perm_mask_checked): Likewise.
6046 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
6047 (vect_gen_perm_mask_checked): Likewise.
6048 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
6049 (vectorizable_store): Likewise.
6050 (vectorizable_load): Likewise.
6051 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
6052 (vectorizable_bswap): Likewise.
6054 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6055 Alan Hayward <alan.hayward@arm.com>
6056 David Sherwood <david.sherwood@arm.com>
6058 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
6059 * tree.c (build_vector): Likewise.
6060 (build_vector_from_ctor): Update accordingly.
6061 (build_vector_from_val): Likewise.
6062 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
6063 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6064 * tree-vect-generic.c (add_rshift): Likewise.
6065 (expand_vector_divmod): Likewise.
6066 (optimize_vector_constructor): Likewise.
6067 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
6068 (vect_transform_slp_perm_load): Likewise.
6069 (vect_schedule_slp_instance): Likewise.
6070 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
6071 (vectorizable_call): Likewise.
6072 (vect_gen_perm_mask_any): Likewise. Add elements in order.
6073 * expmed.c (make_tree): Likewise.
6074 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
6075 a vector passed to build_vector.
6076 (fold_convert_const): Likewise.
6077 (exact_inverse): Likewise.
6078 (fold_ternary_loc): Likewise.
6079 (fold_relational_const): Likewise.
6080 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
6081 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
6082 (const_unop): Likewise. Store the reduction accumulator in a
6083 variable rather than an array.
6084 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
6085 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
6086 the new vector, rather than constructing it after the input arrays.
6087 (native_interpret_vector): Use auto_vec<tree> when building
6088 a vector passed to build_vector. Add elements in order.
6089 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
6090 auto_vec<tree> when building a vector passed to build_vector.
6091 (vect_create_epilog_for_reduction): Likewise.
6092 (vectorizable_induction): Likewise.
6093 (get_initial_def_for_reduction): Likewise. Fix indentation of
6095 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
6097 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
6098 passed to build_vector.
6100 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6101 Alan Hayward <alan.hayward@arm.com>
6102 David Sherwood <david.sherwood@arm.com>
6104 * tree-core.h (tree_base::u): Add an "nelts" field.
6105 (tree_vector): Use VECTOR_CST_NELTS as the length.
6106 * tree.c (tree_size): Likewise.
6107 (make_vector): Initialize VECTOR_CST_NELTS.
6108 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
6109 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
6110 TYPE_VECTOR_SUBPARTS.
6111 * expr.c (const_vector_mask_from_tree): Consistently use "units"
6112 as the number of units, setting it from VECTOR_CST_NELTS.
6113 (const_vector_from_tree): Likewise.
6114 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
6115 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
6116 (fold_negate_expr_1): Likewise.
6117 (fold_convert_const): Likewise.
6118 (const_binop): Likewise. Differentiate the number of output and
6120 (const_unop): Likewise.
6121 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
6122 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
6123 in cases that did the opposite.
6125 2017-09-14 Richard Biener <rguenther@suse.de>
6127 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
6130 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
6132 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
6134 2017-09-14 Jakub Jelinek <jakub@redhat.com>
6137 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
6138 if and where to split a bb, except for splitting before debug insn
6139 sequences followed by non-label real insn. Delete debug insns
6140 in between basic blocks.
6142 * combine.c (make_compound_operation_int): Formatting fixes.
6144 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
6145 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6146 * config/netbsd.h (LINK_EH_SPEC): Likewise.
6147 * config/sol2.h (LINK_EH_SPEC): Likewise.
6148 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6149 * config/s390/linux.h (LINK_SPEC): Likewise.
6150 * config/freebsd.h (LINK_EH_SPEC): Likewise.
6151 * config/openbsd.h (LINK_EH_SPEC): Likewise.
6152 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6153 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
6154 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
6155 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6156 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
6157 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
6159 2017-09-13 Jakub Jelinek <jakub@redhat.com>
6161 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
6163 (ENDFILE_LINUX_SPEC): Likewise.
6164 (LINK_EH_SPEC): Likewise.
6165 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
6166 (LINK_OS_LINUX_SPEC32): Likewise.
6167 (LINK_OS_LINUX_SPEC64): Likewise.
6168 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
6169 (LINK_OS_LINUX_SPEC): Likewise.
6171 2017-09-13 Martin Liska <mliska@suse.cz>
6174 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
6175 CASE_HIGH is NULL_TREE.
6177 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6178 Alan Hayward <alan.hayward@arm.com>
6179 David Sherwood <david.sherwood@arm.com>
6181 * target.def (secondary_memory_needed): New hook.
6182 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
6183 instead of SECONDARY_MEMORY_NEEDED.
6184 (secondary_memory_needed_mode): Likewise.
6185 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
6186 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
6187 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
6188 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
6189 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
6190 * doc/tm.texi: Regenerate.
6191 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
6192 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
6193 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6194 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
6195 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
6196 * config/i386/i386.c (inline_secondary_memory_needed): Put the
6197 mode argument first and change the reg_class arguments to reg_class_t.
6198 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
6199 Make static. Update the call to inline_secondary_memory_needed.
6200 (ix86_register_move_cost): Update the call to
6201 inline_secondary_memory_needed.
6202 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6203 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
6205 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
6206 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6208 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
6209 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
6210 * config/mips/mips.c (mips_secondary_memory_needed): Make static
6211 and match hook interface. Add comment from mips.h.
6212 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6213 * config/mmix/mmix.md (truncdfsf2): Refer to
6214 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6216 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
6217 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
6218 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6219 (pa_secondary_memory_needed): New function.
6220 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
6221 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
6222 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6223 (pdp11_secondary_memory_needed): Make static and match hook interface.
6224 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
6225 * config/powerpcspe/powerpcspe-protos.h
6226 (rs6000_secondary_memory_needed_ptr): Delete.
6227 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
6229 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6230 (rs6000_option_override_internal): Assign to
6231 targetm.secondary_memory_needed rather than
6232 rs6000_secondary_memory_needed_ptr.
6233 (rs6000_secondary_memory_needed): Match hook interface.
6234 (rs6000_debug_secondary_memory_needed): Likewise.
6235 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
6236 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
6237 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
6238 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6239 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
6240 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
6242 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
6243 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6244 (rs6000_option_override_internal): Assign to
6245 targetm.secondary_memory_needed rather than
6246 rs6000_secondary_memory_needed_ptr.
6247 (rs6000_secondary_memory_needed): Match hook interface.
6248 (rs6000_debug_secondary_memory_needed): Likewise.
6249 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
6250 * config/s390/s390.c (s390_secondary_memory_needed): New function.
6251 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6252 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
6253 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6254 (sparc_secondary_memory_needed): New function.
6255 * lra-constraints.c (check_and_process_move): Refer to
6256 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6258 (curr_insn_transform): Likewise.
6259 (process_alt_operands): Use targetm.secondary_memory_needed
6260 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6261 (check_secondary_memory_needed_p): Likewise.
6262 (choose_split_class): Likewise.
6263 * reload.c: Unconditionally include code that was previously
6264 conditional on SECONDARY_MEMORY_NEEDED.
6265 (push_secondary_reload): Use targetm.secondary_memory_needed
6266 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6267 (push_reload): Likewise.
6268 * reload1.c: Unconditionally include code that was previously
6269 conditional on SECONDARY_MEMORY_NEEDED.
6270 (choose_reload_regs): Use targetm.secondary_memory_needed
6271 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6272 (gen_reload): Likewise.
6273 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
6275 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6276 Alan Hayward <alan.hayward@arm.com>
6277 David Sherwood <david.sherwood@arm.com>
6279 * target.def (secondary_memory_needed_mode): New hook:
6280 * targhooks.c (default_secondary_memory_needed_mode): Declare.
6281 * targhooks.h (default_secondary_memory_needed_mode): New function.
6282 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
6283 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
6284 * doc/tm.texi: Regenerate.
6285 * lra-constraints.c (check_and_process_move): Use
6286 targetm.secondary_memory_needed_mode instead of
6287 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
6288 (curr_insn_transform): Likewise.
6289 * reload.c (get_secondary_mem): Likewise.
6290 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6291 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
6293 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6294 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6295 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
6296 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6297 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
6299 * config/powerpcspe/powerpcspe-protos.h
6300 (rs6000_secondary_memory_needed_mode): Delete.
6301 * config/powerpcspe/powerpcspe.c
6302 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6303 (rs6000_secondary_memory_needed_mode): Make static.
6304 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6305 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
6307 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
6309 (rs6000_secondary_memory_needed_mode): Make static.
6310 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6311 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
6312 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6313 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6314 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
6316 (sparc_secondary_memory_needed_mode): New function.
6317 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
6319 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
6321 * config/aarch64/constraints.md (Umq): New constraint.
6322 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
6324 (mov<mode>): Update condition.
6326 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6328 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
6329 when bitposition is the same.
6331 2017-09-13 Richard Biener <rguenther@suse.de>
6333 * dwarf2out.c (output_die_symbol): Remove.
6334 (output_die): Do not output a DIEs symbol.
6336 2017-09-13 Richard Biener <rguenther@suse.de>
6339 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
6340 default-def to avoid breaking iterator update with the weird
6341 interaction with cgraph_update_edges_for_call_stmt_node.
6343 2017-09-13 Richard Biener <rguenther@suse.de>
6345 * tree-cfg.c (verify_gimple_assign_binary): Add verification
6346 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
6347 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
6348 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
6350 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
6352 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
6353 Disable pc relative literal load irrespective of
6354 TARGET_FIX_ERR_A53_84341 for default.
6356 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
6358 * config/sparc/sparc.c (output_return): Output the source location of
6359 the insn in the delay slot, if any.
6360 (output_sibcall): Likewise.
6362 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
6365 * common.opt (-static-pie): New alias.
6366 (shared): Negate static-pie.
6367 (-no-pie): Update help text.
6369 (static-pie): New option.
6370 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
6371 -static-pie support.
6372 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
6373 (LINK_EH_SPEC): Likewise.
6374 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6375 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
6376 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
6377 * gcc.c (LINK_COMMAND_SPEC): Likewise.
6378 (init_gcc_specs): Likewise.
6379 (init_spec): Likewise.
6380 (display_help): Update help message for -pie.
6381 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
6384 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
6386 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
6387 (movdi_aarch64): Likewise.
6388 (movti_aarch64): Likewise.
6390 2017-09-12 Simon Wright <simon@pushface.org>
6393 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
6394 calculation of the minor version, always output as 0.
6396 2017-09-12 Jakub Jelinek <jakub@redhat.com>
6399 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6400 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
6401 on it early, rather than manual conversion late. For
6402 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
6403 instead of performing manual conversion.
6405 2017-09-12 Carl Love <cel@us.ibm.com>
6407 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
6408 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
6410 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6411 VMULOSW): Add definitions.
6412 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6413 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6414 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6415 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6416 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6418 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
6420 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
6422 (movti_aarch64): Likewise.
6423 (movdf_aarch64): Likewise.
6424 (movtf_aarch64): Likewise.
6425 (load_pairdi): Likewise.
6426 (store_pairdi): Likewise.
6427 (load_pairdf): Likewise.
6428 (store_pairdf): Likewise.
6429 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
6430 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
6431 (ldr_got_small_<mode>): Likewise.
6432 (ldr_got_small_28k_<mode>): Likewise.
6433 (ldr_got_tiny): Likewise.
6434 * config/aarch64/iterators.md (ldst_sz): New.
6435 (ldpstp_sz): Likewise.
6436 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
6438 (thunderx_load): Split load_8 to load_16.
6439 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
6441 (thunderx2t99_storepair_basic): Split store_8 to store_16.
6442 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
6443 (xgene1_store_pair): Split store_8 to store_16.
6444 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
6445 (falkor_st_0_st_sd): Split store_8 to store_16.
6447 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
6449 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
6450 and store1/2/3/4 to store_4/8/12/16.
6451 * config/aarch64/aarch64.md: Update for rename.
6452 * config/arm/arm.md: Likewise.: Likewise.
6453 * config/arm/arm.c: Likewise.
6454 * config/arm/thumb1.md: Likewise.
6455 * config/arm/thumb2.md: Likewise.
6456 * config/arm/vfp.md: Likewise.
6457 * config/arm/arm-generic.md: Likewise.
6458 * config/arm/arm1020e.md: Likewise.
6459 * config/arm/arm1026ejs.md: Likewise.
6460 * config/arm/arm1136jfs.md: Likewise.
6461 * config/arm/arm926ejs.md: Likewise.
6462 * config/arm/cortex-a15.md: Likewise.
6463 * config/arm/cortex-a17.md: Likewise.
6464 * config/arm/cortex-a5.md: Likewise.
6465 * config/arm/cortex-a53.md: Likewise.
6466 * config/arm/cortex-a57.md: Likewise.
6467 * config/arm/cortex-a7.md: Likewise.
6468 * config/arm/cortex-a8.md: Likewise.
6469 * config/arm/cortex-a9.md: Likewise.
6470 * config/arm/cortex-m4.md: Likewise.
6471 * config/arm/cortex-m7.md: Likewise.
6472 * config/arm/cortex-r4.md: Likewise.
6473 * config/arm/exynos-m1.md: Likewise.
6474 * config/arm/fa526.md: Likewise.
6475 * config/arm/fa606te.md: Likewise.
6476 * config/arm/fa626te.md: Likewise.
6477 * config/arm/fa726te.md: Likewise.
6478 * config/arm/fmp626.md: Likewise.
6479 * config/arm/iwmmxt.md: Likewise.
6480 * config/arm/ldmstm.md: Likewise.
6481 * config/arm/marvell-pj4.md: Likewise.
6482 * config/arm/xgene1.md: Likewise.
6483 * config/aarch64/thunderx.md: Likewise.
6484 * config/aarch64/thunderx2t99.md: Likewise.
6485 * config/aarch64/falkor.md: Likewise.
6487 2017-09-12 Martin Liska <mliska@suse.cz>
6489 * attribs.c (private_lookup_attribute): New function.
6490 * attribs.h (private_lookup_attribute): Declared here.
6491 (lookup_attribute): Called from this place.
6493 2017-09-12 Richard Biener <rguenther@suse.de>
6495 PR tree-optimization/82157
6496 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
6497 stmts with side-effects.
6499 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6500 Alan Hayward <alan.hayward@arm.com>
6501 David Sherwood <david.sherwood@arm.com>
6503 * target.def (hard_regno_nregs): New hook.
6504 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
6505 * targhooks.h (default_hard_regno_nregs): Declare.
6506 * targhooks.c (default_hard_regno_nregs): New function.
6507 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
6508 (TARGET_HARD_REGNO_NREGS): ...this hook.
6509 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
6510 (CLASS_MAX_NREGS): Likewise.
6511 * doc/tm.texi: Regenerate.
6512 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
6513 instead of HARD_REGNO_NREGS.
6514 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
6515 HARD_REGNO_NREGS in the comment.
6516 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
6517 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
6518 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
6519 Return an unsigned int.
6520 (TARGET_HARD_REGNO_NREGS): Redefine.
6521 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
6522 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
6523 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6524 (arc_hard_regno_nregs): New function.
6525 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
6526 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
6527 (arm_hard_regno_nregs): New function.
6528 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
6529 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
6530 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
6531 (TARGET_HARD_REGNO_NREGS): Redefine.
6532 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
6533 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
6534 (HARD_REGNO_NREGS): Delete.
6535 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
6536 (cr16_hard_regno_nregs): New function.
6537 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
6538 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
6539 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
6540 (cris_hard_regno_nregs): New function.
6541 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
6542 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
6543 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6544 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
6545 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
6546 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
6547 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
6548 (frv_hard_regno_nregs): Make static. Take and return an
6550 (frv_class_max_nregs): Remove outdated copy of documentation.
6551 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
6552 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
6553 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
6554 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
6555 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
6556 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
6557 (TARGET_HARD_REGNO_NREGS): Redefine.
6558 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
6559 (CLASS_MAX_NREGS): Update comment.
6560 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
6561 (ia64_hard_regno_nregs): New function.
6562 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
6563 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
6564 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
6565 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
6566 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
6568 (m32c_hard_regno_nregs): Likewise. Make static.
6569 (TARGET_HARD_REGNO_NREGS): Redefine.
6570 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
6571 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
6572 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
6573 (m68k_hard_regno_nregs): New function.
6574 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
6575 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
6576 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
6577 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
6578 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
6579 Take and return an unsigned int.
6580 (TARGET_HARD_REGNO_NREGS): Redefine.
6581 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
6582 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6583 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
6584 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
6585 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
6586 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
6587 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
6588 (msp430_hard_regno_nregs): Make static. Take and return an
6590 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
6591 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
6592 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
6593 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
6594 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
6595 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
6596 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
6597 (TARGET_HARD_REGNO_NREGS): Redefine.
6598 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
6599 (PA_HARD_REGNO_NREGS): ...this.
6600 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
6601 (PA_HARD_REGNO_NREGS): ...this.
6602 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6603 (pa_hard_regno_nregs): New function.
6604 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
6605 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
6606 (pdp11_hard_regno_nregs): New function.
6607 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
6608 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
6609 (rs6000_hard_regno_nregs_hook): New function.
6610 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
6611 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6612 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
6613 Take and return an unsigned int. Move earlier in file.
6614 (TARGET_HARD_REGNO_NREGS): Redefine.
6615 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
6616 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
6617 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
6618 (rl78_hard_regno_nregs): Make static. Take and return an
6620 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
6621 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
6622 (rs6000_hard_regno_nregs_hook): New function.
6623 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
6624 * config/rx/rx.c (rx_hard_regno_nregs): New function.
6625 (TARGET_HARD_REGNO_NREGS): Redefine.
6626 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
6627 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
6628 instead of HARD_REGNO_NREGS.
6629 (s390_hard_regno_nregs): New function.
6630 (s390_hard_regno_mode_ok): Add comment from s390.h.
6631 (TARGET_HARD_REGNO_NREGS): Redefine.
6632 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
6633 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
6634 (sh_hard_regno_nregs): New function.
6635 (sh_pass_in_reg_p): Use it.
6636 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
6637 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6638 (sparc_hard_regno_nregs): New function.
6639 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
6640 * config/spu/spu.c (spu_hard_regno_nregs): New function.
6641 (spu_function_arg_advance): Use it, supplying a valid register number.
6642 (TARGET_HARD_REGNO_NREGS): Redefine.
6643 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
6644 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
6645 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
6646 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
6647 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
6648 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
6649 (CLASS_MAX_NREGS): Remove copy of old documentation.
6650 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
6651 (visium_hard_regno_nregs): New function.
6652 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
6653 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
6654 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6655 xtensa_hard_regno_nregs): New function.
6656 * system.h (HARD_REGNO_NREGS): Poison.
6658 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6660 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
6661 hard_regno_nregs instead of HARD_REGNO_NREGS.
6662 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
6663 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
6664 (c6x_expand_epilogue): Likewise.
6665 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
6666 (frv_read_iacc_argument): Likewise.
6667 * config/sh/sh.c: Include regs.h.
6668 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
6669 (regs_used): Likewise.
6670 (output_stack_adjust): Likewise.
6671 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
6672 * expmed.c: Include regs.h.
6673 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
6674 * ree.c: Include regs.h.
6675 (combine_reaching_defs): Use hard_regno_nregs instead of
6677 (add_removable_extension): Likewise.
6679 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6681 * regs.h (hard_regno_nregs): Turn into a function.
6682 (end_hard_regno): Update accordingly.
6683 * caller-save.c (setup_save_areas): Likewise.
6684 (save_call_clobbered_regs): Likewise.
6685 (replace_reg_with_saved_mem): Likewise.
6686 (insert_restore): Likewise.
6687 (insert_save): Likewise.
6688 * combine.c (can_change_dest_mode): Likewise.
6689 (move_deaths): Likewise.
6690 (distribute_notes): Likewise.
6691 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
6692 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
6693 (rs6000_split_multireg_move): Likewise.
6694 (rs6000_register_move_cost): Likewise.
6695 (rs6000_memory_move_cost): Likewise.
6696 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
6697 (rs6000_split_multireg_move): Likewise.
6698 (rs6000_register_move_cost): Likewise.
6699 (rs6000_memory_move_cost): Likewise.
6700 * cselib.c (cselib_reset_table): Likewise.
6701 (cselib_lookup_1): Likewise.
6702 * emit-rtl.c (set_mode_and_regno): Likewise.
6703 * function.c (aggregate_value_p): Likewise.
6704 * ira-color.c (setup_profitable_hard_regs): Likewise.
6705 (check_hard_reg_p): Likewise.
6706 (calculate_saved_nregs): Likewise.
6707 (assign_hard_reg): Likewise.
6708 (improve_allocation): Likewise.
6709 (calculate_spill_cost): Likewise.
6710 * ira-emit.c (modify_move_list): Likewise.
6711 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
6712 (ira_hard_reg_in_set_p): Likewise.
6713 * ira.c (setup_reg_mode_hard_regset): Likewise.
6714 (clarify_prohibited_class_mode_regs): Likewise.
6715 (check_allocation): Likewise.
6716 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6717 (lra_setup_reg_renumber): Likewise.
6718 (setup_try_hard_regno_pseudos): Likewise.
6719 (spill_for): Likewise.
6720 (assign_hard_regno): Likewise.
6721 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
6722 * lra-constraints.c (in_class_p): Likewise.
6723 (lra_constraint_offset): Likewise.
6724 (simplify_operand_subreg): Likewise.
6725 (lra_constraints): Likewise.
6726 (split_reg): Likewise.
6727 (split_if_necessary): Likewise.
6728 (invariant_p): Likewise.
6729 (inherit_in_ebb): Likewise.
6730 * lra-lives.c (process_bb_lives): Likewise.
6731 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
6732 (get_hard_regs): Likewise.
6733 (do_remat): Likewise.
6734 * lra-spills.c (assign_spill_hard_regs): Likewise.
6735 * mode-switching.c (create_pre_exit): Likewise.
6736 * postreload.c (reload_combine_recognize_pattern): Likewise.
6737 * recog.c (peep2_find_free_register): Likewise.
6738 * regcprop.c (kill_value_regno): Likewise.
6739 (set_value_regno): Likewise.
6740 (copy_value): Likewise.
6741 (maybe_mode_change): Likewise.
6742 (find_oldest_value_reg): Likewise.
6743 (copyprop_hardreg_forward_1): Likewise.
6744 * regrename.c (check_new_reg_p): Likewise.
6745 (regrename_do_replace): Likewise.
6746 * reload.c (push_reload): Likewise.
6747 (combine_reloads): Likewise.
6748 (find_dummy_reload): Likewise.
6749 (operands_match_p): Likewise.
6750 (find_reloads): Likewise.
6751 (find_equiv_reg): Likewise.
6752 (reload_adjust_reg_for_mode): Likewise.
6753 * reload1.c (count_pseudo): Likewise.
6754 (count_spilled_pseudo): Likewise.
6755 (find_reg): Likewise.
6756 (clear_reload_reg_in_use): Likewise.
6757 (free_for_value_p): Likewise.
6758 (allocate_reload_reg): Likewise.
6759 (choose_reload_regs): Likewise.
6760 (reload_adjust_reg_for_temp): Likewise.
6761 (emit_reload_insns): Likewise.
6762 (delete_output_reload): Likewise.
6763 * rtlanal.c (subreg_get_info): Likewise.
6764 * sched-deps.c (sched_analyze_reg): Likewise.
6765 * sel-sched.c (init_regs_for_mode): Likewise.
6766 (mark_unavailable_hard_regs): Likewise.
6767 (choose_best_reg_1): Likewise.
6768 (verify_target_availability): Likewise.
6769 * valtrack.c (dead_debug_insert_temp): Likewise.
6770 * var-tracking.c (track_loc_p): Likewise.
6771 (emit_note_insn_var_location): Likewise.
6772 * varasm.c (make_decl_rtl): Likewise.
6773 * reginfo.c (choose_hard_reg_mode): Likewise.
6774 (init_reg_modes_target): Refer directly to
6775 this_target_regs->x_hard_regno_nregs.
6777 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6779 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
6780 instead of hard_regno_nregs.
6782 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6784 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
6785 end_hard_regno instead of hard_regno_nregs.
6786 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
6787 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
6788 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
6789 * ira-color.c (improve_allocation): Likewise.
6790 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6791 * lra-lives.c (mark_regno_live): Likewise.
6792 (mark_regno_dead): Likewise.
6793 * lra-remat.c (operand_to_remat): Likewise.
6794 * lra.c (collect_non_operand_hard_regs): Likewise.
6795 * postreload.c (reload_combine_note_store): Likewise.
6796 (move2add_valid_value_p): Likewise.
6797 * reload.c (regno_clobbered_p): Likewise.
6799 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6801 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
6803 * config/v850/v850.c (v850_reorg): Likewise.
6804 * reload.c (refers_to_regno_for_reload_p): Likewise.
6805 (find_equiv_reg): Likewise.
6806 * reload1.c (reload_reg_reaches_end_p): Likewise.
6808 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6810 * caller-save.c (add_used_regs): Use REG_NREGS instead of
6812 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
6813 * config/arm/arm.c (output_move_neon): Likewise.
6814 (arm_attr_length_move_neon): Likewise.
6815 (neon_split_vcombine): Likewise.
6816 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
6817 (c6x_mark_reg_written): Likewise.
6818 (c6x_dwarf_register_span): Likewise.
6819 * config/i386/i386.c (ix86_save_reg): Likewise.
6820 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
6821 (rws_access_reg): Likewise.
6822 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
6823 * mode-switching.c (create_pre_exit): Likewise.
6824 * ree.c (combine_reaching_defs): Likewise.
6825 (add_removable_extension): Likewise.
6826 * regcprop.c (find_oldest_value_reg): Likewise.
6827 (copyprop_hardreg_forward_1): Likewise.
6828 * reload.c (reload_inner_reg_of_subreg): Likewise.
6829 (push_reload): Likewise.
6830 (combine_reloads): Likewise.
6831 (find_dummy_reload): Likewise.
6832 (reload_adjust_reg_for_mode): Likewise.
6833 * reload1.c (find_reload_regs): Likewise.
6834 (forget_old_reloads_1): Likewise.
6835 (reload_reg_free_for_value_p): Likewise.
6836 (reload_adjust_reg_for_temp): Likewise.
6837 (emit_reload_insns): Likewise.
6838 (delete_output_reload): Likewise.
6839 * sel-sched.c (choose_best_reg_1): Likewise.
6840 (choose_best_pseudo_reg): Likewise.
6842 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6843 Alan Hayward <alan.hayward@arm.com>
6844 David Sherwood <david.sherwood@arm.com>
6846 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
6847 * target.def (slow_unaligned_access): New hook.
6848 * targhooks.h (default_slow_unaligned_access): Declare.
6849 * targhooks.c (default_slow_unaligned_access): New function.
6850 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
6851 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
6852 * doc/tm.texi: Regenerate.
6853 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
6854 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
6855 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
6857 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
6858 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
6860 (rs6000_slow_unaligned_access): New function.
6861 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
6862 (expand_block_compare): Likewise.
6863 (expand_strn_compare): Likewise.
6864 (rs6000_rtx_costs): Likewise.
6865 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
6866 (riscv_slow_unaligned_access): Likewise.
6867 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
6868 (riscv_slow_unaligned_access_p): ...this and make static.
6869 (riscv_option_override): Update accordingly.
6870 (riscv_slow_unaligned_access): New function.
6871 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
6872 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
6873 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
6874 (rs6000_slow_unaligned_access): New function.
6875 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
6876 (rs6000_rtx_costs): Likewise.
6877 * config/rs6000/rs6000-string.c (expand_block_compare)
6878 (expand_strn_compare): Use targetm.slow_unaligned_access instead
6879 of SLOW_UNALIGNED_ACCESS.
6880 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
6881 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
6882 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
6883 of SLOW_UNALIGNED_ACCESS.
6884 * expmed.c (simple_mem_bitfield_p): Likewise.
6885 * expr.c (alignment_for_piecewise_move): Likewise.
6886 (emit_group_load_1): Likewise.
6887 (emit_group_store): Likewise.
6888 (copy_blkmode_from_reg): Likewise.
6889 (emit_push_insn): Likewise.
6890 (expand_assignment): Likewise.
6891 (store_field): Likewise.
6892 (expand_expr_real_1): Likewise.
6893 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
6894 * lra-constraints.c (simplify_operand_subreg): Likewise.
6895 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6896 * gimple-ssa-store-merging.c: Likewise in block comment at start
6898 * tree-ssa-strlen.c: Include target.h.
6899 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
6900 of SLOW_UNALIGNED_ACCESS.
6901 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
6903 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6905 PR rtl-optimization/82185
6906 * expmed.c (emit_store_flag_int): Only test tem if it has been
6909 2017-09-12 Richard Biener <rguenther@suse.de>
6912 * match.pd ((FTYPE) N CMP CST): Fix typo.
6914 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
6916 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
6918 (mips_near_type_p): Add 'short_call' attribute as a synonym
6920 * doc/extend.texi (short_call): Document new function attribute.
6922 2017-09-12 Jakub Jelinek <jakub@redhat.com>
6925 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
6926 assertion check that in the condition.
6927 (get_atomic_generic_size): Likewise. Before testing if parameter
6928 has pointer type, if it has array type, call for C++
6929 default_conversion to perform array-to-pointer conversion.
6931 2017-09-12 Richard Biener <rguenther@suse.de>
6933 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
6934 for operations we cannot scalarize.
6936 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
6938 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
6939 vectors heap vectors. Clean up comments.
6940 Make visited_bbs a reference.
6941 (profitable_jump_thread_path): Make GC
6942 vectors heap vectors. Clean up comments.
6944 (convert_and_register_jump_thread_path): Make GC vectors heap
6946 (check_subpath_and_update_thread_path): Same. Clean up comments.
6947 Make visited_bbs a reference.
6948 (handle_phi): Abstract common code to to
6949 register_jump_thread_path_if_profitable.
6950 Rename VAR_BB to DEF_BB.
6952 Make GC vectors heap vectors.
6953 Make visited_bbs a reference.
6954 (handle_assignment): Same.
6955 (register_jump_thread_path_if_profitable): New.
6956 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
6958 Make GC vectors heap vectors. Clean up comments.
6959 Make visited_bbs a reference.
6960 (find_jump_threads_backwards): Make visited_bbs live in the stack.
6961 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
6964 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
6967 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
6968 words of E_DImode object are reachable by xtensa_uimm8x4 access.
6970 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
6972 Revert r251800 and r251799.
6974 2017-09-11 Martin Jambor <mjambor@suse.cz>
6977 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
6978 arguments in advance.
6979 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
6980 use it to find predecessor edges.
6981 (naive_outof_ssa): Collect vector of predecessors.
6983 2017-09-08 Jason Merrill <jason@redhat.com>
6985 PR c++/70029 - ICE with ref-qualifier and -flto
6986 * langhooks.h (struct lang_hooks_for_types): Add
6987 copy_lang_qualifiers.
6988 * attribs.c (build_type_attribute_qual_variant): Use it.
6989 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
6991 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
6992 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
6994 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
6997 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
6998 (*mulsi3_sp64): New instruction.
6999 (mulsi3): New expander.
7001 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
7003 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
7005 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7007 * sancov.c: Include memmodel.h.
7009 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
7012 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
7015 2017-09-07 Carl Love <cel@us.ibm.com>
7017 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
7018 the sldi instruction.
7020 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
7022 * sancov.c: Include tm_p.h.
7024 2017-09-07 Jakub Jelinek <jakub@redhat.com>
7027 * output.h (switch_to_other_text_partition): New declaration.
7028 * varasm.c (switch_to_other_text_partition): New function.
7029 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
7030 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
7031 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
7032 to the other text partition before emitting LCL label and switch back
7033 after emitting the word after it.
7035 2017-09-07 Richard Biener <rguenther@suse.de>
7037 * passes.def (pass_split_crit_edges): Remove instance before PRE.
7038 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
7039 critical edges here, after loop init.
7040 (pass_data_pre): Remove PROP_no_crit_edges flags.
7041 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
7042 for valueization of call args to avoid leaking VN_TOP.
7043 (visit_use): Assert we do not visit default defs.
7044 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
7045 Use error_mark_node to more easily detect leaking VN_TOP.
7046 All default-defs are varying, not VN_TOP. Mark them visited.
7047 (run_scc_vn): Make code match comment.
7049 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7051 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
7052 OPTION_MASK_FLOAT128_KEYWORD.
7053 (POWERPC_MASKS): Likewise.
7054 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
7055 support for the -mfloat128-type option, and make -mfloat128
7056 default on PowerPC Linux systems. Define or undefine
7057 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
7058 Define __float128 to be __ieee128 if IEEE 128-bit support is
7059 enabled, or undefine it.
7060 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
7061 Delete defining __FLOAT128_TYPE__.
7062 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
7063 -mfloat128-type option and make -mfloat128 default on PowerPC
7065 (TARGET_FLOAT128_TYPE): Likewise.
7066 (-mfloat128-type): Likewise.
7067 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7068 Delete the -mfloat128-type option and make -mfloat128 default on
7069 PowerPC Linux systems. Always use __ieee128 to be the keyword for
7070 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
7071 128-bit floating point is enabled. Change tests from using
7072 -mfloat128-type to -mfloat128.
7073 (rs6000_mangle_type): Use the correct mangling for the __float128
7074 type even if normal long double is restricted to 64-bits.
7075 (floatn_mode): Enable the _Float128 type by default on VSX Linux
7077 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
7078 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
7079 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
7081 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
7082 documentation for -mfloat128.
7084 2017-09-06 Olivier Hainque <hainque@adacore.com>
7086 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
7088 2017-09-06 Wish Wu <wishwu007@gmail.com>
7089 Jakub Jelinek <jakub@redhat.com>
7091 * asan.c (initialize_sanitizer_builtins): Add
7092 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
7093 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
7094 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
7095 BT_FN_VOID_UINT64_PTR variables.
7096 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
7097 (BT_FN_VOID_UINT16_UINT16): Likewise.
7098 (BT_FN_VOID_UINT32_UINT32): Likewise.
7099 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
7100 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
7101 (BT_FN_VOID_UINT64_PTR): Likewise.
7102 * common.opt (flag_sanitize_coverage): New variable.
7103 (fsanitize-coverage=trace-pc): Remove.
7104 (fsanitize-coverage=): Add.
7105 * flag-types.h (enum sanitize_coverage_code): New enum.
7106 * fold-const.c (fold_range_test): Disable non-short-circuit
7107 optimization if flag_sanitize_coverage.
7108 (fold_truth_andor): Likewise.
7109 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
7110 * opts.c (COVERAGE_SANITIZER_OPT): Define.
7111 (coverage_sanitizer_opts): New array.
7112 (get_closest_sanitizer_option): Add OPTS argument, handle also
7113 OPT_fsanitize_coverage_.
7114 (parse_sanitizer_options): Adjusted to also handle
7115 OPT_fsanitize_coverage_.
7116 (common_handle_option): Add OPT_fsanitize_coverage_.
7117 * sancov.c (instrument_comparison, instrument_switch): New function.
7118 (sancov_pass): Add trace-cmp support.
7119 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
7120 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
7121 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
7122 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
7123 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
7124 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
7125 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
7126 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
7127 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
7128 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
7130 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7132 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7133 error. Only quit immediately if parsing is complete.
7134 (BEGIN): Initialize fatal_err and parse_done.
7135 (begin fpu, end fpu): Check number of arguments.
7136 (begin arch, end arch): Likewise.
7137 (begin cpu, end cpu): Likewise.
7138 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7139 (optalias): Likewise.
7141 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7143 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7144 * config/arm/arm-isa.h: Delete. Move definitions to ...
7145 * arm-cpus.in: ... here. Use new feature and fgroup values.
7146 * config/arm/arm.c (arm_option_override): Use lower case for feature
7148 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7149 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7150 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7152 (print_isa_bits_for): New function.
7153 (gen_isa): New function.
7154 (gen_comm_data): Use print_isa_bits_for.
7155 (define feature): New keyword.
7156 (define fgroup): New keyword.
7157 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
7158 (arm-isa.h): Add rule to generate file.
7159 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7160 case for feature bit names.
7162 2017-09-06 Richard Biener <rguenther@suse.de>
7164 * tree-ssa-pre.c (NECESSARY): Remove.
7165 (create_expression_by_pieces): Do not touch pass-local flags.
7166 (insert_into_preds_of_block): Likewise.
7167 (do_pre_regular_insertion): Likewise.
7168 (eliminate_insert): Likewise.
7169 (eliminate_dom_walker::before_dom_children): Likewise.
7170 (fini_eliminate): Do not look at inserted_exprs.
7171 (mark_operand_necessary): Remove.
7172 (remove_dead_inserted_code): Replace with simple work-list
7173 algorithm based on inserted_exprs and SSA uses.
7174 (pass_pre::execute): Re-order fini_eliminate and
7175 remove_dead_inserted_code.
7177 2017-09-06 Olivier Hainque <hainque@adacore.com>
7179 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
7180 for VxWorks 7. Adjust surrounding comments.
7182 2017-09-06 Richard Biener <rguenther@suse.de>
7184 * gimple-ssa-strength-reduction.c
7185 (find_candidates_dom_walker::before_dom_children): Also allow
7188 2017-09-06 Richard Biener <rguenther@suse.de>
7190 PR tree-optimization/82108
7191 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
7192 for gap in the non-permutation SLP case.
7194 2017-09-06 Martin Jambor <mjambor@suse.cz>
7196 PR tree-optimization/82078
7197 * tree-sra.c (sort_and_splice_var_accesses): Move call to
7198 add_access_to_work_queue...
7199 (build_accesses_from_assign): ...here.
7200 (propagate_all_subaccesses): Make sure racc is the group
7201 representative, if there is one.
7203 2017-09-06 Jakub Jelinek <jakub@redhat.com>
7206 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
7209 2017-09-06 Richard Biener <rguenther@suse.de>
7211 * gimple-ssa-strength-reduction.c
7212 (find_candidates_dom_walker::before_doom_children): Use a
7213 type and not a mode check.
7215 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
7218 * config/arm/predicates.md (arm_general_adddi_operand): Create new
7220 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
7222 2017-09-05 Jeff Law <law@redhat.com>
7224 PR tree-optimization/64910
7225 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
7226 cases where we have 3 or more operands.
7228 2017-09-05 Jakub Jelinek <jakub@redhat.com>
7231 * omp-low.c (lower_omp_for): Recompute tree invariant if
7232 gimple_omp_for_initial/final is ADDR_EXPR.
7235 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
7236 into gimple val before gimplification fo the COND_EXPR.
7238 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
7240 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
7241 REGION_COPY argument.
7242 (thread_through_all_blocks): Remove unused argument to
7243 duplicate_thread_path.
7245 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7246 Alan Hayward <alan.hayward@arm.com>
7247 David Sherwood <david.sherwood@arm.com>
7249 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
7250 Take a scalar_mode rather than a machine_mode.
7251 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7252 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
7253 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7254 (aarch64_gen_adjusted_ldpstp): Likewise.
7255 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
7257 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7258 Alan Hayward <alan.hayward@arm.com>
7259 David Sherwood <david.sherwood@arm.com>
7261 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
7262 Take a scalar_int_mode instead of a machine_mode.
7263 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
7264 (aarch64_output_scalar_simd_mov_immediate): Likewise.
7265 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
7266 (aarch64_simd_attr_length_rglist): Delete.
7267 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
7268 a scalar_int_mode instead of a machine_mode.
7269 (aarch64_add_offset): Likewise.
7270 (aarch64_internal_mov_immediate): Likewise
7271 (aarch64_add_constant_internal): Likewise.
7272 (aarch64_add_constant): Likewise.
7273 (aarch64_movw_imm): Likewise.
7274 (aarch64_rtx_arith_op_extract_p): Likewise.
7275 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
7276 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
7277 Remove assert that the mode isn't a vector.
7278 (aarch64_output_scalar_simd_mov_immediate): Likewise.
7279 (aarch64_expand_mov_immediate): Update calls after above changes.
7280 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
7281 (aarch64_and_bitmask_imm): Check for scalar integer modes.
7282 (aarch64_move_imm): Likewise.
7283 (aarch64_can_const_movi_rtx_p): Likewise.
7284 (aarch64_strip_extend): Likewise.
7285 (aarch64_extr_rtx_p): Likewise.
7286 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
7287 a CONST_INT when the mode parameter is VOIDmode.
7288 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
7290 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7292 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
7293 * stor-layout.c (bitwise_mode_for_mode): Likewise.
7294 (bitwise_type_for_mode): Update accordingly.
7296 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7298 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
7299 * stor-layout.c (mode_for_size_tree): Likewise.
7300 (mode_for_array): Update accordingly.
7301 (layout_decl): Likewise.
7302 (compute_record_mode): Likewise. Only set the mode once.
7304 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7306 * target.def (get_mask_mode): Change return type to opt_mode.
7308 * doc/tm.texi: Regenerate.
7309 * targhooks.h (default_get_mask_mode): Return an opt_mode.
7310 * targhooks.c (default_get_mask_mode): Likewise.
7311 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
7312 * optabs-query.c (can_vec_mask_load_store_p): Update use of
7313 targetm.get_mask_mode.
7314 * tree.c (build_truth_vector_type): Likewise.
7316 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7318 * machmode.h (mode_for_vector): Return an opt_mode.
7319 * stor-layout.c (mode_for_vector): Likewise.
7320 (mode_for_int_vector): Update accordingly.
7321 (layout_type): Likewise.
7322 * config/i386/i386.c (emit_memmov): Likewise.
7323 (ix86_expand_set_or_movmem): Likewise.
7324 (ix86_expand_vector_init): Likewise.
7325 (ix86_get_mask_mode): Likewise.
7326 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
7328 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
7329 * expmed.c (extract_bit_field_1): Likewise.
7330 * expr.c (expand_expr_real_2): Likewise.
7331 * optabs-query.c (can_vec_perm_p): Likewise.
7332 (can_vec_mask_load_store_p): Likewise.
7333 * optabs.c (expand_vec_perm): Likewise.
7334 * targhooks.c (default_get_mask_mode): Likewise.
7335 * tree-vect-stmts.c (vectorizable_store): Likewise.
7336 (vectorizable_load): Likewise.
7337 (get_vectype_for_scalar_type_and_size): Likewise.
7339 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7341 * machmode.h (mode_for_int_vector): New function.
7342 * stor-layout.c (mode_for_int_vector): Likewise.
7343 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
7344 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
7345 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
7346 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
7347 (s390_expand_vcond): Likewise.
7349 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7351 * machmode.h (opt_machine_mode): New type.
7352 (opt_mode<T>): Allow construction from anything that can be
7354 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
7355 (mode_for_size): Return an opt_machine_mode.
7356 * stor-layout.c (mode_for_size): Likewise.
7357 (mode_for_size_tree): Update call accordingly.
7358 (bitwise_mode_for_mode): Likewise.
7359 (make_fract_type): Likewise.
7360 (make_accum_type): Likewise.
7361 * caller-save.c (replace_reg_with_saved_mem): Update call
7363 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7364 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7365 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7366 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7367 * expmed.c (extract_bit_field_1): Likewise.
7368 * reload.c (get_secondary_mem): Likewise.
7369 * varasm.c (assemble_integer): Likewise.
7370 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
7373 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7375 * machmode.h (decimal_float_mode_for_size): New function.
7376 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
7377 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
7378 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
7379 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
7380 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
7381 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
7383 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7385 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
7386 (get_builtin_sync_mode): Likewise.
7387 (expand_ifn_atomic_compare_exchange): Likewise.
7388 (expand_builtin_atomic_clear): Likewise.
7389 (expand_builtin_atomic_test_and_set): Likewise.
7390 (fold_builtin_atomic_always_lock_free): Likewise.
7391 * calls.c (compute_argument_addresses): Likewise.
7392 (emit_library_call_value_1): Likewise.
7393 (store_one_arg): Likewise.
7394 * combine.c (combine_instructions): Likewise.
7395 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
7396 * config/arm/arm.c (arm_function_value): Likewise.
7397 (aapcs_allocate_return_reg): Likewise.
7398 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
7399 * config/i386/i386.c (construct_container): Likewise.
7400 (ix86_gimplify_va_arg): Likewise.
7401 (ix86_expand_sse_cmp): Likewise.
7402 (emit_memmov): Likewise.
7403 (emit_memset): Likewise.
7404 (expand_small_movmem_or_setmem): Likewise.
7405 (ix86_expand_pextr): Likewise.
7406 (ix86_expand_pinsr): Likewise.
7407 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
7408 * config/microblaze/microblaze.c (microblaze_block_move_straight):
7410 * config/mips/mips.c (mips_function_value_1) Likewise.
7411 (mips_block_move_straight): Likewise.
7412 (mips_expand_ins_as_unaligned_store): Likewise.
7413 * config/powerpcspe/powerpcspe.c
7414 (rs6000_darwin64_record_arg_advance_flush): Likewise.
7415 (rs6000_darwin64_record_arg_flush): Likewise.
7416 * config/rs6000/rs6000.c
7417 (rs6000_darwin64_record_arg_advance_flush): Likewise.
7418 (rs6000_darwin64_record_arg_flush): Likewise.
7419 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
7420 (sparc_function_value_1): Likewise.
7421 * config/spu/spu.c (adjust_operand): Likewise.
7422 (spu_emit_branch_or_set): Likewise.
7423 (arith_immediate_p): Likewise.
7424 * emit-rtl.c (gen_lowpart_common): Likewise.
7425 * expr.c (expand_expr_real_1): Likewise.
7426 * function.c (assign_parm_setup_block): Likewise.
7427 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
7428 * reload1.c (alter_reg): Likewise.
7429 * stor-layout.c (mode_for_vector): Likewise.
7430 (layout_type): Likewise.
7432 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7434 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
7435 (spu_convert_move): Likewise.
7436 * lower-subreg.c (resolve_simple_move): Likewise.
7438 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7441 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
7442 define_insn to a define_expand.
7443 (altivec_vsum2sws_direct): New define_insn.
7444 (altivec_vsumsws): Convert from a define_insn to a define_expand.
7446 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
7448 * config/arm/arm.c (arm_option_params_internal): Improve setting of
7451 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
7456 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
7457 realign stack if stack alignment needed is less than incoming
7460 2017-09-05 Marek Polacek <polacek@redhat.com>
7463 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
7466 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
7468 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
7470 2017-09-05 Richard Biener <rguenther@suse.de>
7472 PR tree-optimization/82084
7473 * fold-const.c (can_native_encode_string_p): Handle wide characters.
7475 2017-09-05 Richard Biener <rguenther@suse.de>
7477 PR tree-optimization/82102
7478 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
7480 2017-09-05 Martin Liska <mliska@suse.cz>
7482 PR tree-optimization/82032
7483 * tree-cfg.c (generate_range_test): New function.
7484 * tree-cfg.h (generate_range_test): Declared here.
7485 * tree-cfgcleanup.c (convert_single_case_switch): New function.
7486 (cleanup_control_expr_graph): Use it.
7487 * tree-switch-conversion.c (try_switch_expansion): Remove
7489 (emit_case_nodes): Use generate_range_test.
7491 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
7494 * config/i386/i386.md (*<btsc><mode>_mask): Add
7495 TARGET_USE_BT to insn constraint.
7496 (*btr<mode>_mask): Ditto.
7498 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
7500 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
7501 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
7503 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
7506 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
7507 TARGET_NEON and TARGET_IWMMXT.
7508 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
7509 TARGET_NEON and TARGET_IWMMXT.
7510 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
7512 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
7514 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
7515 (ix86_rewrite_tls_address): Ditto.
7516 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
7517 (ix86_rewrite_tls_address_1): Ditto.
7518 (ix86_rewrite_tls_address): Ditto.
7519 * config/i386/predicates.md (tls_address_pattern): New predicate.
7520 * config/i386/i386.md (TLS address splitter): New splitter.
7522 2017-09-04 Richard Biener <rguenther@suse.de>
7524 PR tree-optimization/82084
7525 * fold-const.h (can_native_encode_string_p): Declare.
7526 * fold-const.c (can_native_encode_string_p): Factor out from ...
7527 (native_encode_string): ... here.
7528 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
7529 vectorizing stores from constants we later cannot handle.
7531 2017-09-04 Marek Polacek <polacek@redhat.com>
7534 * doc/invoke.texi: Update -Wtautological-compare documentation.
7536 2017-09-04 Jeff Law <law@redhat.com>
7538 PR tree-optimization/64910
7539 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
7540 swap the first and last operand if the last is a constant.
7542 2017-09-04 Marek Polacek <polacek@redhat.com>
7545 * convert.c (do_narrow): When sanitizing signed integer overflows,
7546 bail out for signed types.
7547 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
7549 2017-09-04 Richard Biener <rguenther@suse.de>
7551 PR tree-optimization/82060
7552 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
7553 Move devirtualization after stmt folding and before EH/AB/noreturn
7554 cleanup to get the stmt refs canonicalized. Use a bool instead
7555 of gimple_modified_p since that doesn't work for NOPs. Schedule
7556 NOPs generated by folding for removal.
7558 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7559 Alan Hayward <alan.hayward@arm.com>
7560 David Sherwood <david.sherwood@arm.com>
7562 * coretypes.h (pad_direction): New enum.
7563 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
7564 (FUNCTION_ARG_PADDING): Likewise.
7565 * target.def (function_arg_padding): New hook.
7566 * targhooks.h (default_function_arg_padding): Declare.
7567 * targhooks.c (default_function_arg_padding): New function.
7568 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
7569 (TARGET_FUNCTION_ARG_PADDING): ...this.
7570 * doc/tm.texi: Regenerate.
7571 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
7572 instead of direction.
7573 (compute_argument_addresses): Likewise.
7574 (load_register_parameters): Likewise.
7575 (emit_library_call_value_1): Likewise.
7576 (store_one_arg): Use targetm.calls.function_arg_padding instead
7577 of FUNCTION_ARG_PADDING.
7578 (must_pass_in_stack_var_size_or_pad): Likewise.
7579 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
7580 (emit_group_store): Likewise.
7581 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
7582 instead of FUNCTION_ARG_PADDING.
7583 (emit_push_insn): Likewise, and propagate enum change throughout
7585 * function.h (direction): Delete.
7586 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
7588 * function.c (assign_parm_find_stack_rtl): Likewise.
7589 (assign_parm_setup_block_p): Likewise.
7590 (assign_parm_setup_block): Likewise.
7591 (gimplify_parameters): Likewise.
7592 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
7593 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
7595 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
7596 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7597 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
7598 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
7599 (aarch64_function_arg_padding): ...this new function.
7600 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
7601 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7602 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
7603 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7604 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
7605 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7606 (arm_pad_arg_upward): Replace with...
7607 (arm_function_arg_padding): ...this new function.
7608 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
7610 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
7611 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
7612 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7613 (ia64_hpux_function_arg_padding): Replace with...
7614 (ia64_function_arg_padding): ...this new function. Use pad_direction
7615 instead of direction. Check for TARGET_HPUX.
7616 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
7617 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7618 (iq2000_function_arg_padding): New function.
7619 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
7620 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
7621 (mips_function_arg_padding): ...this new function.
7622 (mips_pad_reg_upward): Update accordingly.
7623 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7624 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
7625 targetm.calls.function_arg_padding.
7626 (FUNCTION_ARG_PADDING): Delete.
7627 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7628 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
7629 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7630 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
7631 (nios2_block_reg_padding): Return pad_direction instead of direction.
7632 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
7633 instead of direction.
7634 (nios2_function_arg_padding): Likewise. Make static.
7635 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7636 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
7637 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
7638 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
7639 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7640 (pa_function_arg_padding): Make static. Return pad_direction instead
7642 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
7643 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7644 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
7645 instead of direction. Use targetm.calls.function_arg_padding.
7646 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
7647 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
7648 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
7649 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
7650 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
7652 (function_arg_padding): Rename to...
7653 (rs6000_function_arg_padding): ...this. Make static. Return
7654 pad_direction instead of direction.
7655 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
7656 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
7657 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7658 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
7659 instead of direction. Use targetm.calls.function_arg_padding.
7660 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
7661 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
7662 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
7663 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
7664 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7665 (function_arg_padding): Rename to...
7666 (rs6000_function_arg_padding): ...this. Make static. Return
7667 pad_direction instead of direction.
7668 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
7669 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
7670 * config/s390/s390.c (s390_function_arg_padding): New function.
7671 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7672 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
7673 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
7674 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7675 (function_arg_padding): Rename to...
7676 (sparc_function_arg_padding): ...this. Make static. Return
7677 pad_direction instead of direction.
7678 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
7679 * config/spu/spu.c (spu_function_arg_padding): New function.
7680 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7681 * system.h (FUNCTION_ARG_PADDING): Poison.
7683 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7684 Alan Hayward <alan.hayward@arm.com>
7685 David Sherwood <david.sherwood@arm.com>
7687 * target.def (modes_tieable_p): New hook.
7688 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
7689 (TARGET_MODES_TIEABLE_P): ...this.
7690 * doc/tm.texi.in: Regenerate.
7691 * hooks.h (hook_bool_mode_mode_true): Declare.
7692 * hooks.c (hook_bool_mode_mode_true): New function.
7693 * combine.c (subst): Use targetm.modes_tieable_p instead of
7695 * dse.c (find_shift_sequence): Likewise.
7696 * expmed.c (extract_low_bits): Likewise.
7697 * lower-subreg.c: Include target.h.
7698 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
7700 * rtlanal.c (rtx_cost): Likewise.
7701 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
7702 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
7703 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
7704 (TARGET_MODES_TIEABLE_P): Redefine.
7705 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
7706 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
7707 (TARGET_MODES_TIEABLE_P): Redefine.
7708 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
7709 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
7710 (arc_modes_tieable_p): New function.
7711 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
7712 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
7713 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
7714 (arm_modes_tieable_p): Make static.
7715 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
7716 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
7717 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
7718 (TARGET_MODES_TIEABLE_P): Redefine.
7719 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
7720 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
7721 (TARGET_MODES_TIEABLE_P): Redefine.
7722 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
7723 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
7724 (cr16_modes_tieable_p): New function.
7725 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
7726 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
7727 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
7728 (TRULY_NOOP_TRUNCATION): Update comment.
7729 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
7730 (TRULY_NOOP_TRUNCATION): Update comment.
7731 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
7732 (frv_modes_tieable_p): New function.
7733 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
7734 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
7735 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
7736 (TARGET_MODES_TIEABLE_P): Redefine.
7737 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
7738 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
7739 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
7740 (TARGET_MODES_TIEABLE_P): Redefine.
7741 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
7742 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
7743 (ia64_modes_tieable_p): New function.
7744 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
7745 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
7746 (iq2000_modes_tieable_p): New function.
7747 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
7748 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
7749 (lm32_modes_tieable_p): New function.
7750 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
7751 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
7752 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
7753 (TARGET_MODES_TIEABLE_P): Redefine.
7754 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
7755 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
7756 (m32r_modes_tieable_p): New function.
7757 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
7758 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
7759 (m68k_modes_tieable_p): New function.
7760 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
7761 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
7762 (mcore_modes_tieable_p): New function.
7763 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
7764 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
7766 (TARGET_MODES_TIEABLE_P): Redefine.
7767 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
7768 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
7769 * config/mips/mips.c (mips_modes_tieable_p): Make static.
7770 (TARGET_MODES_TIEABLE_P): Redefine.
7771 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
7772 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
7773 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
7774 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
7775 (mn10300_modes_tieable_p): ...this and make static.
7776 (TARGET_MODES_TIEABLE_P): Redefine.
7777 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
7778 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
7779 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
7780 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
7781 (msp430_modes_tieable_p): Make static.
7782 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
7783 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
7784 (TARGET_MODES_TIEABLE_P): Redefine.
7785 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
7786 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
7787 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
7788 (TARGET_MODES_TIEABLE_P): Redefine.
7789 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
7790 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
7791 * config/pa/pa.c (pa_modes_tieable_p): Make static.
7792 (TARGET_MODES_TIEABLE_P): Redefine.
7793 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
7794 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
7795 (pdp11_modes_tieable_p): New function.
7796 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
7797 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
7798 (rs6000_modes_tieable_p): New function.
7799 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
7800 * config/powerpcspe/powerpcspe.md: Update comment.
7801 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
7802 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
7803 (TARGET_MODES_TIEABLE_P): Redefine.
7804 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
7805 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
7806 (rl78_modes_tieable_p): New function.
7807 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
7808 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
7809 (rs6000_modes_tieable_p): New function.
7810 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
7811 * config/rs6000/rs6000.md: Update comment.
7812 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
7813 * config/rx/rx.c (rx_modes_tieable_p): New function.
7814 (TARGET_MODES_TIEABLE_P): Redefine.
7815 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
7816 * config/s390/s390.c (s390_modes_tieable_p): New function.
7817 (TARGET_MODES_TIEABLE_P): Redefine.
7818 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
7819 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
7820 (sh_modes_tieable_p): New function.
7821 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
7822 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
7823 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
7824 (sparc_modes_tieable_p): Make static.
7825 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
7826 * config/spu/spu.c (spu_modes_tieable_p): New function.
7827 (TARGET_MODES_TIEABLE_P): Redefine.
7828 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
7829 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
7830 (TARGET_MODES_TIEABLE_P): Redefine.
7831 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
7832 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
7833 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
7834 * config/v850/v850.c (v850_modes_tieable_p): New function.
7835 (TARGET_MODES_TIEABLE_P): Redefine.
7836 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
7837 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
7838 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
7839 (visium_modes_tieable_p): New function.
7840 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
7841 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
7842 (xtensa_modes_tieable_p): New function.
7843 * system.h (MODES_TIEABLE_P): Poison.
7845 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7846 Alan Hayward <alan.hayward@arm.com>
7847 David Sherwood <david.sherwood@arm.com>
7849 * target.def (hard_regno_mode_ok): New hook.
7850 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
7851 (TARGET_HARD_REGNO_MODE_OK): ...this.
7852 * doc/tm.texi.in: Regenerate.
7853 * hooks.h (hook_bool_uint_mode_true): Declare.
7854 * hooks.c (hook_bool_uint_mode_true): New function.
7855 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
7857 * genpreds.c (write_insn_preds_c): Add an include of target.h.
7858 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
7859 instead of HARD_REGNO_MODE_OK.
7860 * caller-save.c: Include target.h.
7861 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
7863 * combine.c (can_combine_p): Likewise.
7864 (combinable_i3pat): Likewise.
7865 (can_change_dest_mode): Likewise.
7866 * expr.c (init_expr_target): Likewise.
7867 (convert_move): Likewise.
7868 (convert_modes): Likewise.
7869 * ira.c (setup_prohibited_class_mode_regs): Likewise.
7870 (setup_prohibited_mode_move_regs): Likewise.
7871 * ira.h (target_ira): Likewise.
7872 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7873 * lra-constraints.c (process_alt_operands): Likewise.
7874 (split_reg): Likewise.
7875 * recog.c (peep2_find_free_register): Likewise.
7876 * ree.c (combine_reaching_defs): Likewise.
7877 * regcprop.c (maybe_mode_change): Likewise.
7878 * reginfo.c (init_reg_sets_1): Likewise.
7879 (choose_hard_reg_mode): Likewise.
7880 (simplifiable_subregs): Likewise.
7881 * regrename.c (check_new_reg_p): Likewise.
7882 * reload.c (find_valid_class): Likewise.
7883 (find_valid_class_1): Likewise.
7884 (reload_inner_reg_of_subreg): Likewise.
7885 (push_reload): Likewise.
7886 (combine_reloads): Likewise.
7887 (find_dummy_reload): Likewise.
7888 (find_reloads): Likewise.
7889 * reload1.c (find_reg): Likewise.
7890 (set_reload_reg): Likewise.
7891 (allocate_reload_reg): Likewise.
7892 (choose_reload_regs): Likewise.
7893 (reload_adjust_reg_for_temp): Likewise.
7894 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
7895 (simplify_subreg_regno): Likewise.
7896 * sel-sched.c (init_regs_for_mode): Likewise.
7897 * varasm.c (make_decl_rtl): Likewise.
7898 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
7899 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
7901 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
7902 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
7903 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7904 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
7905 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
7906 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7907 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
7908 (arc_mode_class): Delete.
7909 (HARD_REGNO_MODE_OK): Delete.
7910 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7911 (arc_hard_regno_mode_ok): Rename old array to...
7912 (arc_hard_regno_mode_ok_modes): ...this.
7913 (arc_conditional_register_usage): Update accordingly.
7914 (arc_mode_class): Make static.
7915 (arc_hard_regno_mode_ok): New function.
7916 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
7917 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
7918 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7919 (arm_hard_regno_mode_ok): Make static.
7920 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
7922 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
7923 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
7924 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
7926 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7927 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
7928 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
7929 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
7930 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
7931 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7932 * config/bfin/predicates.md (valid_reg_operand): Use
7933 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
7934 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
7935 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
7936 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7937 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
7938 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
7939 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7940 (cr16_hard_regno_mode_ok): Make static and return a bool.
7941 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
7942 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7943 (cris_hard_regno_mode_ok): New function.
7944 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
7945 (epiphany_mode_class): Delete.
7946 (HARD_REGNO_MODE_OK): Delete.
7947 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
7948 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7949 (hard_regno_mode_ok): Rename to...
7950 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
7951 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
7952 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
7954 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
7955 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
7956 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7957 (frv_hard_regno_mode_ok): Make static and return a bool.
7958 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
7960 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
7961 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
7962 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
7963 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
7965 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7966 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
7967 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
7968 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
7970 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7971 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
7972 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7973 (ia64_hard_regno_mode_ok): New function.
7974 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
7975 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7976 (iq2000_hard_regno_mode_ok): New function.
7977 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
7978 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7979 (lm32_hard_regno_mode_ok): New function.
7980 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
7981 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
7982 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
7983 instead of HARD_REGNO_MODE_OK.
7984 (m32c_hard_regno_ok): Rename to...
7985 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
7986 (m32c_cannot_change_mode_class): Update accordingly.
7987 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7988 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
7989 (m32r_mode_class): Delete.
7990 (HARD_REGNO_MODE_OK): Delete.
7991 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7992 (m32r_hard_regno_mode_ok): Rename to...
7993 (m32r_hard_regno_modes): ...this.
7994 (m32r_mode_class): Make static.
7995 (m32r_hard_regno_mode_ok): New function.
7996 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
7997 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
7998 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7999 (m68k_hard_regno_mode_ok): Make static.
8000 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
8001 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8002 (mcore_hard_regno_mode_ok): New function.
8003 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
8004 (HARD_REGNO_MODE_OK): Delete.
8005 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
8007 (microblaze_hard_regno_mode_ok_p): ...this and make static.
8008 (microblaze_hard_regno_mode_ok): New function.
8009 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8010 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
8011 (mips_hard_regno_mode_ok): Delete.
8012 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
8013 (mips_hard_regno_mode_ok_p): ...this and make static.
8014 (mips_hard_regno_mode_ok_p): Rename to...
8015 (mips_hard_regno_mode_ok_uncached): ...this.
8016 (mips_hard_regno_mode_ok): New function.
8017 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
8018 of HARD_REGNO_MODE_OK.
8019 (mips_option_override): Update after above name changes.
8020 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8021 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
8022 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
8023 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
8024 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
8025 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8026 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
8027 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
8028 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
8029 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8030 (msp430_hard_regno_mode_ok): Make static and return a bool.
8031 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
8032 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
8033 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
8035 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8036 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
8037 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
8038 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
8039 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
8040 (PA_HARD_REGNO_MODE_OK): ...this
8041 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
8042 (PA_HARD_REGNO_MODE_OK): ...this.
8043 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8044 (pa_hard_regno_mode_ok): New function.
8045 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
8046 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8047 (pdp11_hard_regno_mode_ok): New function.
8048 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
8049 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
8051 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
8053 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8054 (rs6000_hard_regno_mode_ok): Rename to...
8055 (rs6000_hard_regno_mode_ok_uncached): ...this.
8056 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8057 (rs6000_hard_regno_mode_ok): New function.
8058 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
8059 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
8060 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
8061 (riscv_hard_regno_mode_ok): ...this and make static.
8062 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8063 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
8064 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
8065 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8066 (rl78_hard_regno_mode_ok): Make static and return bool.
8067 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
8068 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
8070 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
8071 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8072 (rs6000_hard_regno_mode_ok): Rename to...
8073 (rs6000_hard_regno_mode_ok_uncached): ...this.
8074 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8075 (rs6000_hard_regno_mode_ok): New function.
8076 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
8077 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
8078 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8079 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
8080 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
8081 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
8082 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8083 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
8084 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
8085 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8086 (sh_hard_regno_mode_ok): Make static.
8087 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
8088 instead of HARD_REGNO_MODE_OK.
8089 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
8090 (sparc_mode_class): Delete.
8091 (HARD_REGNO_MODE_OK): Delete.
8092 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8093 (hard_regno_mode_classes): Make static.
8094 (sparc_mode_class): Likewise.
8095 (sparc_hard_regno_mode_ok): New function.
8096 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
8097 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
8098 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
8100 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8101 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
8102 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
8103 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
8104 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
8105 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8106 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
8107 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
8108 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8109 (visium_hard_regno_mode_ok): New function.
8110 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
8111 instead of HARD_REGNO_MODE_OK.
8112 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
8113 (HARD_REGNO_MODE_OK): Delete.
8114 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
8115 (xtensa_hard_regno_mode_ok_p): ...this and make static.
8116 (xtensa_option_override): Update accordingly.
8117 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8118 (xtensa_hard_regno_mode_ok): New function.
8119 * system.h (HARD_REGNO_MODE_OK): Poison.
8121 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8122 Alan Hayward <alan.hayward@arm.com>
8123 David Sherwood <david.sherwood@arm.com>
8125 * target.def (hard_regno_call_part_clobbered): New hook.
8126 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
8127 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
8128 * doc/tm.texi: Regenerate.
8129 * hooks.h (hook_bool_uint_mode_false): Declare.
8130 * hooks.c (hook_bool_uint_mode_false): New function.
8131 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8132 * cselib.c (cselib_process_insn): Use
8133 targetm.hard_regno_call_part_clobbered instead of
8134 HARD_REGNO_CALL_PART_CLOBBERED.
8135 * ira-conflicts.c (ira_build_conflicts): Likewise.
8136 * ira-costs.c (ira_tune_allocno_costs): Likewise.
8137 * lra-constraints.c (need_for_call_save_p): Likewise.
8138 * lra-lives.c: Include target.h.
8139 (check_pseudos_live_through_calls): Use
8140 targetm.hard_regno_call_part_clobbered instead of
8141 HARD_REGNO_CALL_PART_CLOBBERED.
8142 * regcprop.c: Include target.h.
8143 (copyprop_hardreg_forward_1): Use
8144 targetm.hard_regno_call_part_clobbered instead of
8145 HARD_REGNO_CALL_PART_CLOBBERED.
8146 * reginfo.c (choose_hard_reg_mode): Likewise.
8147 * regrename.c (check_new_reg_p): Likewise.
8148 * reload.c (find_equiv_reg): Likewise.
8149 * reload1.c (emit_reload_insns): Likewise.
8150 * sched-deps.c (deps_analyze_insn): Likewise.
8151 * sel-sched.c (init_regs_for_mode): Likewise.
8152 (mark_unavailable_hard_regs): Likewise.
8153 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
8154 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8155 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
8157 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8158 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8159 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
8161 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
8163 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8164 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8165 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
8167 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8168 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8169 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
8171 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8172 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
8174 * config/powerpcspe/powerpcspe.c
8175 (rs6000_hard_regno_call_part_clobbered): New function.
8176 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8177 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8178 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
8180 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8181 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8182 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
8184 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8185 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8186 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
8188 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8189 Alan Hayward <alan.hayward@arm.com>
8190 David Sherwood <david.sherwood@arm.com>
8192 * rtl.h (subreg_memory_offset): Declare.
8193 * emit-rtl.c (subreg_memory_offset): New function.
8194 * expmed.c (store_bit_field_1): Use it.
8195 * expr.c (undefined_operand_subword_p): Likewise.
8196 * simplify-rtx.c (simplify_subreg): Likewise.
8198 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
8200 PR rtl-optimization/57448
8203 * optabs.c (expand_atomic_load): Place compiler memory barriers if
8204 using atomic_load pattern.
8205 (expand_atomic_store): Likewise.
8207 2017-09-04 Jakub Jelinek <jakub@redhat.com>
8210 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
8211 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
8212 handling. Use replace_call_with_value with NULL instead of
8213 gsi_replace, unlink_stmt_vdef and release_defs.
8215 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
8218 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
8220 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8223 * rtl.h (emit_library_call_value_1): Declare.
8224 (emit_library_call): Replace declaration with a series of overloads.
8225 Remove the parameter count argument.
8226 (emit_library_call_value): Likewise.
8227 * calls.c (emit_library_call_value_1): Make global. Replace varargs
8228 with an "rtx_mode_t *".
8229 (emit_library_call_value): Delete.
8230 (emit_library_call): Likewise.
8231 * asan.c (asan_emit_stack_protection): Update calls accordingly.
8232 (asan_emit_allocas_unpoison): Likewise.
8233 * builtins.c (expand_builtin_powi): Likewise.
8234 (expand_asan_emit_allocas_unpoison): Likewise.
8235 * cfgexpand.c (expand_main_function): Likewise.
8236 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
8237 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
8238 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
8239 * config/arm/arm.c (arm_trampoline_init): Likewise.
8240 (arm_call_tls_get_addr): Likewise.
8241 (arm_expand_divmod_libfunc): Likewise.
8242 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
8243 (smulsi3_highpart): Likewise.
8244 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
8245 (c6x_expand_compare): Likewise.
8246 (c6x_expand_movmem): Likewise.
8247 * config/frv/frv.c (frv_trampoline_init): Likewise.
8248 * config/i386/i386.c (ix86_trampoline_init): Likewise.
8249 (ix86_expand_divmod_libfunc): Likewise.
8250 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
8251 (ia64_expand_compare): Likewise.
8252 (ia64_profile_hook): Likewise.
8253 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
8254 (nonlocal_goto): Likewise.
8255 (restore_stack_nonlocal): Likewise.
8256 * config/m32r/m32r.c (block_move_call): Likewise.
8257 (m32r_trampoline_init): Likewise.
8258 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
8259 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
8260 (m68k_call_m68k_read_tp): Likewise.
8261 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
8262 (microblaze_expand_divide): Likewise.
8263 * config/mips/mips.h (mips_args): Likewise.
8264 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
8265 (MIPS_ICACHE_SYNC): Likewise.
8266 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
8267 (nios2_trampoline_init): Likewise.
8268 * config/pa/pa.c (hppa_tls_call): Likewise.
8269 (pa_trampoline_init): Likewise.
8270 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
8271 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
8272 (expand_strn_compare): Likewise.
8273 (rs6000_generate_compare): Likewise.
8274 (rs6000_expand_float128_convert): Likewise.
8275 (output_profile_hook): Likewise.
8276 (rs6000_trampoline_init): Likewise.
8277 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
8278 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
8279 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
8280 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
8281 (rs6000_generate_compare): Likewise.
8282 (rs6000_expand_float128_convert): Likewise.
8283 (output_profile_hook): Likewise.
8284 (rs6000_trampoline_init): Likewise.
8285 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
8286 * config/sh/sh.c (sh_trampoline_init): Likewise.
8287 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
8288 (sparc_emit_float_lib_cmp): Likewise.
8289 (sparc32_initialize_trampoline): Likewise.
8290 (sparc64_initialize_trampoline): Likewise.
8291 (sparc_profile_hook): Likewise.
8292 * config/spu/spu.c (ea_load_store): Likewise.
8293 * config/spu/spu.md (floatunssidf2): Likewise.
8294 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
8295 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
8296 * config/visium/visium.c (expand_block_move_4): Likewise.
8297 (expand_block_move_2): Likewise.
8298 (expand_block_move_1): Likewise.
8299 (expand_block_set_4): Likewise.
8300 (expand_block_set_2): Likewise.
8301 (expand_block_set_1): Likewise.
8302 (visium_trampoline_init): Likewise.
8303 (visium_profile_hook): Likewise.
8304 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
8305 (xtensa_setup_frame_addresses): Likewise.
8306 (xtensa_trampoline_init): Likewise.
8307 * except.c (sjlj_emit_function_enter): Likewise.
8308 (sjlj_emit_function_exit): Likewise.
8309 * explow.c (allocate_dynamic_stack_space): Likewise.
8310 (probe_stack_range): Likewise.
8311 * expr.c (convert_mode_scalar): Likewise.
8312 * optabs.c (expand_binop): Likewise.
8313 (expand_twoval_binop_libfunc): Likewise.
8314 (expand_unop): Likewise.
8315 (prepare_cmp_insn): Likewise.
8316 (prepare_float_lib_cmp): Likewise.
8317 (expand_float): Likewise.
8318 (expand_fix): Likewise.
8319 (expand_fixed_convert): Likewise.
8320 (maybe_emit_sync_lock_test_and_set): Likewise.
8321 (expand_atomic_compare_and_swap): Likewise.
8322 (expand_mem_thread_fence): Likewise.
8323 (expand_atomic_fetch_op): Likewise.
8325 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
8327 * doc/generic.texi (OpenACC): Adjust URL.
8328 * doc/invoke.texi (C Dialect Options): Ditto.
8330 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
8332 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
8333 predicate for operand 1. Add (m,<S>) constraint.
8334 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
8335 Prevent memory operand 1 with register operand 2.
8337 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
8339 PR rtl-optimization/82024
8340 * combine.c (try_combine): If the combination result is a PARALLEL,
8341 and we only need to retain the SET in there that would be placed
8342 at I2, check that we can place that at I3 instead, before doing so.
8344 2017-09-01 Jakub Jelinek <jakub@redhat.com>
8347 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
8349 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
8350 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
8353 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
8354 Jeff Law <law@redhat.com>
8356 * varasm.c (bss_initializer_p): Do not put constants into .bss
8357 (categorize_decl_for_section): Handle bss_initializer_p returning
8358 false when DECL_INITIAL is NULL.
8360 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8363 * config/s390/s390.c (s390_can_inline_p): New function.
8365 2017-09-01 Jeff Law <law@redhat.com>
8367 PR tree-optimization/82052
8368 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8369 Always initialize the returned slot after a hash table miss
8370 when INSERT is true.
8372 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
8374 * config/s390/s390.md (mem_signal_fence): Remove.
8375 * doc/md.texi (mem_signal_fence): Remove.
8376 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
8378 * target-insns.def (mem_signal_fence): Remove.
8380 2017-09-01 Jakub Jelinek <jakub@redhat.com>
8383 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
8386 * asan.c (create_odr_indicator): Strip name encoding from assembler
8387 name before appending it after __odr_asan_.
8389 2017-09-01 Martin Liska <mliska@suse.cz>
8391 PR tree-optimization/82059
8392 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
8393 frequency only when an edge is redirected.
8395 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8397 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
8398 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
8399 (arc_conditional_register_usage): Remove ARC600 lp_count
8401 (arc_file_start): Emit Tag_ARC_CPU_variation.
8402 (arc_can_use_doloop_p): New conditions to use ZOLs.
8403 (hwloop_fail): New function.
8404 (hwloop_optimize): Likewise.
8405 (hwloop_pattern_reg): Likewise.
8406 (arc_doloop_hooks): New struct, to be used with reorg_loops.
8407 (arc_reorg_loops): New function, calls reorg_loops.
8408 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
8409 (arc600_corereg_hazard): Remove ZOL checking, case handled by
8411 (arc_loop_hazard): Remove function, functionality moved into
8413 (arc_hazard): Remove arc_loop_hazard call.
8414 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
8415 into hwloop_optimize.
8416 (arc_label_align): Remove ZOL handling.
8417 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
8418 * config/arc/arc.md (doloop_begin): Remove pattern.
8419 (doloop_begin_i): Likewise.
8420 (doloop_end_i): Likewise.
8421 (doloop_fallback): Likewise.
8422 (doloop_fallback_m): Likewise.
8423 (doloop_end): Reimplement expand.
8424 (arc_lp): New pattern for LP instruction.
8425 (loop_end): New pattern.
8426 (loop_fail): Likewise.
8427 (decrement_and_branch_until_zero): Likewise.
8428 * config/arc/arc.opt (mlpc-width): New option.
8429 * doc/invoke.texi (mlpc-width): Document option.
8431 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8433 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
8434 (arc_ccfsm_advance): Fix checking for delay slots.
8435 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
8437 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8439 * config/arc/arc.md (movqi_insn): Add stores to save constant long
8441 (movhi_insn): Update store instruction constraint which are saving
8442 6-bit short immediates.
8443 (movsi_insn): Consider also short scaled load operations.
8444 (zero_extendhisi2_i): Use Usd constraint instead of T.
8445 (extendhisi2_i): Add q constraint.
8446 (arc_clzsi2): Add type and length attributes.
8447 (arc_ctzsi2): Likewise.
8448 * config/arc/constraints.md (Usc): Update constraint, the
8449 assembler can parse two relocations for a single instruction.
8451 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8453 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
8454 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
8456 2017-08-31 Olivier Hainque <hainque@adacore.com>
8458 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
8459 match as powerpc-wrs-vxworks*.
8461 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
8463 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
8464 register constraint for by-element operand.
8465 (aarch64_mls_elt_merge<mode>): Likewise.
8467 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8469 * config/arc/arc.c (arc_can_follow_jump): Check for short
8472 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8474 * config.gcc: Use g.opt for arc.
8475 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
8476 functionality moved to ...
8477 (legitimate_scaled_address_p): New function, ...here.
8478 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
8479 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
8480 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
8482 (arc_override_options): Handle G option.
8483 (arc_output_pic_addr_const): Correct function definition.
8484 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
8485 (arc_decl_anon_ns_mem_p): Delete.
8486 (arc_in_small_data_p): Overhaul this function to take into
8487 consideration the value given via G option.
8488 (arc_rewrite_small_data_1): Renamed and corrected old
8489 arc_rewrite_small_data function.
8490 (arc_rewrite_small_data): New function.
8491 (small_data_pattern): Don't use pic_offset_table_rtx.
8492 * config/arc/arc.h (CC1_SPEC): Recognize G option.
8493 * config/arc/simdext.md (movmisalignv2hi): Use
8494 prepare_move_operands function.
8496 (movmisalign*): Likewise.
8497 * doc/invoke.texi (ARC options): Document -G option.
8499 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8501 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
8503 * config/arc/arc.c (arc_print_operand): Output scalled address for
8504 sdata whenever is possible.
8505 (arc_in_small_data_p): Allow sdata for 64bit datum when double
8506 load/stores are available.
8507 (compact_sda_memory_operand): Check for the alignment required by
8508 code density instructions.
8509 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
8511 * config/arc/constraints.md (Usd): Update constraint.
8512 (Us0): New constraint.
8513 (Usc): Update constraint.
8515 2017-08-31 Richard Biener <rguenther@suse.de>
8518 * dwarf2out.c (dwarf2out_early_global_decl): Process each
8521 2017-08-31 Tamar Christina <tamar.christina@arm.com>
8523 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
8524 Resize type_signature.
8526 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8527 Alan Hayward <alan.hayward@arm.com>
8528 David Sherwood <david.sherwood@arm.com>
8530 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
8531 subregs whose inner modes can be stored in GPRs.
8532 (aarch64_classify_index): Likewise.
8534 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8535 Alan Hayward <alan.hayward@arm.com>
8536 David Sherwood <david.sherwood@arm.com>
8538 * config/aarch64/iterators.md (V_cmp_result): Rename to...
8539 (V_INT_EQUIV): ...this.
8540 (v_cmp_result): Rename to...
8541 (v_int_equiv): ...this.
8542 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
8543 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
8544 (copysign<mode>3): Likewise.
8545 (aarch64_simd_bsl<mode>_internal): Likewise.
8546 (aarch64_simd_bsl<mode>): Likewise.
8547 (vec_cmp<mode><mode>): Likewise.
8548 (vcond<mode><mode>): Likewise.
8549 (vcond<v_cmp_mixed><mode>): Likewise.
8550 (vcondu<mode><v_cmp_mixed>): Likewise.
8551 (aarch64_cm<optab><mode>): Likewise.
8552 (aarch64_cmtst<mode>): Likewise.
8553 (aarch64_fac<optab><mode>): Likewise.
8554 (vec_perm_const<mode>): Likewise.
8555 (vcond_mask_<mode><v_cmp_result>): Rename to...
8556 (vcond_mask_<mode><v_int_equiv>): ...this.
8557 (vec_cmp<mode><v_cmp_result>): Rename to...
8558 (vec_cmp<mode><v_int_equiv>): ...this.
8560 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8561 Alan Hayward <alan.hayward@arm.com>
8562 David Sherwood <david.sherwood@arm.com>
8564 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
8566 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
8567 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
8568 (UNSPEC_LD4_DREG): New unspecs.
8569 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
8570 (aarch64_ld2<mode>_dreg_be): Replace with...
8571 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
8573 (aarch64_ld3<mode>_dreg_le)
8574 (aarch64_ld3<mode>_dreg_be): Replace with...
8575 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
8577 (aarch64_ld4<mode>_dreg_le)
8578 (aarch64_ld4<mode>_dreg_be): Replace with...
8579 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
8582 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8584 PR tree-optimization/81987
8585 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
8586 insert an initializer in a location not dominated by the stride
8589 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8591 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
8592 on the entire header of the finally block in the fallthru case.
8594 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8596 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
8598 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
8600 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
8601 rs6000_emit_move_from_cr and call renamed function.
8602 (rs6000_emit_prologue): Call renamed functions.
8603 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
8604 movesi_from_cr, remove volatile CRs.
8606 2017-08-30 Jon Beniston <jon@beniston.com>
8607 Richard Biener <rguenther@suse.de>
8609 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
8610 of VECTOR_MODE_P check.
8611 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
8612 element vector types.
8614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8616 * df.h (df_read_modify_subreg_p): Remove in favor of...
8617 * rtl.h (read_modify_subreg_p): ...this new function. Take a
8618 const_rtx instead of an rtx.
8619 * cprop.c (local_cprop_find_used_regs): Update accordingly.
8620 * df-problems.c (df_word_lr_mark_ref): Likewise.
8621 * ira-lives.c (mark_pseudo_reg_live): Likewise.
8622 (mark_pseudo_reg_dead): Likewise.
8623 (mark_ref_dead): Likewise.
8624 * reginfo.c (init_subregs_of_mode): Likewise.
8625 * sched-deps.c (sched_analyze_1): Likewise.
8626 * df-scan.c (df_def_record_1): Likewise.
8627 (df_uses_record): Likewise.
8628 (df_read_modify_subreg_p): Remove in favor of...
8629 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
8630 const_rtx instead of an rtx.
8632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8633 Alan Hayward <alan.hayward@arm.com>
8634 David Sherwood <david.sherwood@arm.com>
8636 * rtl.h (partial_subreg_p): New function.
8637 * caller-save.c (save_call_clobbered_regs): Use it.
8638 * calls.c (expand_call): Likewise.
8639 * combine.c (combinable_i3pat): Likewise.
8640 (simplify_set): Likewise.
8641 (make_extraction): Likewise.
8642 (make_compound_operation_int): Likewise.
8643 (gen_lowpart_or_truncate): Likewise.
8644 (force_to_mode): Likewise.
8645 (make_field_assignment): Likewise.
8646 (reg_truncated_to_mode): Likewise.
8647 (record_truncated_value): Likewise.
8648 (move_deaths): Likewise.
8649 * cse.c (record_jump_cond): Likewise.
8650 (cse_insn): Likewise.
8651 * cselib.c (cselib_lookup_1): Likewise.
8652 * expmed.c (extract_bit_field_using_extv): Likewise.
8653 * function.c (assign_parm_setup_reg): Likewise.
8654 * ifcvt.c (noce_convert_multiple_sets): Likewise.
8655 * ira-build.c (create_insn_allocnos): Likewise.
8656 * lra-coalesce.c (merge_pseudos): Likewise.
8657 * lra-constraints.c (match_reload): Likewise.
8658 (simplify_operand_subreg): Likewise.
8659 (curr_insn_transform): Likewise.
8660 * lra-lives.c (process_bb_lives): Likewise.
8661 * lra.c (new_insn_reg): Likewise.
8662 (lra_substitute_pseudo): Likewise.
8663 * regcprop.c (mode_change_ok): Likewise.
8664 (maybe_mode_change): Likewise.
8665 (copyprop_hardreg_forward_1): Likewise.
8666 * reload.c (push_reload): Likewise.
8667 (find_reloads): Likewise.
8668 (find_reloads_subreg_address): Likewise.
8669 * reload1.c (alter_reg): Likewise.
8670 (eliminate_regs_1): Likewise.
8671 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8673 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
8675 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
8676 back to if statements, including unpack.
8678 2017-08-30 Martin Liska <mliska@suse.cz>
8681 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
8683 (func_checker::compare_asm_inputs_outputs): ... this function.
8684 (func_checker::compare_gimple_asm): Use the function to compare
8685 also ASM constrains.
8686 * ipa-icf-gimple.h: Rename the function.
8688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8689 Alan Hayward <alan.hayward@arm.com>
8690 David Sherwood <david.sherwood@arm.com>
8692 * coretypes.h (complex_mode): New type.
8693 * gdbhooks.py (build_pretty_printer): Handle it.
8694 * machmode.h (complex_mode): New class.
8695 (complex_mode::includes_p): New function.
8696 (is_complex_int_mode): Likewise.
8697 (is_complex_float_mode): Likewise.
8698 * genmodes.c (get_mode_class): Handle complex mode classes.
8699 * function.c (expand_function_end): Use is_complex_int_mode.
8701 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8702 Alan Hayward <alan.hayward@arm.com>
8703 David Sherwood <david.sherwood@arm.com>
8705 * coretypes.h (scalar_mode_pod): New typedef.
8706 * gdbhooks.py (build_pretty_printer): Handle it.
8707 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
8708 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
8709 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
8710 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
8713 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8714 Alan Hayward <alan.hayward@arm.com>
8715 David Sherwood <david.sherwood@arm.com>
8717 * machmode.h (mode_for_vector): Take a scalar_mode instead
8719 * stor-layout.c (mode_for_vector): Likewise.
8720 * explow.c (promote_mode): Use as_a <scalar_mode>.
8721 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
8723 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8724 Alan Hayward <alan.hayward@arm.com>
8725 David Sherwood <david.sherwood@arm.com>
8727 * target.def (preferred_simd_mode): Take a scalar_mode
8728 instead of a machine_mode.
8729 * targhooks.h (default_preferred_simd_mode): Likewise.
8730 * targhooks.c (default_preferred_simd_mode): Likewise.
8731 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
8732 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
8733 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
8734 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
8735 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
8736 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
8737 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
8738 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
8740 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
8741 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
8742 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
8743 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
8744 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
8745 * doc/tm.texi: Regenerate.
8746 * optabs-query.c (can_vec_mask_load_store_p): Return false for
8749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8750 Alan Hayward <alan.hayward@arm.com>
8751 David Sherwood <david.sherwood@arm.com>
8753 * target.def (scalar_mode_supported_p): Take a scalar_mode
8754 instead of a machine_mode.
8755 * targhooks.h (default_scalar_mode_supported_p): Likewise.
8756 * targhooks.c (default_scalar_mode_supported_p): Likewise.
8757 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
8758 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8759 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
8760 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
8761 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
8762 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
8763 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
8764 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
8765 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
8766 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
8767 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
8768 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
8770 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
8771 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
8772 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8773 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8774 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
8776 * doc/tm.texi: Regenerate.
8778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8779 Alan Hayward <alan.hayward@arm.com>
8780 David Sherwood <david.sherwood@arm.com>
8782 * coretypes.h (opt_scalar_mode): New typedef.
8783 * gdbhooks.py (build_pretty_printers): Handle it.
8784 * machmode.h (mode_iterator::get_2xwider): Add overload for
8786 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
8788 * expr.c (convert_mode_scalar): Likewise.
8789 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8790 * optabs.c (expand_float): Likewise.
8791 (expand_fix): Likewise.
8792 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8794 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8795 Alan Hayward <alan.hayward@arm.com>
8796 David Sherwood <david.sherwood@arm.com>
8798 * optabs.c (expand_float): Explicitly check for scalars before
8799 using a branching expansion.
8800 (expand_fix): Likewise.
8802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8803 Alan Hayward <alan.hayward@arm.com>
8804 David Sherwood <david.sherwood@arm.com>
8806 * expr.c (convert_mode): Split scalar handling out into...
8807 (convert_mode_scalar): ...this new function. Treat the modes
8810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8811 Alan Hayward <alan.hayward@arm.com>
8812 David Sherwood <david.sherwood@arm.com>
8814 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
8816 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
8818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8819 Alan Hayward <alan.hayward@arm.com>
8820 David Sherwood <david.sherwood@arm.com>
8822 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
8823 rather than a machine_mode.
8824 (fixed_from_string): Likewise.
8825 (fixed_convert): Likewise.
8826 (fixed_convert_from_int): Likewise.
8827 (fixed_convert_from_real): Likewise.
8828 (real_convert_from_fixed): Likewise.
8829 * fixed-value.c (fixed_from_double_int): Likewise.
8830 (fixed_from_string): Likewise.
8831 (fixed_convert): Likewise.
8832 (fixed_convert_from_int): Likewise.
8833 (fixed_convert_from_real): Likewise.
8834 (real_convert_from_fixed): Likewise.
8835 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
8837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8838 Alan Hayward <alan.hayward@arm.com>
8839 David Sherwood <david.sherwood@arm.com>
8841 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
8842 of separate mode class checks. Do not allow vector modes here.
8843 (immed_wide_int_const): Use as_a <scalar_mode>.
8844 * explow.c (trunc_int_for_mode): Likewise.
8845 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
8846 (wi::shwi): Likewise.
8847 (wi::min_value): Likewise.
8848 (wi::max_value): Likewise.
8849 * dwarf2out.c (loc_descriptor): Likewise.
8850 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
8853 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8854 Alan Hayward <alan.hayward@arm.com>
8855 David Sherwood <david.sherwood@arm.com>
8857 * tree.h (SCALAR_TYPE_MODE): New macro.
8858 * expr.c (expand_expr_addr_expr_1): Use it.
8859 (expand_expr_real_2): Likewise.
8860 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
8861 (fold_convert_const_fixed_from_int): Likewise.
8862 (fold_convert_const_fixed_from_real): Likewise.
8863 (native_encode_fixed): Likewise
8864 (native_encode_complex): Likewise
8865 (native_encode_vector): Likewise.
8866 (native_interpret_fixed): Likewise.
8867 (native_interpret_real): Likewise.
8868 (native_interpret_complex): Likewise.
8869 (native_interpret_vector): Likewise.
8870 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
8871 (simd_clone_adjust_argument_types): Likewise.
8872 (simd_clone_init_simd_arrays): Likewise.
8873 (simd_clone_adjust): Likewise.
8874 * stor-layout.c (layout_type): Likewise.
8875 * tree.c (build_minus_one_cst): Likewise.
8876 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
8877 * tree-inline.c (estimate_move_cost): Likewise.
8878 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
8879 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
8880 (vectorizable_reduction): Likewise.
8881 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
8882 (vect_recog_mixed_size_cond_pattern): Likewise.
8883 (check_bool_pattern): Likewise.
8884 (adjust_bool_pattern): Likewise.
8885 (search_type_for_mask_1): Likewise.
8886 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
8887 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8888 (vectorizable_load): Likewise.
8889 (vectorizable_store): Likewise.
8890 * ubsan.c (ubsan_encode_value): Likewise.
8891 * varasm.c (output_constant): Likewise.
8893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8894 Alan Hayward <alan.hayward@arm.com>
8895 David Sherwood <david.sherwood@arm.com>
8897 * coretypes.h (scalar_mode): New class.
8898 * machmode.h (scalar_mode): Likewise.
8899 (scalar_mode::includes_p): New function.
8900 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
8901 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
8902 * genmodes.c (get_mode_class): Handle remaining scalar modes.
8903 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
8904 * expmed.c (store_bit_field_1): Likewise.
8905 (extract_bit_field_1): Likewise.
8906 * expr.c (write_complex_part): Likewise.
8907 (read_complex_part): Likewise.
8908 (emit_move_complex_push): Likewise.
8909 (expand_expr_real_2): Likewise.
8910 * function.c (assign_parm_setup_reg): Likewise.
8911 (assign_parms_unsplit_complex): Likewise.
8912 * optabs.c (expand_binop): Likewise.
8913 * rtlanal.c (subreg_get_info): Likewise.
8914 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8915 * varasm.c (output_constant_pool_2): Likewise.
8917 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8918 Alan Hayward <alan.hayward@arm.com>
8919 David Sherwood <david.sherwood@arm.com>
8921 * expmed.c (extract_high_half): Use scalar_int_mode and remove
8923 (expmed_mult_highpart_optab): Likewise.
8924 (expmed_mult_highpart): Likewise.
8926 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8927 Alan Hayward <alan.hayward@arm.com>
8928 David Sherwood <david.sherwood@arm.com>
8930 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
8931 instead of a machine_mode.
8932 (builtin_memset_read_str): Likewise.
8933 * builtins.c (c_readstr): Likewise.
8934 (builtin_memcpy_read_str): Likewise.
8935 (builtin_strncpy_read_str): Likewise.
8936 (builtin_memset_read_str): Likewise.
8937 (builtin_memset_gen_str): Likewise.
8938 (expand_builtin_signbit): Use scalar_int_mode for local variables.
8939 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
8940 instead of a machine_mode.
8941 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
8943 (make_extraction): Likewise.
8944 (try_widen_shift_mode): Take and return scalar_int_modes instead
8946 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
8947 a scalar_int_mode instead of a machine_mode.
8948 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
8949 (avr_addr_space_pointer_mode): Likewise.
8950 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
8951 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
8952 (msp430_unwind_word_mode): Likewise.
8953 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
8954 (spu_addr_space_pointer_mode): Likewise.
8955 (spu_addr_space_address_mode): Likewise.
8956 (spu_libgcc_cmp_return_mode): Likewise.
8957 (spu_libgcc_shift_count_mode): Likewise.
8958 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
8959 (rl78_addr_space_pointer_mode): Likewise.
8960 (fl78_unwind_word_mode): Likewise.
8961 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
8963 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
8964 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
8965 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
8966 (mips_valid_pointer_mode): Likewise.
8967 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
8968 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
8969 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
8971 (ft32_addr_space_address_mode): Likewise.
8972 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
8973 scalar_int_mode instead of a machine_mode.
8974 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
8976 (m32c_addr_space_address_mode): Likewise.
8977 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
8978 (rs6000_eh_return_filter_mode): Likewise.
8979 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
8980 (rs6000_eh_return_filter_mode): Likewise.
8981 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
8982 (s390_libgcc_shift_count_mode): Likewise.
8983 (s390_unwind_word_mode): Likewise.
8984 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
8986 * target.def (mode_rep_extended): Likewise.
8987 (valid_pointer_mode): Likewise.
8988 (addr_space.valid_pointer_mode): Likewise.
8989 (eh_return_filter_mode): Return a scalar_int_mode rather than
8991 (libgcc_cmp_return_mode): Likewise.
8992 (libgcc_shift_count_mode): Likewise.
8993 (unwind_word_mode): Likewise.
8994 (addr_space.pointer_mode): Likewise.
8995 (addr_space.address_mode): Likewise.
8996 * doc/tm.texi: Regenerate.
8997 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
8999 (do_jump): Use scalar_int_mode for local variables.
9000 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
9001 rather than a machine_mode.
9002 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
9003 (scompare_loc_descriptor_wide): Likewise.
9004 (scompare_loc_descriptor_narrow): Likewise.
9005 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
9007 * except.c (sjlj_emit_dispatch_table): Likewise.
9008 (expand_builtin_eh_copy_values): Likewise.
9009 * explow.c (convert_memory_address_addr_space_1): Likewise.
9010 Take a scalar_int_mode rather than a machine_mode.
9011 (convert_memory_address_addr_space): Take a scalar_int_mode rather
9012 than a machine_mode.
9013 (memory_address_addr_space): Use scalar_int_mode for local variables.
9014 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
9015 rather than a machine_mode.
9016 * expmed.c (mask_rtx): Likewise.
9017 (init_expmed_one_conv): Likewise.
9018 (expand_mult_highpart_adjust): Likewise.
9019 (extract_high_half): Likewise.
9020 (expmed_mult_highpart_optab): Likewise.
9021 (expmed_mult_highpart): Likewise.
9022 (expand_smod_pow2): Likewise.
9023 (expand_sdiv_pow2): Likewise.
9024 (emit_store_flag_int): Likewise.
9025 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
9027 (extract_low_bits): Likewise.
9028 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
9030 * expr.c (pieces_addr::adjust): Likewise.
9031 (can_store_by_pieces): Likewise.
9032 (store_by_pieces): Likewise.
9033 (clear_by_pieces_1): Likewise.
9034 (expand_expr_addr_expr_1): Likewise.
9035 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
9036 (expand_expr_real_1): Likewise.
9037 (try_casesi): Likewise.
9038 * final.c (shorten_branches): Likewise.
9039 * fold-const.c (fold_convert_const_int_from_fixed): Change the
9040 type of "mode" to machine_mode.
9041 * internal-fn.c (expand_arith_overflow_result_store): Take a
9042 scalar_int_mode rather than a machine_mode.
9043 (expand_mul_overflow): Use scalar_int_mode for local variables.
9044 * loop-doloop.c (doloop_modify): Likewise.
9045 (doloop_optimize): Likewise.
9046 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
9047 than a machine_mode.
9048 (expand_doubleword_shift_condmove): Likewise.
9049 (expand_doubleword_shift): Likewise.
9050 (expand_doubleword_clz): Likewise.
9051 (expand_doubleword_popcount): Likewise.
9052 (expand_doubleword_parity): Likewise.
9053 (expand_absneg_bit): Use scalar_int_mode for local variables.
9054 (prepare_float_lib_cmp): Likewise.
9055 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
9056 rather than a machine_mode.
9057 (convert_memory_address_addr_space): Likewise.
9058 (get_mode_bounds): Likewise.
9059 (get_address_mode): Return a scalar_int_mode rather than a
9061 * rtlanal.c (get_address_mode): Likewise.
9062 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
9063 than a machine_mode.
9064 * targhooks.c (default_mode_rep_extended): Likewise.
9065 (default_valid_pointer_mode): Likewise.
9066 (default_addr_space_valid_pointer_mode): Likewise.
9067 (default_eh_return_filter_mode): Return a scalar_int_mode rather
9068 than a machine_mode.
9069 (default_libgcc_cmp_return_mode): Likewise.
9070 (default_libgcc_shift_count_mode): Likewise.
9071 (default_unwind_word_mode): Likewise.
9072 (default_addr_space_pointer_mode): Likewise.
9073 (default_addr_space_address_mode): Likewise.
9074 * targhooks.h (default_eh_return_filter_mode): Likewise.
9075 (default_libgcc_cmp_return_mode): Likewise.
9076 (default_libgcc_shift_count_mode): Likewise.
9077 (default_unwind_word_mode): Likewise.
9078 (default_addr_space_pointer_mode): Likewise.
9079 (default_addr_space_address_mode): Likewise.
9080 (default_mode_rep_extended): Take a scalar_int_mode rather than
9082 (default_valid_pointer_mode): Likewise.
9083 (default_addr_space_valid_pointer_mode): Likewise.
9084 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
9086 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
9087 rather than a machine_mode.
9088 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
9089 for local variables.
9090 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9091 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
9092 than a machine_mode.
9094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9095 Alan Hayward <alan.hayward@arm.com>
9096 David Sherwood <david.sherwood@arm.com>
9098 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
9099 the mode argument to scalar_int_mode.
9100 (do_jump_by_parts_zero_rtx): Likewise.
9101 (do_jump_by_parts_equality_rtx): Likewise.
9102 (do_jump_by_parts_greater): Take a mode argument.
9103 (do_jump_by_parts_equality): Likewise.
9104 (do_jump_1): Update calls accordingly.
9106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9107 Alan Hayward <alan.hayward@arm.com>
9108 David Sherwood <david.sherwood@arm.com>
9110 * is-a.h (safe_dyn_cast): New function.
9111 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
9112 (jump_table_for_label): Likewise.
9113 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
9114 instead of an rtx_insn *.
9115 (shorten_branches): Use dyn_cast instead of LABEL_P and
9116 JUMP_TABLE_DATA_P. Use jump_table_for_label and
9117 rtx_jump_table_data::get_data_mode.
9118 (final_scan_insn): Likewise.
9120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9121 Alan Hayward <alan.hayward@arm.com>
9122 David Sherwood <david.sherwood@arm.com>
9124 * combine.c (try_combine): Use is_a <scalar_int_mode> when
9125 trying to combine a full-register integer set with a subreg
9128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9129 Alan Hayward <alan.hayward@arm.com>
9130 David Sherwood <david.sherwood@arm.com>
9132 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
9133 that is always either address_mode or pointer_mode.
9135 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9136 Alan Hayward <alan.hayward@arm.com>
9137 David Sherwood <david.sherwood@arm.com>
9139 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
9140 when the two are known to be equal.
9142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9143 Alan Hayward <alan.hayward@arm.com>
9144 David Sherwood <david.sherwood@arm.com>
9146 * simplify-rtx.c (simplify_const_unary_operation): Use
9147 is_a <scalar_int_mode> instead of checking for a nonzero
9148 precision. Forcibly convert op_mode to a scalar_int_mode
9149 in that case. More clearly differentiate the operand and
9150 result modes and use the former when deciding what the value
9151 of a count-bits operation should be. Use is_int_mode instead
9152 of checking for a MODE_INT. Remove redundant check for whether
9153 this mode has a zero precision.
9155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9156 Alan Hayward <alan.hayward@arm.com>
9157 David Sherwood <david.sherwood@arm.com>
9159 * optabs.c (widen_leading): Change the type of the mode argument
9160 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
9161 (widen_bswap): Likewise.
9162 (expand_parity): Likewise.
9163 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
9164 (expand_ffs): Likewise.
9165 (epand_unop): Check for scalar integer modes before calling the
9168 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9169 Alan Hayward <alan.hayward@arm.com>
9170 David Sherwood <david.sherwood@arm.com>
9172 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
9174 (expand_expr_real_1): Update call accordingly.
9176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9177 Alan Hayward <alan.hayward@arm.com>
9178 David Sherwood <david.sherwood@arm.com>
9180 * expmed.c (store_bit_field_using_insv): Add op0_mode and
9181 value_mode arguments. Use scalar_int_mode internally.
9182 (store_bit_field_1): Rename the new integer mode from imode
9183 to op0_mode and use it instead of GET_MODE (op0). Update calls
9184 to store_split_bit_field, store_bit_field_using_insv and
9185 store_fixed_bit_field.
9186 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
9187 Use scalar_int_mode internally. Use a bit count rather than a mode
9188 when calculating the largest bit size for get_best_mode.
9189 Update calls to store_split_bit_field and store_fixed_bit_field_1.
9190 (store_fixed_bit_field_1): Add mode and value_mode arguments.
9191 Remove assertion that OP0 has a scalar integer mode.
9192 (store_split_bit_field): Add op0_mode and value_mode arguments.
9193 Update calls to extract_fixed_bit_field.
9194 (extract_bit_field_using_extv): Add an op0_mode argument.
9195 Use scalar_int_mode internally.
9196 (extract_bit_field_1): Rename the new integer mode from imode to
9197 op0_mode and use it instead of GET_MODE (op0). Update calls to
9198 extract_split_bit_field, extract_bit_field_using_extv and
9199 extract_fixed_bit_field.
9200 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
9201 to extract_split_bit_field and extract_fixed_bit_field_1.
9202 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
9203 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
9205 (extract_split_bit_field): Add an op0_mode argument. Update call
9206 to extract_fixed_bit_field.
9208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9209 Alan Hayward <alan.hayward@arm.com>
9210 David Sherwood <david.sherwood@arm.com>
9212 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
9213 * explow.c (hard_function_value): Likewise.
9214 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
9215 convert_to_mode call outside the loop.
9216 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
9217 for the mode iterator. Require the mode specified by max_pieces
9219 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
9221 (copy_blkmode_to_reg): Likewise.
9222 (set_storage_via_setmem): Likewise.
9223 * optabs.c (prepare_cmp_insn): Likewise.
9224 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9225 * stor-layout.c (finish_bitfield_representative): Likewise.
9227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9228 Alan Hayward <alan.hayward@arm.com>
9229 David Sherwood <david.sherwood@arm.com>
9231 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
9232 * expr.c (convert_move): Use them.
9233 (convert_modes): Likewise.
9234 (store_expr_with_bounds): Likewise.
9236 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9237 Alan Hayward <alan.hayward@arm.com>
9238 David Sherwood <david.sherwood@arm.com>
9240 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
9241 parameter for the mode of "x". Remove the "known_x", "known_mode"
9242 and "known_ret" arguments. Change the type of the mode argument
9244 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
9245 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
9246 (reg_num_sign_bit_copies_for_combine): Likewise.
9247 * rtlanal.c (nonzero_bits1): Likewise.
9248 (num_sign_bit_copies1): Likewise.
9249 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
9250 (reg_num_sign_bit_copies_general): Likewise.
9251 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
9252 (reg_nonzero_bits_general): Likewise.
9254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9255 Alan Hayward <alan.hayward@arm.com>
9256 David Sherwood <david.sherwood@arm.com>
9258 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
9259 than in subroutines. Return 1 for non-integer modes.
9260 (cached_num_sign_bit_copies): Change the type of the mode parameter
9262 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
9263 classes. Handle CONST_INT_P first and then check whether X also
9264 has a scalar integer mode. Check the same thing for inner registers
9265 of a SUBREG and for values that are being extended or truncated.
9267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9268 Alan Hayward <alan.hayward@arm.com>
9269 David Sherwood <david.sherwood@arm.com>
9271 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
9272 in subroutines. Return the mode mask for non-integer modes.
9273 (cached_nonzero_bits): Change the type of the mode parameter
9275 (nonzero_bits1): Likewise. Remove early exit for other mode
9276 classes. Handle CONST_INT_P first and then check whether X
9277 also has a scalar integer mode.
9279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9280 Alan Hayward <alan.hayward@arm.com>
9281 David Sherwood <david.sherwood@arm.com>
9283 * expr.c (widest_int_mode_for_size): Make the comment match the code.
9284 Return a scalar_int_mode and assert that the size is greater than
9286 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
9287 (op_by_pieces_d::op_by_pieces_d): Likewise.
9288 (op_by_pieces_d::run): Likewise.
9289 (can_store_by_pieces): Likewise.
9291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9292 Alan Hayward <alan.hayward@arm.com>
9293 David Sherwood <david.sherwood@arm.com>
9295 * combine.c (extract_left_shift): Add a mode argument and update
9297 (make_compound_operation_int): Change the type of the mode parameter
9298 to scalar_int_mode and update the call to extract_left_shift.
9300 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9301 Alan Hayward <alan.hayward@arm.com>
9302 David Sherwood <david.sherwood@arm.com>
9304 * combine.c (simplify_and_const_int): Change the type of the mode
9305 parameter to scalar_int_mode.
9306 (simplify_and_const_int_1): Likewise. Update recursive call.
9308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9309 Alan Hayward <alan.hayward@arm.com>
9310 David Sherwood <david.sherwood@arm.com>
9312 * combine.c (simplify_compare_const): Check that the mode is a
9313 scalar_int_mode (rather than VOIDmode) before testing its
9315 (simplify_comparison): Move COMPARISON_P handling out of the
9316 loop and restrict the latter part of the loop to scalar_int_modes.
9317 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
9318 and when considering SUBREG_REGs. Use is_int_mode instead of
9319 checking GET_MODE_CLASS against MODE_INT.
9321 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9322 Alan Hayward <alan.hayward@arm.com>
9323 David Sherwood <david.sherwood@arm.com>
9325 * combine.c (try_widen_shift_mode): Move check for equal modes to...
9326 (simplify_shift_const_1): ...here. Use scalar_int_mode for
9327 shift_unit_mode and for modes involved in scalar shifts.
9329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9330 Alan Hayward <alan.hayward@arm.com>
9331 David Sherwood <david.sherwood@arm.com>
9333 * combine.c (force_int_to_mode): New function, split out from...
9334 (force_to_mode): ...here. Keep xmode up-to-date and use it
9335 instead of GET_MODE (x).
9337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9338 Alan Hayward <alan.hayward@arm.com>
9339 David Sherwood <david.sherwood@arm.com>
9341 * optabs-query.h (extraction_insn::struct_mode): Change type to
9342 opt_scalar_int_mode and update comment.
9343 (extraction_insn::field_mode): Change type to scalar_int_mode.
9344 (extraction_insn::pos_mode): Likewise.
9345 * combine.c (make_extraction): Update accordingly.
9346 * optabs-query.c (get_traditional_extraction_insn): Likewise.
9347 (get_optab_extraction_insn): Likewise.
9348 * recog.c (simplify_while_replacing): Likewise.
9349 * expmed.c (narrow_bit_field_mem): Change the type of the mode
9350 parameter to opt_scalar_int_mode.
9352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9353 Alan Hayward <alan.hayward@arm.com>
9354 David Sherwood <david.sherwood@arm.com>
9356 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
9357 to a scalar_int_mode instead of a machine_mode.
9358 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
9359 (get_best_mode): Return a boolean and use a pointer argument to store
9360 the selected mode. Replace the limit mode parameter with a bit limit.
9361 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
9362 for the values returned by bit_field_mode_iterator::next_mode.
9363 (store_bit_field): Update call to get_best_mode.
9364 (store_fixed_bit_field): Likewise.
9365 (extract_fixed_bit_field): Likewise.
9366 * expr.c (optimize_bitfield_assignment_op): Likewise.
9367 * fold-const.c (optimize_bit_field_compare): Likewise.
9368 (fold_truth_andor_1): Likewise.
9369 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
9370 Update for new type of m_mode.
9371 (get_best_mode): As above.
9373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9374 Alan Hayward <alan.hayward@arm.com>
9375 David Sherwood <david.sherwood@arm.com>
9377 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
9378 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
9379 (store_bit_field): Check is_a <scalar_int_mode> before calling
9380 strict_volatile_bitfield_p.
9381 (extract_bit_field): Likewise.
9383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9384 Alan Hayward <alan.hayward@arm.com>
9385 David Sherwood <david.sherwood@arm.com>
9387 * target.def (cstore_mode): Return a scalar_int_mode.
9388 * doc/tm.texi: Regenerate.
9389 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
9390 * targhooks.h (default_cstore_mode): Likewise.
9391 * targhooks.c (default_cstore_mode): Likewise, using a forced
9393 * expmed.c (emit_cstore): Expect the target of the cstore to be
9396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9397 Alan Hayward <alan.hayward@arm.com>
9398 David Sherwood <david.sherwood@arm.com>
9400 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
9402 (niter_desc): Likewise mode.
9403 (iv_analyze): Add a mode parameter.
9405 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
9406 and change its type to scalar_int_mode.
9407 * loop-iv.c: Update commentary at head of file.
9408 (iv_constant): Pass the mode paraeter before the rtx it describes
9409 and change its type to scalar_int_mode. Remove VOIDmode handling.
9410 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
9411 (iv_extend): Likewise.
9412 (shorten_into_mode): Likewise.
9413 (iv_add): Use scalar_int_mode.
9414 (iv_mult): Likewise.
9415 (iv_shift): Likewise.
9416 (canonicalize_iv_subregs): Likewise.
9417 (get_biv_step_1): Pass the outer_mode parameter before the rtx
9418 it describes and change its mode to scalar_int_mode. Also change
9419 the type of the returned inner_mode to scalar_int_mode.
9420 (get_biv_step): Likewise, turning outer_mode from a pointer
9421 into a direct parameter. Update call to get_biv_step_1.
9422 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
9423 iv_constant and get_biv_step.
9424 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
9425 and change its type to scalar_int_mode. Don't initialise iv->mode
9426 to VOIDmode and remove later checks for its still being VOIDmode.
9427 Update calls to iv_analyze_op and iv_analyze_expr. Check
9428 is_a <scalar_int_mode> when changing the mode under consideration.
9429 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
9430 Update call to iv_analyze_expr.
9431 (iv_analyze_op): Add a mode parameter. Reject subregs whose
9432 inner register is not also a scalar_int_mode. Update call to
9434 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
9435 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
9436 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
9437 separate mode class checks. Update calls to iv_analyze. Remove
9438 fix-up of VOIDmodes after iv_analyze_biv.
9439 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
9440 don't have a scalar_int_mode. Update call to biv_p.
9442 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9443 Alan Hayward <alan.hayward@arm.com>
9444 David Sherwood <david.sherwood@arm.com>
9446 * cfgexpand.c (convert_debug_memory_address): Use
9447 as_a <scalar_int_mode>.
9448 * combine.c (expand_compound_operation): Likewise.
9449 (make_extraction): Likewise.
9450 (change_zero_ext): Likewise.
9451 (simplify_comparison): Likewise.
9452 * cse.c (cse_insn): Likewise.
9453 * dwarf2out.c (minmax_loc_descriptor): Likewise.
9454 (mem_loc_descriptor): Likewise.
9455 (loc_descriptor): Likewise.
9456 * expmed.c (init_expmed_one_mode): Likewise.
9457 (synth_mult): Likewise.
9458 (emit_store_flag_1): Likewise.
9459 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
9460 of a comparison with size.
9461 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
9462 (reduce_to_bit_field_precision): Likewise.
9463 * function.c (expand_function_end): Likewise.
9464 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
9465 * loop-doloop.c (doloop_modify): Likewise.
9466 * optabs.c (expand_binop): Likewise.
9467 (expand_unop): Likewise.
9468 (expand_copysign_absneg): Likewise.
9469 (prepare_cmp_insn): Likewise.
9470 (maybe_legitimize_operand): Likewise.
9471 * recog.c (const_scalar_int_operand): Likewise.
9472 * rtlanal.c (get_address_mode): Likewise.
9473 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9474 (simplify_cond_clz_ctz): Likewise.
9475 * tree-nested.c (get_nl_goto_field): Likewise.
9476 * tree.c (build_vector_type_for_mode): Likewise.
9477 * var-tracking.c (use_narrower_mode): Likewise.
9479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9480 Alan Hayward <alan.hayward@arm.com>
9481 David Sherwood <david.sherwood@arm.com>
9483 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
9484 * builtins.c (expand_builtin_signbit): Use it.
9485 * cfgexpand.c (expand_debug_expr): Likewise.
9486 * dojump.c (do_jump): Likewise.
9487 (do_compare_and_jump): Likewise.
9488 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
9489 * expmed.c (make_tree): Likewise.
9490 * expr.c (expand_expr_real_2): Likewise.
9491 (expand_expr_real_1): Likewise.
9492 (try_casesi): Likewise.
9493 * fold-const-call.c (fold_const_call_ss): Likewise.
9494 * fold-const.c (unextend): Likewise.
9495 (extract_muldiv_1): Likewise.
9496 (fold_single_bit_test): Likewise.
9497 (native_encode_int): Likewise.
9498 (native_encode_string): Likewise.
9499 (native_interpret_int): Likewise.
9500 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
9501 * internal-fn.c (expand_addsub_overflow): Likewise.
9502 (expand_neg_overflow): Likewise.
9503 (expand_mul_overflow): Likewise.
9504 (expand_arith_overflow): Likewise.
9505 * match.pd: Likewise.
9506 * stor-layout.c (layout_type): Likewise.
9507 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9508 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
9509 * tree-ssanames.c (get_range_info): Likewise.
9510 * tree-switch-conversion.c (array_value_type) Likewise.
9511 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
9512 (vect_recog_divmod_pattern): Likewise.
9513 (vect_recog_mixed_size_cond_pattern): Likewise.
9514 * tree-vrp.c (extract_range_basic): Likewise.
9515 (simplify_float_conversion_using_ranges): Likewise.
9516 * tree.c (int_fits_type_p): Likewise.
9517 * ubsan.c (instrument_bool_enum_load): Likewise.
9518 * varasm.c (mergeable_string_section): Likewise.
9519 (narrowing_initializer_constant_valid_p): Likewise.
9520 (output_constant): Likewise.
9522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9523 Alan Hayward <alan.hayward@arm.com>
9524 David Sherwood <david.sherwood@arm.com>
9526 * machmode.h (NARROWEST_INT_MODE): New macro.
9527 * expr.c (alignment_for_piecewise_move): Use it instead of
9528 GET_CLASS_NARROWEST_MODE (MODE_INT).
9529 (push_block): Likewise.
9530 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
9532 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9534 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9535 Alan Hayward <alan.hayward@arm.com>
9536 David Sherwood <david.sherwood@arm.com>
9538 * postreload.c (move2add_valid_value_p): Change the type of the
9539 mode parameter to scalar_int_mode.
9540 (move2add_use_add2_insn): Add a mode parameter and use it instead
9542 (move2add_use_add3_insn): Likewise.
9543 (reload_cse_move2add): Update accordingly.
9545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9546 Alan Hayward <alan.hayward@arm.com>
9547 David Sherwood <david.sherwood@arm.com>
9549 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
9551 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
9552 * optabs.c (expand_unop): Likewise.
9554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9555 Alan Hayward <alan.hayward@arm.com>
9556 David Sherwood <david.sherwood@arm.com>
9558 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
9559 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
9560 (popcount_loc_descriptor): Likewise.
9561 (bswap_loc_descriptor): Likewise.
9562 (rotate_loc_descriptor): Likewise.
9563 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
9564 calling the functions above.
9566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9567 Alan Hayward <alan.hayward@arm.com>
9568 David Sherwood <david.sherwood@arm.com>
9570 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
9572 (try_combine): Likewise.
9573 (simplify_if_then_else): Likewise.
9574 * cse.c (cse_insn): Likewise.
9575 * dwarf2out.c (mem_loc_descriptor): Likewise.
9576 * emit-rtl.c (gen_lowpart_common): Likewise.
9577 * simplify-rtx.c (simplify_truncation): Likewise.
9578 (simplify_binary_operation_1): Likewise.
9579 (simplify_const_relational_operation): Likewise.
9580 (simplify_ternary_operation): Likewise.
9581 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
9583 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9584 Alan Hayward <alan.hayward@arm.com>
9585 David Sherwood <david.sherwood@arm.com>
9587 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
9588 * reload.c (push_reload): Likewise.
9589 (find_reloads): Likewise.
9591 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9592 Alan Hayward <alan.hayward@arm.com>
9593 David Sherwood <david.sherwood@arm.com>
9595 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
9596 (make_compound_operation_int): Likewise.
9597 (change_zero_ext): Likewise.
9598 * expr.c (convert_move): Likewise.
9599 (convert_modes): Likewise.
9600 * fwprop.c (forward_propagate_subreg): Likewise.
9601 * loop-iv.c (get_biv_step_1): Likewise.
9602 * optabs.c (widen_operand): Likewise.
9603 * postreload.c (move2add_valid_value_p): Likewise.
9604 * recog.c (simplify_while_replacing): Likewise.
9605 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9606 (simplify_binary_operation_1): Likewise. Remove redundant
9607 mode equality check.
9609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9610 Alan Hayward <alan.hayward@arm.com>
9611 David Sherwood <david.sherwood@arm.com>
9613 * combine.c (combine_simplify_rtx): Add checks for
9614 is_a <scalar_int_mode>.
9615 (simplify_if_then_else): Likewise.
9616 (make_field_assignment): Likewise.
9617 (simplify_comparison): Likewise.
9618 * ifcvt.c (noce_try_bitop): Likewise.
9619 * loop-invariant.c (canonicalize_address_mult): Likewise.
9620 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9623 Alan Hayward <alan.hayward@arm.com>
9624 David Sherwood <david.sherwood@arm.com>
9626 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
9627 is_a <scalar_int_mode> instead of != BLKmode.
9629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9630 Alan Hayward <alan.hayward@arm.com>
9631 David Sherwood <david.sherwood@arm.com>
9633 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
9634 instead of != VOIDmode.
9635 * combine.c (if_then_else_cond): Likewise.
9636 (change_zero_ext): Likewise.
9637 * dwarf2out.c (mem_loc_descriptor): Likewise.
9638 (loc_descriptor): Likewise.
9639 * rtlanal.c (canonicalize_condition): Likewise.
9640 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
9642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9643 Alan Hayward <alan.hayward@arm.com>
9644 David Sherwood <david.sherwood@arm.com>
9646 * simplify-rtx.c (simplify_binary_operation_1): Use
9647 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
9649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9650 Alan Hayward <alan.hayward@arm.com>
9651 David Sherwood <david.sherwood@arm.com>
9653 * wide-int.h (int_traits<unsigned char>) New class.
9654 (int_traits<unsigned short>) Likewise.
9655 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
9656 Use GET_MODE_UNIT_PRECISION and remove redundant test for
9658 * combine.c (set_nonzero_bits_and_sign_copies): Use
9659 is_a <scalar_int_mode>.
9660 (find_split_point): Likewise.
9661 (combine_simplify_rtx): Likewise.
9662 (simplify_logical): Likewise.
9663 (expand_compound_operation): Likewise.
9664 (expand_field_assignment): Likewise.
9665 (make_compound_operation): Likewise.
9666 (extended_count): Likewise.
9667 (change_zero_ext): Likewise.
9668 (simplify_comparison): Likewise.
9669 * dwarf2out.c (scompare_loc_descriptor): Likewise.
9670 (ucompare_loc_descriptor): Likewise.
9671 (minmax_loc_descriptor): Likewise.
9672 (mem_loc_descriptor): Likewise.
9673 (loc_descriptor): Likewise.
9674 * expmed.c (init_expmed_one_mode): Likewise.
9675 * lra-constraints.c (lra_constraint_offset): Likewise.
9676 * optabs.c (prepare_libcall_arg): Likewise.
9677 * postreload.c (move2add_note_store): Likewise.
9678 * reload.c (operands_match_p): Likewise.
9679 * rtl.h (load_extend_op): Likewise.
9680 * rtlhooks.c (gen_lowpart_general): Likewise.
9681 * simplify-rtx.c (simplify_truncation): Likewise.
9682 (simplify_unary_operation_1): Likewise.
9683 (simplify_binary_operation_1): Likewise.
9684 (simplify_const_binary_operation): Likewise.
9685 (simplify_const_relational_operation): Likewise.
9686 (simplify_subreg): Likewise.
9687 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9688 * var-tracking.c (adjust_mems): Likewise.
9689 (prepare_call_arguments): Likewise.
9691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9692 Alan Hayward <alan.hayward@arm.com>
9693 David Sherwood <david.sherwood@arm.com>
9695 * machmode.h (is_int_mode): New fuction.
9696 * combine.c (find_split_point): Use it.
9697 (combine_simplify_rtx): Likewise.
9698 (simplify_if_then_else): Likewise.
9699 (simplify_set): Likewise.
9700 (simplify_shift_const_1): Likewise.
9701 (simplify_comparison): Likewise.
9702 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
9703 * cse.c (notreg_cost): Likewise.
9704 (cse_insn): Likewise.
9705 * cselib.c (cselib_lookup_1): Likewise.
9706 * dojump.c (do_jump_1): Likewise.
9707 (do_compare_rtx_and_jump): Likewise.
9708 * dse.c (get_call_args): Likewise.
9709 * dwarf2out.c (rtl_for_decl_init): Likewise.
9710 (native_encode_initializer): Likewise.
9711 * expmed.c (emit_store_flag_1): Likewise.
9712 (emit_store_flag): Likewise.
9713 * expr.c (convert_modes): Likewise.
9714 (store_field): Likewise.
9715 (expand_expr_real_1): Likewise.
9716 * fold-const.c (fold_read_from_constant_string): Likewise.
9717 * gimple-ssa-sprintf.c (get_format_string): Likewise.
9718 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
9719 * optabs.c (expand_binop): Likewise.
9720 (expand_unop): Likewise.
9721 (expand_abs_nojump): Likewise.
9722 (expand_one_cmpl_abs_nojump): Likewise.
9723 * simplify-rtx.c (mode_signbit_p): Likewise.
9724 (val_signbit_p): Likewise.
9725 (val_signbit_known_set_p): Likewise.
9726 (val_signbit_known_clear_p): Likewise.
9727 (simplify_relational_operation_1): Likewise.
9728 * tree.c (vector_type_mode): Likewise.
9730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9731 Alan Hayward <alan.hayward@arm.com>
9732 David Sherwood <david.sherwood@arm.com>
9734 * machmode.h (smallest_mode_for_size): Fix formatting.
9735 (smallest_int_mode_for_size): New function.
9736 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
9737 instead of smallest_mode_for_size.
9738 * combine.c (make_extraction): Likewise.
9739 * config/arc/arc.c (arc_expand_movmem): Likewise.
9740 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
9741 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9742 * config/s390/s390.c (s390_expand_insv): Likewise.
9743 * config/sparc/sparc.c (assign_int_registers): Likewise.
9744 * config/spu/spu.c (spu_function_value): Likewise.
9745 (spu_function_arg): Likewise.
9746 * coverage.c (get_gcov_type): Likewise.
9747 (get_gcov_unsigned_t): Likewise.
9748 * dse.c (find_shift_sequence): Likewise.
9749 * expmed.c (store_bit_field_1): Likewise.
9750 * expr.c (convert_move): Likewise.
9751 (store_field): Likewise.
9752 * internal-fn.c (expand_arith_overflow): Likewise.
9753 * optabs-query.c (get_best_extraction_insn): Likewise.
9754 * optabs.c (expand_twoval_binop_libfunc): Likewise.
9755 * stor-layout.c (layout_type): Likewise.
9756 (initialize_sizetypes): Likewise.
9757 * targhooks.c (default_get_mask_mode): Likewise.
9758 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
9760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9761 Alan Hayward <alan.hayward@arm.com>
9762 David Sherwood <david.sherwood@arm.com>
9764 * machmode.h (opt_mode::else_blk): New function.
9765 (int_mode_for_mode): Declare.
9766 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
9767 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
9769 * cfgexpand.c (expand_debug_expr): Likewise.
9770 * combine.c (gen_lowpart_or_truncate): Likewise.
9771 (gen_lowpart_for_combine): Likewise.
9772 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
9773 * config/avr/avr.c (avr_to_int_mode): Likewise.
9774 (avr_out_plus_1): Likewise.
9775 (avr_out_plus): Likewise.
9776 (avr_out_round): Likewise.
9777 * config/i386/i386.c (ix86_split_to_parts): Likewise.
9778 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9779 (s390_expand_vcond): Likewise.
9780 * config/spu/spu.c (spu_split_immediate): Likewise.
9781 (spu_expand_mov): Likewise.
9782 * dse.c (get_stored_val): Likewise.
9783 * expmed.c (store_bit_field_1): Likewise.
9784 (convert_extracted_bit_field): Use int_mode_for_mode instead of
9786 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
9787 (extract_low_bits): Likewise.
9788 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
9789 handling rather than repeating the check.
9790 (emit_group_store): Likewise.
9791 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
9792 * optabs.c (expand_absneg_bit): Likewise.
9793 (expand_copysign_absneg): Likewise.
9794 (expand_copysign_bit): Likewise.
9795 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
9796 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
9797 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
9798 * var-tracking.c (prepare_call_arguments): Likewise.
9799 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
9800 int_mode_for_mode instead of mode_for_size.
9801 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9804 Alan Hayward <alan.hayward@arm.com>
9805 David Sherwood <david.sherwood@arm.com>
9807 * machmode.h (int_mode_for_size): New function.
9808 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
9809 instead of mode_for_size.
9810 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
9812 * combine.c (expand_field_assignment): Use int_mode_for_size
9813 instead of mode_for_size.
9814 (make_extraction): Likewise.
9815 (simplify_shift_const_1): Likewise.
9816 (simplify_comparison): Likewise.
9817 * dojump.c (do_jump): Likewise.
9818 * dwarf2out.c (mem_loc_descriptor): Likewise.
9819 * emit-rtl.c (init_derived_machine_modes): Likewise.
9820 * expmed.c (flip_storage_order): Likewise.
9821 (convert_extracted_bit_field): Likewise.
9822 * expr.c (copy_blkmode_from_reg): Likewise.
9823 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
9824 * internal-fn.c (expand_mul_overflow): Likewise.
9825 * lower-subreg.c (simple_move): Likewise.
9826 * optabs-libfuncs.c (init_optabs): Likewise.
9827 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9828 * tree.c (vector_type_mode): Likewise.
9829 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
9830 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
9831 * tree-vect-generic.c (expand_vector_parallel): Likewise.
9832 * tree-vect-stmts.c (vectorizable_load): Likewise.
9833 (vectorizable_store): Likewise.
9835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9836 Alan Hayward <alan.hayward@arm.com>
9837 David Sherwood <david.sherwood@arm.com>
9839 * coretypes.h (pod_mode): New type.
9840 (scalar_int_mode_pod): New typedef.
9841 * machmode.h (pod_mode): New class.
9842 (int_n_data_t::m): Change type to scalar_int_mode_pod.
9843 * genmodes.c (emit_mode_int_n): Update accordingly.
9844 * lower-subreg.h (target_lower_subreg): Change type to
9845 scalar_int_mode_pod.
9846 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
9847 scalar_int_mode_pod.
9849 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9850 Alan Hayward <alan.hayward@arm.com>
9851 David Sherwood <david.sherwood@arm.com>
9853 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
9854 machine_mode to scalar_int_mode.
9855 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
9856 (rs6000_option_override_internal): Remove cast to int.
9857 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
9858 machine_mode to scalar_int_mode.
9859 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
9860 (rs6000_option_override_internal): Remove cast to int.
9861 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
9862 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
9864 * config/s390/s390.c (s390_expand_builtin): Likewise.
9865 * coretypes.h (scalar_int_mode): New type.
9866 (opt_scalar_int_mode): New typedef.
9867 * machmode.h (scalar_int_mode): New class.
9868 (scalar_int_mode::includes_p): New function.
9869 (byte_mode): Change type to scalar_int_mode.
9870 (word_mode): Likewise.
9871 (ptr_mode): Likewise.
9872 * emit-rtl.c (byte_mode): Likewise.
9873 (word_mode): Likewise.
9874 (ptr_mode): Likewise.
9875 (init_derived_machine_modes): Update accordingly.
9876 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
9877 and MODE_PARTIAL_INT.
9878 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
9879 opt_scalar_int_mode.
9881 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9882 Alan Hayward <alan.hayward@arm.com>
9883 David Sherwood <david.sherwood@arm.com>
9885 * target.def (libgcc_floating_mode_supported_p): Take a
9887 * doc/tm.texi: Regenerate.
9888 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
9890 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
9891 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
9894 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9895 Alan Hayward <alan.hayward@arm.com>
9896 David Sherwood <david.sherwood@arm.com>
9898 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
9899 * doc/tm.texi: Regenerate.
9900 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
9901 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
9902 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
9903 * targhooks.h (default_floatn_mode): Likewise.
9904 * targhooks.c (default_floatn_mode): Likewise.
9905 * tree.c (build_common_tree_nodes): Update accordingly.
9907 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9908 Alan Hayward <alan.hayward@arm.com>
9909 David Sherwood <david.sherwood@arm.com>
9911 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
9912 (mode_iterator::iterate_p): Likewise.
9913 (mode_iterator::get_wider): Likewise.
9914 * expr.c (init_expr_target): Use opt_scalar_float_mode.
9916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9917 Alan Hayward <alan.hayward@arm.com>
9918 David Sherwood <david.sherwood@arm.com>
9920 * coretypes.h (opt_scalar_float_mode): New typedef.
9921 * machmode.h (float_mode_for_size): New function.
9922 * emit-rtl.c (double_mode): Delete.
9923 (init_emit_once): Use float_mode_for_size.
9924 * stor-layout.c (layout_type): Likewise.
9925 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
9927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9928 Alan Hayward <alan.hayward@arm.com>
9929 David Sherwood <david.sherwood@arm.com>
9931 * output.h (assemble_real): Take a scalar_float_mode.
9932 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
9933 * config/arm/arm.md (consttable_4): Likewise.
9934 (consttable_8): Likewise.
9935 (consttable_16): Likewise.
9936 * config/mips/mips.md (consttable_float): Likewise.
9937 * config/s390/s390.c (s390_output_pool_entry): Likewise.
9938 * varasm.c (assemble_real): Take a scalar_float_mode.
9939 (output_constant_pool_2): Update accordingly.
9940 (output_constant): Likewise.
9942 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9943 Alan Hayward <alan.hayward@arm.com>
9944 David Sherwood <david.sherwood@arm.com>
9946 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
9947 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
9948 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
9949 (native_encode_real): Likewise.
9950 (native_interpret_real): Likewise.
9951 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
9952 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9954 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9955 Alan Hayward <alan.hayward@arm.com>
9956 David Sherwood <david.sherwood@arm.com>
9958 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
9959 <scalar_float_mode>. Simplify.
9960 (gen_extend_conv_libfunc): Likewise.
9962 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9963 Alan Hayward <alan.hayward@arm.com>
9964 David Sherwood <david.sherwood@arm.com>
9966 * coretypes.h (scalar_float_mode): New type.
9967 * machmode.h (mode_traits::from_int): Use machine_mode if
9968 USE_ENUM_MODES is defined.
9969 (is_a): New function.
9971 (dyn_cast): Likewise.
9972 (scalar_float_mode): New class.
9973 (scalar_float_mode::includes_p): New function.
9974 (is_float_mode): Likewise.
9975 * gdbhooks.py (MachineModePrinter): New class.
9976 (build_pretty_printer): Use it for scalar_float_mode.
9977 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
9978 (format_helper::format_helper): Turn into a template.
9979 * genmodes.c (get_mode_class): New function.
9980 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
9981 or machine_mode if none.
9982 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
9983 as_a <scalar_float_mode>.
9984 * dwarf2out.c (mem_loc_descriptor): Likewise.
9985 (insert_float): Likewise.
9986 (add_const_value_attribute): Likewise.
9987 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9988 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
9989 (expand_unop): Update accordingly.
9990 (expand_abs_nojump): Likewise.
9991 (expand_copysign_absneg): Take a scalar_float_mode.
9992 (expand_copysign_bit): Likewise.
9993 (expand_copysign): Update accordingly.
9995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9996 Alan Hayward <alan.hayward@arm.com>
9997 David Sherwood <david.sherwood@arm.com>
9999 * coretypes.h (opt_mode): New class.
10000 * machmode.h (opt_mode): Likewise.
10001 (opt_mode::else_void): New function.
10002 (opt_mode::require): Likewise.
10003 (opt_mode::exists): Likewise.
10004 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
10005 (GET_MODE_2XWIDER_MODE): Likewise.
10006 (mode_iterator::get_wider): Update accordingly.
10007 (mode_iterator::get_2xwider): Likewise.
10008 (mode_iterator::get_known_wider): Likewise, turning into a template.
10009 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
10010 forcing a wider mode to exist.
10011 * config/cr16/cr16.h (LONG_REG_P): Likewise.
10012 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10013 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
10014 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
10015 * lower-subreg.c (init_lower_subreg): Likewise.
10016 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
10017 on the final iteration.
10018 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
10019 a wider mode exists before asking for a move pattern.
10020 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
10021 forcing a wider mode to exist.
10022 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
10023 returning false if no such mode exists.
10024 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
10025 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
10026 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
10027 Avoid checking for a MODE_INT if we already know the mode is not a
10029 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
10030 forcing a wider mode to exist.
10031 (expmed_mult_highpart_optab): Likewise.
10032 (expmed_mult_highpart): Likewise.
10033 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
10035 * lto-streamer-in.c (lto_input_mode_table): Likewise.
10036 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
10037 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10038 * internal-fn.c (expand_mul_overflow): Update use of
10039 GET_MODE_2XWIDER_MODE.
10040 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10041 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
10042 GET_MODE_WIDER_MODE.
10043 (convert_plusminus_to_widen): Likewise.
10044 * tree-switch-conversion.c (array_value_type): Likewise.
10045 * var-tracking.c (emit_note_insn_var_location): Likewise.
10046 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10047 Return false inside rather than outside the loop if no wider mode
10049 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
10050 and GET_MODE_2XWIDER_MODE
10051 (can_compare_p): Use else_void.
10052 * gdbhooks.py (OptMachineModePrinter): New class.
10053 (build_pretty_printer): Use it for opt_mode.
10055 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10056 Alan Hayward <alan.hayward@arm.com>
10057 David Sherwood <david.sherwood@arm.com>
10059 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
10060 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
10062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10063 Alan Hayward <alan.hayward@arm.com>
10064 David Sherwood <david.sherwood@arm.com>
10066 * machmode.h (mode_traits): New structure.
10067 (get_narrowest_mode): New function.
10068 (mode_iterator::start): Likewise.
10069 (mode_iterator::iterate_p): Likewise.
10070 (mode_iterator::get_wider): Likewise.
10071 (mode_iterator::get_known_wider): Likewise.
10072 (mode_iterator::get_2xwider): Likewise.
10073 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
10074 (FOR_EACH_MODE): Likewise.
10075 (FOR_EACH_MODE_FROM): Likewise.
10076 (FOR_EACH_MODE_UNTIL): Likewise.
10077 (FOR_EACH_WIDER_MODE): Likewise.
10078 (FOR_EACH_2XWIDER_MODE): Likewise.
10079 * builtins.c (expand_builtin_strlen): Use new mode iterators.
10080 * combine.c (simplify_comparison): Likewise
10081 * config/i386/i386.c (type_natural_mode): Likewise.
10082 * cse.c (cse_insn): Likewise.
10083 * dse.c (find_shift_sequence): Likewise.
10084 * emit-rtl.c (init_derived_machine_modes): Likewise.
10085 (init_emit_once): Likewise.
10086 * explow.c (hard_function_value): Likewise.
10087 * expmed.c (extract_fixed_bit_field_1): Likewise.
10088 (extract_bit_field_1): Likewise.
10089 (expand_divmod): Likewise.
10090 (emit_store_flag_1): Likewise.
10091 * expr.c (init_expr_target): Likewise.
10092 (convert_move): Likewise.
10093 (alignment_for_piecewise_move): Likewise.
10094 (widest_int_mode_for_size): Likewise.
10095 (emit_block_move_via_movmem): Likewise.
10096 (copy_blkmode_to_reg): Likewise.
10097 (set_storage_via_setmem): Likewise.
10098 (compress_float_constant): Likewise.
10099 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10100 * optabs-query.c (get_best_extraction_insn): Likewise.
10101 * optabs.c (expand_binop): Likewise.
10102 (expand_twoval_unop): Likewise.
10103 (expand_twoval_binop): Likewise.
10104 (widen_leading): Likewise.
10105 (widen_bswap): Likewise.
10106 (expand_parity): Likewise.
10107 (expand_unop): Likewise.
10108 (prepare_cmp_insn): Likewise.
10109 (prepare_float_lib_cmp): Likewise.
10110 (expand_float): Likewise.
10111 (expand_fix): Likewise.
10112 (expand_sfix_optab): Likewise.
10113 * postreload.c (move2add_use_add2_insn): Likewise.
10114 * reg-stack.c (reg_to_stack): Likewise.
10115 * reginfo.c (choose_hard_reg_mode): Likewise.
10116 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10117 * stor-layout.c (mode_for_size): Likewise.
10118 (smallest_mode_for_size): Likewise.
10119 (mode_for_vector): Likewise.
10120 (finish_bitfield_representative): Likewise.
10121 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
10122 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
10123 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10124 * var-tracking.c (prepare_call_arguments): Likewise.
10126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10127 Alan Hayward <alan.hayward@arm.com>
10128 David Sherwood <david.sherwood@arm.com>
10130 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
10131 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
10132 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
10133 * machmode.h (mode_size): Move earlier in file.
10134 (mode_precision): Likewise.
10135 (mode_inner): Likewise.
10136 (mode_nunits): Likewise.
10137 (mode_unit_size): Likewise.
10138 (unit_unit_precision): Likewise.
10139 (mode_wider): Likewise.
10140 (mode_2xwider): Likewise.
10141 (machine_mode): New class.
10142 (mode_to_bytes): New function.
10143 (mode_to_bits): Likewise.
10144 (mode_to_precision): Likewise.
10145 (mode_to_inner): Likewise.
10146 (mode_to_unit_size): Likewise.
10147 (mode_to_unit_precision): Likewise.
10148 (mode_to_nunits): Likewise.
10149 (GET_MODE_SIZE): Use mode_to_bytes.
10150 (GET_MODE_BITSIZE): Use mode_to_bits.
10151 (GET_MODE_PRECISION): Use mode_to_precision.
10152 (GET_MODE_INNER): Use mode_to_inner.
10153 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
10154 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
10155 (GET_MODE_NUNITS): Use mode_to_nunits.
10156 * system.h (ALWAYS_INLINE): New macro.
10157 * config/powerpcspe/powerpcspe-c.c
10158 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
10159 int for arg1_mode and arg2_mode.
10161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10162 Alan Hayward <alan.hayward@arm.com>
10163 David Sherwood <david.sherwood@arm.com>
10165 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10166 Prefix mode names with E_ in case statements.
10167 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10168 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
10169 (aarch64_split_simd_move): Likewise.
10170 (aarch64_gen_storewb_pair): Likewise.
10171 (aarch64_gen_loadwb_pair): Likewise.
10172 (aarch64_gen_store_pair): Likewise.
10173 (aarch64_gen_load_pair): Likewise.
10174 (aarch64_get_condition_code_1): Likewise.
10175 (aarch64_constant_pool_reload_icode): Likewise.
10176 (get_rsqrte_type): Likewise.
10177 (get_rsqrts_type): Likewise.
10178 (get_recpe_type): Likewise.
10179 (get_recps_type): Likewise.
10180 (aarch64_gimplify_va_arg_expr): Likewise.
10181 (aarch64_simd_container_mode): Likewise.
10182 (aarch64_emit_load_exclusive): Likewise.
10183 (aarch64_emit_store_exclusive): Likewise.
10184 (aarch64_expand_compare_and_swap): Likewise.
10185 (aarch64_gen_atomic_cas): Likewise.
10186 (aarch64_emit_bic): Likewise.
10187 (aarch64_emit_atomic_swap): Likewise.
10188 (aarch64_emit_atomic_load_op): Likewise.
10189 (aarch64_evpc_trn): Likewise.
10190 (aarch64_evpc_uzp): Likewise.
10191 (aarch64_evpc_zip): Likewise.
10192 (aarch64_evpc_ext): Likewise.
10193 (aarch64_evpc_rev): Likewise.
10194 (aarch64_evpc_dup): Likewise.
10195 (aarch64_gen_ccmp_first): Likewise.
10196 (aarch64_gen_ccmp_next): Likewise.
10197 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10198 (alpha_emit_xfloating_libcall): Likewise.
10199 (emit_insxl): Likewise.
10200 (alpha_arg_type): Likewise.
10201 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
10202 (arc_preferred_simd_mode): Likewise.
10203 (arc_secondary_reload): Likewise.
10204 (get_arc_condition_code): Likewise.
10205 (arc_print_operand): Likewise.
10206 (arc_legitimate_constant_p): Likewise.
10207 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10208 * config/arc/arc.md (casesi_load): Likewise.
10209 (casesi_compact_jump): Likewise.
10210 * config/arc/predicates.md (proper_comparison_operator): Likewise.
10211 (cc_use_register): Likewise.
10212 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10213 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10214 (arm_init_iwmmxt_builtins): Likewise.
10215 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
10216 (neon_expand_vector_init): Likewise.
10217 (arm_attr_length_move_neon): Likewise.
10218 (maybe_get_arm_condition_code): Likewise.
10219 (arm_emit_vector_const): Likewise.
10220 (arm_preferred_simd_mode): Likewise.
10221 (arm_output_iwmmxt_tinsr): Likewise.
10222 (thumb1_output_casesi): Likewise.
10223 (thumb2_output_casesi): Likewise.
10224 (arm_emit_load_exclusive): Likewise.
10225 (arm_emit_store_exclusive): Likewise.
10226 (arm_expand_compare_and_swap): Likewise.
10227 (arm_evpc_neon_vuzp): Likewise.
10228 (arm_evpc_neon_vzip): Likewise.
10229 (arm_evpc_neon_vrev): Likewise.
10230 (arm_evpc_neon_vtrn): Likewise.
10231 (arm_evpc_neon_vext): Likewise.
10232 (arm_validize_comparison): Likewise.
10233 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
10234 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
10235 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
10236 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
10237 (c6x_preferred_simd_mode): Likewise.
10238 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
10239 (epiphany_rtx_costs): Likewise.
10240 * config/epiphany/predicates.md (proper_comparison_operator):
10242 * config/frv/frv.c (condexec_memory_operand): Likewise.
10243 (frv_emit_move): Likewise.
10244 (output_move_single): Likewise.
10245 (output_condmove_single): Likewise.
10246 (frv_hard_regno_mode_ok): Likewise.
10247 (frv_matching_accg_mode): Likewise.
10248 * config/h8300/h8300.c (split_adds_subs): Likewise.
10249 (h8300_rtx_costs): Likewise.
10250 (h8300_print_operand): Likewise.
10251 (compute_mov_length): Likewise.
10252 (output_logical_op): Likewise.
10253 (compute_logical_op_length): Likewise.
10254 (compute_logical_op_cc): Likewise.
10255 (h8300_shift_needs_scratch_p): Likewise.
10256 (output_a_shift): Likewise.
10257 (compute_a_shift_length): Likewise.
10258 (compute_a_shift_cc): Likewise.
10259 (expand_a_rotate): Likewise.
10260 (output_a_rotate): Likewise.
10261 * config/i386/i386.c (classify_argument): Likewise.
10262 (function_arg_advance_32): Likewise.
10263 (function_arg_32): Likewise.
10264 (function_arg_64): Likewise.
10265 (function_value_64): Likewise.
10266 (ix86_gimplify_va_arg): Likewise.
10267 (ix86_legitimate_constant_p): Likewise.
10268 (put_condition_code): Likewise.
10269 (split_double_mode): Likewise.
10270 (ix86_avx256_split_vector_move_misalign): Likewise.
10271 (ix86_expand_vector_logical_operator): Likewise.
10272 (ix86_split_idivmod): Likewise.
10273 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
10274 (ix86_build_const_vector): Likewise.
10275 (ix86_build_signbit_mask): Likewise.
10276 (ix86_match_ccmode): Likewise.
10277 (ix86_cc_modes_compatible): Likewise.
10278 (ix86_expand_branch): Likewise.
10279 (ix86_expand_sse_cmp): Likewise.
10280 (ix86_expand_sse_movcc): Likewise.
10281 (ix86_expand_int_sse_cmp): Likewise.
10282 (ix86_expand_vec_perm_vpermi2): Likewise.
10283 (ix86_expand_vec_perm): Likewise.
10284 (ix86_expand_sse_unpack): Likewise.
10285 (ix86_expand_int_addcc): Likewise.
10286 (ix86_split_to_parts): Likewise.
10287 (ix86_vectorize_builtin_gather): Likewise.
10288 (ix86_vectorize_builtin_scatter): Likewise.
10289 (avx_vpermilp_parallel): Likewise.
10290 (inline_memory_move_cost): Likewise.
10291 (ix86_tieable_integer_mode_p): Likewise.
10292 (x86_maybe_negate_const_int): Likewise.
10293 (ix86_expand_vector_init_duplicate): Likewise.
10294 (ix86_expand_vector_init_one_nonzero): Likewise.
10295 (ix86_expand_vector_init_one_var): Likewise.
10296 (ix86_expand_vector_init_concat): Likewise.
10297 (ix86_expand_vector_init_interleave): Likewise.
10298 (ix86_expand_vector_init_general): Likewise.
10299 (ix86_expand_vector_set): Likewise.
10300 (ix86_expand_vector_extract): Likewise.
10301 (emit_reduc_half): Likewise.
10302 (ix86_emit_i387_round): Likewise.
10303 (ix86_mangle_type): Likewise.
10304 (ix86_expand_round_sse4): Likewise.
10305 (expand_vec_perm_blend): Likewise.
10306 (canonicalize_vector_int_perm): Likewise.
10307 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
10308 (expand_vec_perm_1): Likewise.
10309 (expand_vec_perm_interleave3): Likewise.
10310 (expand_vec_perm_even_odd_pack): Likewise.
10311 (expand_vec_perm_even_odd_1): Likewise.
10312 (expand_vec_perm_broadcast_1): Likewise.
10313 (ix86_vectorize_vec_perm_const_ok): Likewise.
10314 (ix86_expand_vecop_qihi): Likewise.
10315 (ix86_expand_mul_widen_hilo): Likewise.
10316 (ix86_expand_sse2_abs): Likewise.
10317 (ix86_expand_pextr): Likewise.
10318 (ix86_expand_pinsr): Likewise.
10319 (ix86_preferred_simd_mode): Likewise.
10320 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
10321 * config/i386/sse.md (*andnot<mode>3): Likewise.
10322 (<mask_codefor><code><mode>3<mask_name>): Likewise.
10323 (*<code><mode>3): Likewise.
10324 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
10325 (ia64_expand_atomic_op): Likewise.
10326 (ia64_arg_type): Likewise.
10327 (ia64_mode_to_int): Likewise.
10328 (ia64_scalar_mode_supported_p): Likewise.
10329 (ia64_vector_mode_supported_p): Likewise.
10330 (expand_vec_perm_broadcast): Likewise.
10331 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
10332 (iq2000_function_arg_advance): Likewise.
10333 (iq2000_function_arg): Likewise.
10334 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
10335 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
10336 (m68k_libcall_value): Likewise.
10337 (m68k_function_value): Likewise.
10338 (sched_attr_op_type): Likewise.
10339 * config/mcore/mcore.c (mcore_output_move): Likewise.
10340 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
10342 (microblaze_function_arg): Likewise.
10343 * config/mips/mips.c (mips16_build_call_stub): Likewise.
10344 (mips_print_operand): Likewise.
10345 (mips_mode_ok_for_mov_fmt_p): Likewise.
10346 (mips_vector_mode_supported_p): Likewise.
10347 (mips_preferred_simd_mode): Likewise.
10348 (mips_expand_vpc_loongson_even_odd): Likewise.
10349 (mips_expand_vec_unpack): Likewise.
10350 (mips_expand_vi_broadcast): Likewise.
10351 (mips_expand_vector_init): Likewise.
10352 (mips_expand_vec_reduc): Likewise.
10353 (mips_expand_msa_cmp): Likewise.
10354 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
10355 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
10356 (cc_flags_for_mode): Likewise.
10357 * config/msp430/msp430.c (msp430_print_operand): Likewise.
10358 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
10359 (nds32_output_casesi_pc_relative): Likewise.
10360 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10361 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
10362 (nvptx_gen_unpack): Likewise.
10363 (nvptx_gen_pack): Likewise.
10364 (nvptx_gen_shuffle): Likewise.
10365 (nvptx_gen_wcast): Likewise.
10366 (nvptx_preferred_simd_mode): Likewise.
10367 * config/pa/pa.c (pa_secondary_reload): Likewise.
10368 * config/pa/predicates.md (base14_operand): Likewise.
10369 * config/powerpcspe/powerpcspe-c.c
10370 (altivec_resolve_overloaded_builtin): Likewise.
10371 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
10373 (rs6000_preferred_simd_mode): Likewise.
10374 (output_vec_const_move): Likewise.
10375 (rs6000_expand_vector_extract): Likewise.
10376 (rs6000_split_vec_extract_var): Likewise.
10377 (reg_offset_addressing_ok_p): Likewise.
10378 (rs6000_legitimate_offset_address_p): Likewise.
10379 (rs6000_legitimize_address): Likewise.
10380 (rs6000_emit_set_const): Likewise.
10381 (rs6000_const_vec): Likewise.
10382 (rs6000_emit_move): Likewise.
10383 (spe_build_register_parallel): Likewise.
10384 (rs6000_darwin64_record_arg_recurse): Likewise.
10385 (swap_selector_for_mode): Likewise.
10386 (spe_init_builtins): Likewise.
10387 (paired_init_builtins): Likewise.
10388 (altivec_init_builtins): Likewise.
10389 (do_load_for_compare): Likewise.
10390 (rs6000_generate_compare): Likewise.
10391 (rs6000_expand_float128_convert): Likewise.
10392 (emit_load_locked): Likewise.
10393 (emit_store_conditional): Likewise.
10394 (rs6000_output_function_epilogue): Likewise.
10395 (rs6000_handle_altivec_attribute): Likewise.
10396 (rs6000_function_value): Likewise.
10397 (emit_fusion_gpr_load): Likewise.
10398 (emit_fusion_p9_load): Likewise.
10399 (emit_fusion_p9_store): Likewise.
10400 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
10401 (fusion_gpr_mem_load): Likewise.
10402 (fusion_addis_mem_combo_load): Likewise.
10403 (fusion_addis_mem_combo_store): Likewise.
10404 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
10405 (fusion_gpr_mem_load): Likewise.
10406 (fusion_addis_mem_combo_load): Likewise.
10407 (fusion_addis_mem_combo_store): Likewise.
10408 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10410 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
10411 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10412 (rs6000_preferred_simd_mode): Likewise.
10413 (output_vec_const_move): Likewise.
10414 (rs6000_expand_vector_extract): Likewise.
10415 (rs6000_split_vec_extract_var): Likewise.
10416 (reg_offset_addressing_ok_p): Likewise.
10417 (rs6000_legitimate_offset_address_p): Likewise.
10418 (rs6000_legitimize_address): Likewise.
10419 (rs6000_emit_set_const): Likewise.
10420 (rs6000_const_vec): Likewise.
10421 (rs6000_emit_move): Likewise.
10422 (rs6000_darwin64_record_arg_recurse): Likewise.
10423 (swap_selector_for_mode): Likewise.
10424 (paired_init_builtins): Likewise.
10425 (altivec_init_builtins): Likewise.
10426 (rs6000_expand_float128_convert): Likewise.
10427 (emit_load_locked): Likewise.
10428 (emit_store_conditional): Likewise.
10429 (rs6000_output_function_epilogue): Likewise.
10430 (rs6000_handle_altivec_attribute): Likewise.
10431 (rs6000_function_value): Likewise.
10432 (emit_fusion_gpr_load): Likewise.
10433 (emit_fusion_p9_load): Likewise.
10434 (emit_fusion_p9_store): Likewise.
10435 * config/rx/rx.c (rx_gen_move_template): Likewise.
10436 (flags_from_mode): Likewise.
10437 * config/s390/predicates.md (s390_alc_comparison): Likewise.
10438 (s390_slb_comparison): Likewise.
10439 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
10440 (s390_vector_mode_supported_p): Likewise.
10441 (s390_cc_modes_compatible): Likewise.
10442 (s390_match_ccmode_set): Likewise.
10443 (s390_canonicalize_comparison): Likewise.
10444 (s390_emit_compare_and_swap): Likewise.
10445 (s390_branch_condition_mask): Likewise.
10446 (s390_rtx_costs): Likewise.
10447 (s390_secondary_reload): Likewise.
10448 (__SECONDARY_RELOAD_CASE): Likewise.
10449 (s390_expand_cs): Likewise.
10450 (s390_preferred_simd_mode): Likewise.
10451 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
10452 * config/sh/sh.c (sh_print_operand): Likewise.
10453 (dump_table): Likewise.
10454 (sh_secondary_reload): Likewise.
10455 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10456 * config/sh/sh.md (casesi_worker_1): Likewise.
10457 (casesi_worker_2): Likewise.
10458 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
10459 (fcc_comparison_operator): Likewise.
10460 * config/sparc/sparc.c (sparc_expand_move): Likewise.
10461 (emit_soft_tfmode_cvt): Likewise.
10462 (sparc_preferred_simd_mode): Likewise.
10463 (output_cbranch): Likewise.
10464 (sparc_print_operand): Likewise.
10465 (sparc_expand_vec_perm_bmask): Likewise.
10466 (vector_init_bshuffle): Likewise.
10467 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10468 (spu_vector_mode_supported_p): Likewise.
10469 (spu_expand_insv): Likewise.
10470 (spu_emit_branch_or_set): Likewise.
10471 (spu_handle_vector_attribute): Likewise.
10472 (spu_builtin_splats): Likewise.
10473 (spu_builtin_extract): Likewise.
10474 (spu_builtin_promote): Likewise.
10475 (spu_expand_sign_extend): Likewise.
10476 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10477 (tilegx_simd_int): Likewise.
10478 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
10479 (tilepro_simd_int): Likewise.
10480 * config/v850/v850.c (const_double_split): Likewise.
10481 (v850_print_operand): Likewise.
10482 (ep_memory_offset): Likewise.
10483 * config/vax/vax.c (vax_rtx_costs): Likewise.
10484 (vax_output_int_move): Likewise.
10485 (vax_output_int_add): Likewise.
10486 (vax_output_int_subtract): Likewise.
10487 * config/visium/predicates.md (visium_branch_operator): Likewise.
10488 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
10489 (visium_print_operand_address): Likewise.
10490 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10491 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
10492 (xtensa_expand_conditional_branch): Likewise.
10493 (xtensa_copy_incoming_a7): Likewise.
10494 (xtensa_output_literal): Likewise.
10495 * dfp.c (decimal_real_maxval): Likewise.
10496 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10499 Alan Hayward <alan.hayward@arm.com>
10500 David Sherwood <david.sherwood@arm.com>
10502 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
10503 (mode_nunits_inline): Likewise.
10504 (mode_inner_inline): Likewise.
10505 (mode_unit_size_inline): Likewise.
10506 (mode_unit_precision_inline): Likewise.
10507 (emit_insn_modes_h): Likewise. Also emit a #define of the
10509 (emit_mode_wider): Add an E_ prefix to mode names.
10510 (emit_mode_complex): Likewise.
10511 (emit_mode_inner): Likewise.
10512 (emit_mode_adjustments): Likewise.
10513 (emit_mode_int_n): Likewise.
10514 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
10515 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
10516 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
10517 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
10518 (CRC32_BUILTIN, ENTRY): Likewise.
10519 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
10520 (aarch64_pop_regs): Likewise.
10521 (aarch64_process_components): Likewise.
10522 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
10523 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
10524 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
10525 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
10526 * config/arm/arm.c (arm_init_libfuncs): Likewise.
10527 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
10529 * config/i386/i386-builtin.def (pcmpestr): Likewise.
10530 (pcmpistr): Likewise.
10531 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
10532 * config/mmix/mmix.c (mmix_output_condition): Likewise.
10533 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10535 * config/rl78/rl78.c (mduc_regs): Likewise.
10536 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10537 (htm_expand_builtin): Likewise.
10538 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
10539 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
10540 * config/xtensa/xtensa.c (print_operand): Likewise.
10541 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
10542 (NUM_MODE_VECTOR_INT): Likewise.
10543 * genoutput.c (null_operand): Likewise.
10544 (output_operand_data): Likewise.
10545 * genrecog.c (print_parameter_value): Likewise.
10546 * lra.c (debug_operand_data): Likewise.
10548 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10550 * dwarf2out.c (scompare_loc_descriptor_wide)
10551 (scompare_loc_descriptor_narrow): New functions, split out from...
10552 (scompare_loc_descriptor): ...here.
10553 * expmed.c (emit_store_flag_int): New function, split out from...
10554 (emit_store_flag): ...here.
10556 2017-08-30 Richard Biener <rguenther@suse.de>
10558 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
10559 (dwarf2out_early_finish): Move setting of AT_pubnames from
10560 early debug output to early finish.
10562 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10564 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
10565 and -mdata-region to the assembler.
10567 2017-08-30 Richard Biener <rguenther@suse.de>
10569 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
10571 (gen_subprogram_die): Add DW_AT_object_pointer only early.
10572 (dwarf2out_early_global_decl): Only generate a DIE for the
10573 abstract origin if it doesn't already exist or is a declaration DIE.
10574 (resolve_addr): Do not add the linkage name twice when
10575 generating a stub DIE for the DW_TAG_GNU_call_site target.
10577 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10579 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10580 Use machine_mode rather than int for arg1_mode.
10582 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
10585 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
10586 that the second argument of the built-in functions to unpack
10587 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
10588 switch statement instead a lot of if statements.
10589 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
10590 Allow 64-bit values to be in Altivec registers as well as
10591 traditional floating point registers.
10592 (pack<mode>, FMOVE128_VSX iterator): Likewise.
10594 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
10596 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
10597 MAX_REGS_PER_ADDRESS == 1.
10599 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
10601 * config/i386/i386.opt (flag_fentry): Do not init to -1.
10602 * config/i386/i386.c (ix86_option_override_internal): Simplify
10603 setting of opts->x_flag_entry.
10605 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10606 Jakub Jelinek <jakub@redhat.com>
10607 Richard Biener <rguenther@suse.de>
10609 PR tree-optimization/81503
10610 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
10611 folded constant fits in the target type; reorder tests for clarity.
10613 2017-08-29 Martin Liska <mliska@suse.cz>
10615 * passes.def: Include pass_lower_switch.
10616 * stmt.c (dump_case_nodes): Remove and move to
10617 tree-switch-conversion.
10618 (case_values_threshold): Likewise.
10619 (expand_switch_as_decision_tree_p): Likewise.
10620 (emit_case_decision_tree): Likewise.
10621 (expand_case): Likewise.
10622 (balance_case_nodes): Likewise.
10623 (node_has_low_bound): Likewise.
10624 (node_has_high_bound): Likewise.
10625 (node_is_bounded): Likewise.
10626 (emit_case_nodes): Likewise.
10627 (struct simple_case_node): New struct.
10628 (add_case_node): Remove.
10629 (emit_case_dispatch_table): Use vector instead of case_list.
10630 (reset_out_edges_aux): Remove.
10631 (compute_cases_per_edge): Likewise.
10632 (expand_case): Build list of simple_case_node.
10633 (expand_sjlj_dispatch_table): Use it.
10634 * tree-switch-conversion.c (struct case_node): Moved from
10635 stmt.c and adjusted.
10636 (emit_case_nodes): Likewise.
10637 (node_has_low_bound): Likewise.
10638 (node_has_high_bound): Likewise.
10639 (node_is_bounded): Likewise.
10640 (case_values_threshold): Likewise.
10641 (reset_out_edges_aux): Likewise.
10642 (compute_cases_per_edge): Likewise.
10643 (add_case_node): Likewise.
10644 (dump_case_nodes): Likewise.
10645 (balance_case_nodes): Likewise.
10646 (expand_switch_as_decision_tree_p): Likewise.
10647 (emit_jump): Likewise.
10648 (emit_case_decision_tree): Likewise.
10649 (try_switch_expansion): Likewise.
10650 (do_jump_if_equal): Likewise.
10651 (emit_cmp_and_jump_insns): Likewise.
10652 (fix_phi_operands_for_edge): New function.
10653 (record_phi_operand_mapping): Likewise.
10654 (class pass_lower_switch): New pass.
10655 (pass_lower_switch::execute): New function.
10656 (make_pass_lower_switch): Likewise.
10657 (conditional_probability):
10658 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
10659 * tree-pass.h: Add make_pass_lower_switch.
10661 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10664 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
10667 2017-08-29 Richard Biener <rguenther@suse.de>
10669 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
10670 we do not add a DW_AT_inline attribute twice.
10671 (gen_subprogram_die): Remove code setting DW_AT_inline on
10672 DECL_ABSTRACT_P nodes.
10674 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
10676 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
10677 calls to internal functions.
10678 (gimplify_modify_expr): Likewise.
10679 * tree-call-cdce.c (use_internal_fn): Likewise.
10680 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
10681 (convert_to_divmod): Set the nothrow flag.
10682 * tree-if-conv.c (predicate_mem_writes): Likewise.
10683 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10684 (vectorizable_call): Likewise.
10685 (vectorizable_store): Likewise.
10686 (vectorizable_load): Likewise.
10687 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
10688 (vect_recog_mask_conversion_pattern): Likewise.
10690 2017-08-29 Martin Liska <mliska@suse.cz>
10693 * gcc.c (driver_handle_option): Add new argument.
10694 * opts-common.c (handle_option): Pass
10695 target_option_override_hook.
10696 * opts-global.c (lang_handle_option): Add new option.
10697 (set_default_handlers): Add new argument.
10698 (decode_options): Likewise.
10699 * opts.c (target_handle_option): Likewise.
10700 (common_handle_option): Call target_option_override_hook.
10701 * opts.h (struct cl_option_handler_func): Add hook for
10702 target option override.
10703 (struct cl_option_handlers): Likewise.
10704 (set_default_handlers): Add new argument.
10705 (decode_options): Likewise.
10706 (common_handle_option): Likewise.
10707 (target_handle_option): Likewise.
10708 * toplev.c (toplev::main): Pass targetm.target_option.override
10711 2017-08-29 Richard Biener <rguenther@suse.de>
10712 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10714 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
10715 life to the active subtree.
10717 2017-08-28 Jeff Law <law@redhat.com>
10719 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
10720 derive_equivalences.
10721 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
10722 Code moved into....
10723 (edge_info::derive_equivalences): New private member function
10725 * tree-ssa-dom.c (class edge_info): Changed from a struct
10726 to a class. Add ctor/dtor, methods and data members.
10727 (edge_info::edge_info): Renamed from allocate_edge_info.
10728 Initialize additional members.
10729 (edge_info::~edge_info): New.
10730 (free_dom_edge_info): Delete the edge info.
10731 (record_edge_info): Use new class & associated member functions.
10732 Tighten forms for testing for edge equivalences.
10733 (record_temporary_equivalences): Iterate over the simple
10734 equivalences rather than assuming there's only one per edge.
10735 (cprop_into_successor_phis): Iterate over the simple
10736 equivalences rather than assuming there's only one per edge.
10737 (optimize_stmt): Use operand_equal_p rather than pointer
10738 equality for mini-DSE code.
10740 2017-08-28 Nathan Sidwell <nathan@acm.org>
10742 * gcc.c (execute): Fold SIGPIPE handling into switch
10743 statement. Adjust internal error message.
10745 2017-08-28 Richard Biener <rguenther@suse.de>
10748 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
10749 Do nothing for removed DIEs.
10751 2017-08-28 Richard Biener <rguenther@suse.de>
10753 PR tree-optimization/81977
10754 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
10757 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
10760 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
10761 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
10762 using targetm.gen_mem_thread_fence.
10764 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
10767 * config/i386/i386.md (*<btsc><mode>): Change operand 2
10768 predicate to register_operand. Reorder operands.
10769 (*btr<mode>): Ditto.
10770 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
10771 (*btr<mode>_mask): Ditto.
10773 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
10775 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
10776 * config/rs6000/xmmintrin.h: New file.
10777 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
10779 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10782 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
10783 parameter and_insn and return it.
10784 (recombine_lvx_pattern): Insert a copy to ensure availability of
10785 the base register of the copied masking operation at the point of
10786 the instruction replacement.
10787 (recombine_stvx_pattern): Likewise.
10789 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
10791 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
10792 undocumented switches.
10793 (-mpower9-dform-vector): Likewise.
10794 (-mpower9-dform): Likewise.
10795 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
10796 comments to delete references to -mpower9-dform* switches.
10797 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
10798 Delete reference to -mpower9-dform* switches, test for
10799 -mpower9-vector instead.
10800 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
10801 (OTHER_P9_VECTOR_MASKS): Likewise.
10802 (POWERPC_MASKS): Likewise.
10803 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
10804 tests against -mpower9-dform* to -mpower9-vector. Delete code
10805 that checked for -mpower9-dform* consistancy with other options.
10806 Add test for -mpower9-misc to enable other power9 switches.
10807 (rs6000_init_hard_regno_mode_ok): Likewise.
10808 (rs6000_option_override_internal): Likewise.
10809 (rs6000_emit_prologue): Likewise.
10810 (rs6000_emit_epilogue): Likewise.
10811 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
10812 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
10813 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
10815 (emit_fusion_p9_store): Likewise.
10816 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
10817 resetting these macros if the assembler does not support ISA 3.0
10819 (TARGET_P9_DFORM_VECTOR): Likewise.
10820 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
10821 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
10823 2017-08-25 Alan Modra <amodra@gmail.com>
10825 PR rtl-optimization/81747
10826 * cse.c (cse_extended_basic_block): Don't attempt to record
10827 equivalences for degenerate conditional branches that branch
10828 to their fall-through.
10830 2017-08-24 Martin Sebor <msebor@redhat.com>
10832 PR middle-end/81908
10833 * gimple-fold.c (size_must_be_zero_p): New function.
10834 (gimple_fold_builtin_memory_op): Call it.
10836 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
10838 * config/rs6000/mm_malloc.h: New file.
10840 2017-08-24 Bin Cheng <bin.cheng@arm.com>
10842 PR tree-optimization/81913
10843 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
10844 analysis when either IVs in condition can wrap.
10846 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
10848 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
10849 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
10851 2017-08-24 Richard Biener <rguenther@suse.de>
10854 * targhooks.c (default_target_can_inline_p): Properly
10855 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
10856 is present and always compare.
10857 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
10858 infer -mfpmath=sse from TARGET_SSE_P.
10859 (ix86_can_inline_p): Properly use target_option_default_node when
10860 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
10862 2017-08-24 Richard Biener <rguenther@suse.de>
10865 * dwarf2out.c (output_die): Handle flag_generate_offload like
10867 (output_comp_unit): Likewise.
10868 (gen_array_type_die): Likewise.
10869 (dwarf2out_early_finish): Likewise.
10870 (note_variable_value_in_expr): Likewise.
10871 (dwarf2out_finish): Likewise. Adjust assert.
10872 * cgraphunit.c (symbol_table::compile): Move setting of
10873 flag_generate_offload earlier ...
10874 (symbol_table::finalize_compilation_unit): ... here, before
10875 early debug finalization.
10877 2017-08-24 Richard Biener <rguenther@suse.de>
10879 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
10880 and ipa-fnsummary.h.
10881 (ix86_can_inline_p): When ix86_fpmath flags do not match
10882 check whether the callee uses FP math at all.
10884 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
10886 PR middle-end/81931
10887 * tree-ssanames.c (get_nonzero_bits): Use element_precision
10888 instead of TYPE_PRECISION.
10890 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
10891 Alan Hayward <alan.hayward@arm.com>
10892 David Sherwood <david.sherwood@arm.com>
10894 * combine.c (make_extraction): Use subreg_offset_from_lsb.
10896 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
10898 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
10900 (ix86_frame::stack_realign_allocate): New field.
10901 (struct machine_frame_state): Modify comments.
10902 (machine_frame_state::sp_realigned_fp_end): New field.
10903 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
10904 layout calculation.
10905 (sp_valid_at): Add assertion to assure no attempt to access invalid
10906 offset of a realigned stack.
10907 (fp_valid_at): Likewise.
10908 (choose_baseaddr): Modify comments.
10909 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
10910 ix86_expand_prologue.
10911 (ix86_expand_prologue): Modify stack realignment and allocation.
10912 (ix86_expand_epilogue): Modify comments.
10913 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
10914 avx2_runtime, avx512f, and avx512f_runtime.
10916 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
10918 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
10919 (mstackrealign): Do not init to -1.
10920 * config/i386/i386.c (ix86_option_override_internal):
10921 Check opts_set, not opts when setting default value of
10922 opts->x_ix86_force_align_arg_pointer.
10924 2017-08-23 Richard Biener <rguenther@suse.de>
10926 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
10927 lang_hooks.decl_printable_name.
10928 * print-rtl-function.c (print_rtx_function): Likewise.
10929 * tree-pretty-print.c (dump_function_header): Likewise.
10931 2017-08-23 Richard Biener <rguenther@suse.de>
10934 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
10935 -g0 at compile-time.
10937 2017-08-23 Tamar Christina <tamar.christina@arm.com>
10939 PR middle-end/19706
10940 * doc/sourcebuild.texi (Other hardware attributes):
10943 2017-08-23 Tamar Christina <tamar.christina@arm.com>
10945 PR middle-end/19706
10946 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
10947 Fix single-use check.
10949 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
10951 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
10953 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
10955 * doc/install.texi: Modify to add more details on running selected
10958 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
10960 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
10961 is combined with -mabi=ms.
10962 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
10965 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10967 PR tree-optimization/81488
10968 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
10969 and cached_basis fields.
10970 (MAX_SPREAD): New constant.
10971 (alloc_cand_and_find_basis): Initialize new fields.
10972 (clear_visited): New function.
10973 (create_phi_basis_1): Rename from create_phi_basis, set visited
10974 and cached_basis fields.
10975 (create_phi_basis): New wrapper function.
10976 (phi_add_costs_1): Rename from phi_add_costs, add spread
10977 parameter, set visited field, short-circuit when limits reached.
10978 (phi_add_costs): New wrapper function.
10979 (record_phi_increments_1): Rename from record_phi_increments, set
10981 (record_phi_increments): New wrapper function.
10982 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
10983 (phi_incr_cost): New wrapper function.
10984 (all_phi_incrs_profitable_1): Rename from
10985 all_phi_incrs_profitable, set visited field.
10986 (all_phi_incrs_profitable): New wrapper function.
10988 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10989 Alan Hayward <alan.hayward@arm.com>
10990 David Sherwood <david.sherwood@arm.com>
10992 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
10993 that takes the outer and inner modes.
10994 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
10995 comparison as the canonical test for a paradoxical subreg.
10996 * combine.c (simplify_set): Use paradoxical_subreg_p.
10997 (make_extraction): Likewise.
10998 (force_to_mode): Likewise.
10999 (rtx_equal_for_field_assignment_p): Likewise.
11000 (gen_lowpart_for_combine): Likewise.
11001 (simplify_comparison): Likewise.
11002 * cse.c (equiv_constant): Likewise.
11003 * expmed.c (store_bit_field_1): Likewise.
11004 * final.c (alter_subreg): Likewise.
11005 * fwprop.c (propagate_rtx): Likewise.
11006 (forward_propagate_subreg): Likewise.
11007 * ira-conflicts.c (ira_build_conflicts): Likewise.
11008 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
11009 * lra-constraints.c (curr_insn_transform): Likewise.
11010 (split_reg): Likewise.
11011 * lra-eliminations.c (move_plus_up): Likewise.
11012 (lra_eliminate_regs_1): Likewise.
11013 * recog.c (general_operand): Likewise.
11014 * ree.c (combine_reaching_defs): Likewise.
11015 * reload.c (push_reload): Likewise.
11016 (find_reloads): Likewise.
11017 * reload1.c (elimination_effects): Likewise.
11018 (compute_reload_subreg_offset): Likewise.
11019 (choose_reload_regs): Likewise.
11020 * rtlanal.c (subreg_lsb_1): Likewise.
11021 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11022 (simplify_subreg): Likewise.
11023 * var-tracking.c (track_loc_p): Likewise.
11024 * emit-rtl.c (byte_lowpart_offset): Likewise.
11025 (paradoxical_subreg_p): Delete out-of-line definition.
11027 2017-08-22 Jeff Law <law@redhat.com>
11029 PR tree-optimization/81741
11030 PR tree-optimization/71947
11031 * tree-ssa-dom.c: Include tree-inline.h.
11032 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
11033 equivalences if one is more expensive to compute than the other.
11034 * tree-ssa-scopedtables.h (class const_or_copies): Make
11035 record_const_or_copy_raw method private.
11036 (class avail_exprs_stack): New method simplify_binary_operation.
11037 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
11038 avail_exprs_stack::simplify_binary_operation as needed.
11039 (avail_exprs_stack::simplify_binary_operation): New function.
11041 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
11043 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
11044 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
11045 (DOT_SYMBOLS): Likewise.
11046 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
11047 (RELOCATABLE_NEEDS_FIXUP): Likewise.
11048 (RS6000_ABI_NAME): Likewise.
11049 (TARGET_CMODEL): Likewise.
11050 (TOC_SECTION_ASM_OP): Likewise.
11051 (SET_CMODEL): New macro.
11052 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
11054 2017-08-22 Richard Biener <rguenther@suse.de>
11056 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
11057 to allow for free-lang-data replacements similar to verify_type_variant.
11059 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11060 Alan Hayward <alan.hayward@arm.com>
11061 David Sherwood <david.sherwood@arm.com>
11063 * config/aarch64/aarch64.md (casesi): Use DImode rather than
11064 VOIDmode for the LABEL_REF.
11066 2017-08-22 Richard Biener <rguenther@suse.de>
11068 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
11070 2017-08-22 Richard Biener <rguenther@suse.de>
11072 * common.opt (feliminate-dwarf2-dups): Ignore.
11073 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
11074 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
11075 same_die_p_wrap, compute_section_prefix,
11076 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
11077 (comdat_symbol_id, comdat_symbol_number): Likewise.
11078 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
11080 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
11081 (output_die): Mark unreachable path unreachable.
11082 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
11083 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
11084 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
11085 (dwarf2out_early_finish): Likewise.
11087 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
11089 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
11091 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
11094 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
11095 not VAR_P. Filter attribute warnings with OPT_Wattributes.
11096 (avr_attribute_table) <io, io_low, address>: Initialize
11097 .decl_required with true.
11099 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
11101 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
11102 undocumented debugging options.
11103 (-mvsx-scalar-double): Likewise.
11104 (-mallow-df-permute): Likewise.
11105 (-mvectorize-builtins): Likewise.
11106 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11107 (rs6000_builtin_vectorized_function): Likewise.
11108 (rs6000_builtin_md_vectorized_function): Likewise.
11109 (rs6000_opt_vars): Likewise.
11111 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
11114 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
11115 (*btrq_imm): Rename from *btrq.
11116 (*btcq_imm): Rename from *btcq.
11117 (btsc): New code attribute.
11118 (*<btsc><mode>): New insn pattern.
11119 (*btr<mode>): Ditto.
11120 (*<btsc><mode>_mask): New insn_and_split pattern.
11121 (*btr<mode>_mask): Ditto.
11123 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11124 Alan Hayward <alan.hayward@arm.com>
11125 David Sherwood <david.sherwood@arm.com>
11127 * function.c (pad_below): Simplify padding calculation.
11129 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11130 Alan Hayward <alan.hayward@arm.com>
11131 David Sherwood <david.sherwood@arm.com>
11133 * target.def (function_prologue): Remove frame size argument.
11134 (function_epilogue): Likewise.
11135 * doc/tm.texi: Regenerate.
11136 * final.c (final_start_function): Update call to function_prologue.
11137 (final_end_function): Update call to function_epilogue.
11138 (default_function_pro_epilogue): Remove frame size argument.
11139 * output.h (default_function_pro_epilogue): Likewise.
11140 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
11141 (arm_output_function_prologue): Likewise.
11142 * config/frv/frv.c (frv_function_prologue): Likewise.
11143 (frv_function_epilogue): Likewise.
11144 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
11145 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
11146 (ia64_output_function_epilogue): Likewise.
11147 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
11148 (m32r_output_function_epilogue): Likewise.
11149 * config/microblaze/microblaze.c (microblaze_function_prologue)
11150 (microblaze_function_epilogue): Likewise.
11151 * config/mips/mips.c (mips_output_function_prologue): Likewise.
11152 (mips_output_function_epilogue): Likewise.
11153 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
11154 (mmix_target_asm_function_epilogue): Likewise.
11155 * config/msp430/msp430.c (msp430_start_function): Likewise.
11156 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
11157 (nds32_asm_function_epilogue): Likewise.
11158 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
11159 * config/pa/pa.c (pa_output_function_prologue): Likewise.
11160 (pa_output_function_epilogue): Likewise.
11161 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
11162 (rs6000_output_function_epilogue): Likewise.
11163 * config/rl78/rl78.c (rl78_start_function): Likewise.
11164 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
11165 (rs6000_output_function_epilogue): Likewise.
11166 * config/rx/rx.c (rx_output_function_prologue): Likewise.
11167 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
11168 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
11169 (sparc_asm_function_epilogue): Likewise.
11171 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11173 * tree.h (type_has_mode_precision_p): New function.
11174 * convert.c (convert_to_integer_1): Use it.
11175 * expr.c (expand_expr_real_2): Likewise.
11176 (expand_expr_real_1): Likewise.
11177 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
11178 * match.pd: Likewise.
11179 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
11180 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
11181 * tree-tailcall.c (process_assignment): Likewise.
11182 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11183 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
11184 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
11185 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11186 (vectorizable_assignment): Likewise.
11187 (vectorizable_shift): Likewise.
11188 (vectorizable_operation): Likewise.
11189 * tree-vrp.c (register_edge_assert_for_2): Likewise.
11191 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
11193 * match.pd: Add pow (C, x) simplification.
11195 2017-08-21 Richard Biener <rguenther@suse.de>
11197 PR tree-optimization/81900
11198 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
11199 for blocks with abnormal predecessors.
11200 (compute_antic): Do not set visited flag prematurely.
11202 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
11205 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
11207 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11209 * stor-layout.h (vector_type_mode): Move to...
11210 * tree.h (vector_type_mode): ...here.
11211 * stor-layout.c (vector_type_mode): Move to...
11212 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
11214 2017-08-21 Richard Biener <rguenther@suse.de>
11216 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
11217 register_external_die hooks.
11218 (debug_false_tree_charstarstar_uhwistar): Declare.
11219 (debug_nothing_tree_charstar_uhwi): Likewise.
11220 * debug.c (do_nothing_debug_hooks): Adjust.
11221 (debug_false_tree_charstarstar_uhwistar): New do nothing.
11222 (debug_nothing_tree_charstar_uhwi): Likewise.
11223 * dbxout.c (dbx_debug_hooks): Adjust.
11224 (xcoff_debug_hooks): Likewise.
11225 * sdbout.c (sdb_debug_hooks): Likewise.
11226 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11227 * dwarf2out.c (macinfo_label_base): New global.
11228 (dwarf2out_register_external_die): New function for the
11229 register_external_die hook.
11230 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
11231 (dwarf2_debug_hooks): Use them.
11232 (dwarf2_lineno_debug_hooks): Adjust.
11233 (struct die_struct): Add with_offset flag.
11234 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
11235 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
11236 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
11237 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
11238 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
11239 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
11240 defining section names for the early LTO debug variants.
11241 (reset_indirect_string): New helper.
11242 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
11243 (print_dw_val): Add support for offsetted symbol references.
11244 (get_ultimate_context): Split out from is_cxx.
11245 (is_cxx): Use get_ultimate_context.
11246 (is_fortran): Add decl overload.
11247 (compute_comp_unit_symbol): Split out worker from
11248 compute_section_prefix.
11249 (compute_section_prefix): Call compute_comp_unit_symbol and
11250 set comdat_type_p here.
11251 (output_die): Skip DIE symbol output for the LTO added one.
11252 Handle DIE symbol references with offset.
11253 (output_comp_unit): Guard section name mangling properly.
11254 For LTO debug sections emit a symbol at the section beginning
11255 which we use to refer to its DIEs.
11256 (add_abstract_origin_attribute): For DIEs registered via
11257 dwarf2out_register_external_die directly refer to the early
11258 DIE rather than indirectly through the shadow one we created.
11259 Remove obsolete call to dwarf2out_abstract_function for
11260 non-function/block origins.
11261 (gen_array_type_die): When generating early LTO debug do
11262 not emit DW_AT_string_length.
11263 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
11264 late when in LTO. As suggested place a gcc_unreachable for
11265 the DECL_ABSTRACT_P case.
11266 (gen_subprogram_die): Avoid another specification DIE
11267 for early built declarations/definitions for the late LTO case.
11268 (gen_variable_die): Add type references for late duplicated VLA dies
11270 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
11271 we have the abstract instance already.
11272 (process_scope_var): Adjust decl DIE contexts in LTO which
11273 first puts them in limbo.
11274 (gen_decl_die): Do not generate type DIEs late apart from
11275 types for VLAs or for decls we do not yet have a DIE. Do not
11276 call dwarf2out_abstract_function late.
11277 (dwarf2out_early_global_decl): Make sure to create DIEs
11278 for abstract instances of a decl first.
11279 (dwarf2out_late_global_decl): Adjust comment.
11280 (output_macinfo_op): With multiple macro sections use
11281 macinfo_label_base to distinguish labels.
11282 (output_macinfo): Likewise. Update macinfo_label_base.
11283 Pass in the line info label.
11284 (note_variable_value_in_expr): When generating LTO resolve
11285 all variable values here by generating DIEs as needed.
11286 (init_sections_and_labels): Add early LTO debug flag parameter
11287 and generate different sections and names if set. Add generation
11288 counter for the labels so we can have multiple of them.
11289 (reset_dies): Helper to allow DIEs to be output multiple times.
11290 (dwarf2out_finish): When outputting DIEs to the fat part of an
11291 LTO object first reset DIEs.
11292 (dwarf2out_early_finish): Output early DIEs when generating LTO.
11293 (modified_type_die): Check for decl_ultimate_origin being self
11295 (gen_type_die_with_usage): Likewise.
11296 (gen_typedef_die): Allow decl_ultimate_origin being self.
11297 (set_decl_abstract_flags): Remove.
11298 (set_block_abstract_flags): Likewise.
11299 (dwarf2out_abstract_function): Treat the early generated DIEs
11300 as the abstract copy and only add DW_AT_inline and
11301 DW_AT_artificial here and call set_decl_origin_self.
11302 If the DIE has an abstract origin don't do anything.
11303 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
11304 if we have none yet (Go fails to build one, PR78628).
11305 (variably_modified_type_p): Prevent endless recursion for Ada
11306 cyclic pointer types.
11307 * lto-streamer-in.c: Include debug.h.
11308 (dref_queue): New global.
11309 (lto_read_tree_1): Stream in DIE references.
11310 (lto_input_tree): Register DIE references.
11311 (input_function): Stream DECL_DEBUG_ARGS.
11312 * lto-streamer-out.c: Include debug.h.
11313 (lto_write_tree_1): Output DIE references.
11314 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
11315 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
11316 (output_function): Stream DECL_DEBUG_ARGS.
11317 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
11318 Stream DECL_ABSTRACT_ORIGIN.
11319 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
11320 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
11321 DECL_CONTEXT for file-scope decls.
11322 * lto-streamer.h (struct dref_entry): Declare.
11323 (dref_queue): Likewise.
11324 * cfgexpand.c (pass_expand::execute): Do not call the
11325 outlining_inline_function hook here.
11326 * lto-wrapper.c (debug_obj): New global.
11327 (tool_cleanup): Unlink it if required.
11328 (debug_objcopy): New function.
11329 (run_gcc): Handle early debug sections in the IL files by
11330 extracting them to separate files, partially linkin them and
11331 feeding the result back as result to the linker.
11332 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
11333 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
11334 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
11335 sections into a separate segment.
11336 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
11338 (darwin_asm_dwarf_section): Likewise.
11339 (darwin_asm_output_dwarf_offset): Likewise.
11340 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
11342 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11343 Alan Hayward <alan.hayward@arm.com>
11344 David Sherwood <david.sherwood@arm.com>
11346 * read-md.h (md_reader::record_potential_iterator_use): Replace
11347 pointer argument with an rtx and an index.
11348 * read-rtl.c (iterator_group::apply_iterator): Likewise.
11349 (apply_mode_iterator): Likewise.
11350 (apply_code_iterator): Likewise.
11351 (apply_int_iterator): Likewise.
11352 (apply_subst_iterator): Likewise.
11353 (record_iterator_use): Likewise.
11354 (record_attribute_use): Likewise.
11355 (md_reader::record_potential_iterator_use): Likewise. Update calls
11356 to record_iterator_use and apply_iterator.
11357 (iterator_use): Replace ptr with x and index.
11358 (attribute_use): Likewise.
11359 (apply_attribute_uses): Update calls to apply_iterator.
11360 (apply_iterators): Likewise. Update initialization of iterator_use.
11361 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
11362 and record_potential_iterator_use.
11363 (rtx_reader::read_rtx_operand): Likewise.
11365 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11366 Alan Hayward <alan.hayward@arm.com>
11367 David Sherwood <david.sherwood@arm.com>
11369 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
11372 2017-08-21 Richard Biener <rguenther@suse.de>
11374 PR middle-end/81884
11375 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
11376 at struct end conservatively when comparing common bases.
11378 2017-08-21 Richard Biener <rguenther@suse.de>
11380 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
11381 (mem_ref_in_stmt): Remove.
11382 (determine_max_movement): Use ref index to get at the reference.
11383 (invariantness_dom_walker::before_dom_children): Deal with
11384 lim data already initialized.
11385 (gather_mem_refs_stmt): Initialize lim data and record ref index.
11387 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
11389 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
11390 (TARGET_ISA_ROUND): Ditto.
11391 (TARGET_ROUND): Ditto.
11392 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
11393 * config/i386/i386.md: Ditto.
11394 * config/i386/sse.md: Ditto.
11395 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
11396 with OPTION_MASK_ISA_SSE4_1.
11398 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
11401 * doc/extend.texi (x86 Built-in Functions): Correct the name of
11402 __builtin_ia32_lzcnt_u16.
11404 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
11407 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
11408 (rs6000_set_current_function): Rewrite function to use it.
11410 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
11413 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
11414 and TYPE_WARN_IF_NOT_ALIGN.
11415 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
11416 (handle_warn_if_not_align): New.
11417 (place_union_field): Call handle_warn_if_not_align.
11418 (place_field): Call handle_warn_if_not_align.
11419 Copy TYPE_WARN_IF_NOT_ALIGN.
11420 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
11421 (layout_type): Likewise.
11422 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
11424 (tree_decl_common): Add warn_if_not_align.
11425 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
11426 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
11427 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
11428 (DECL_WARN_IF_NOT_ALIGN): Likewise.
11429 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
11430 * doc/extend.texi: Document warn_if_not_aligned attribute.
11431 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
11433 2017-08-17 Martin Liska <mliska@suse.cz>
11436 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
11437 (get_data_dependence): Use it as pointer type.
11438 (distribute_loop): Likewise.
11440 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11442 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
11443 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
11444 (p8_vmrgow_<mode>_direct): New define_insn.
11445 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
11446 handle endianness for vmrgew and vmrgow permute patterns.
11448 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
11450 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
11451 * config/rs6000/rs6000-cpus.def: Remove comment.
11452 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
11453 (POWERPC_MASKS): Likewise.
11454 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
11455 use of TARGET_VSX_TIMODE.
11456 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
11457 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
11458 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
11459 (rs6000_option_override_internal): Remove dead code.
11460 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
11461 (rs6000_legitimize_reload_address): Likewise.
11462 (rs6000_legitimate_address_p): Likewise.
11463 (rs6000_opt_masks): Delete "vsx-timode".
11464 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
11465 from function comment.
11466 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
11467 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
11468 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
11470 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
11471 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
11472 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
11473 (VSX_TI): Likewise.
11475 (define_peephole2): Likewise.
11477 2017-08-17 Martin Sebor <msebor@redhat.com>
11480 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
11481 past the end of an array.
11482 (test_pp_format): Add test cases.
11484 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
11486 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
11487 missing ECF_NOTHROW flags.
11489 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
11492 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
11493 operands residing in integer registers.
11494 (*vsx_le_perm_load_<mode>): Likewise.
11495 (*vsx_le_perm_store_<mode>): Likewise.
11496 (define_peephole2): Add peepholes to optimize the above.
11498 2017-08-17 Marek Polacek <polacek@redhat.com>
11500 PR middle-end/81814
11501 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
11502 to mimic what shorten_compare did. Change the return type to bool.
11503 (fold_cond_expr_with_comparison): Update call to
11504 operand_equal_for_comparison_p.
11505 (fold_ternary_loc): Likewise.
11507 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
11509 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
11511 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
11513 2017-08-17 Richard Biener <rguenther@suse.de>
11515 * tree-ssa-structalias.c (solve_graph): When propagating
11516 to successors update the graphs succ edges and avoid duplicate work.
11518 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
11521 * config/i386/i386.c (ix86_option_override_internal): Save target
11522 specific options after ix86_stack_protector_guard_reg was changed.
11524 2017-08-17 Richard Biener <rguenther@suse.de>
11526 PR tree-optimization/81827
11527 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
11528 (new_var_info): Initialize it conservatively.
11529 (get_call_vi): Mark register vars.
11530 (new_scalar_tmp_constraint_exp): Likewise.
11531 (handle_rhs_call): Likewise.
11532 (handle_const_call): Likewise.
11533 (create_function_info_for): Likewise.
11534 (solve_constraints): Sort varinfos to separate register from
11535 non-register vars to pack points-to solution bitmaps during
11538 2017-08-17 Marek Polacek <polacek@redhat.com>
11540 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
11542 2017-08-17 Richard Biener <rguenther@suse.de>
11544 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
11545 to true when overflow is undefined and we saturated the result.
11547 2017-08-17 Alan Modra <amodra@gmail.com>
11550 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
11551 Don't use store multiple if only one reg needs saving.
11552 (interesting_frame_related_regno): New function.
11553 (rs6000_frame_related): Don't emit frame info for regs that
11555 (rs6000_emit_epilogue): Likewise.
11557 2017-08-16 Nathan Sidwell <nathan@acm.org>
11559 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
11560 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
11561 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
11562 * tree.c (free_lang_data_in_type): Use else-if chain. Always
11563 clear TYPE_LANG_1. Remove obsolete member-function stripping.
11564 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
11565 (verify_type): Adjust for TYPE_BINFO move.
11566 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
11567 process TYPE_BINFO directly.
11568 (hash_tree): Likewise.
11569 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11571 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11574 2017-08-16 David Malcolm <dmalcolm@redhat.com>
11576 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
11578 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
11581 * config/i386/i386.md (*anddi_1_btr): Change predicates of
11582 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
11583 Add ix86_binary_operator_ok to insn constraint.
11584 (*iordi_1_bts): Ditto.
11585 (*xordi_1_btc): Ditto.
11586 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
11587 Update corresponding peephole2 pattern.
11591 2017-08-16 Bin Cheng <bin.cheng@arm.com>
11593 PR tree-optimization/81832
11594 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
11595 copy loop header which has IFN_LOOP_DIST_ALIAS call.
11597 2017-08-16 Marek Polacek <polacek@redhat.com>
11600 * fold-const.c (fold_indirect_ref_1): Restore original behavior
11601 regarding size_zero_node.
11603 2017-08-16 Martin Liska <mliska@suse.cz>
11606 * config.gcc: Respect previously set extra_objs in case
11609 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
11611 PR tree-optimization/81835
11612 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
11613 the phi SSA_NAME. Check that the condition in a COND_EXPR does
11614 not depend on the phi.
11616 2017-08-16 Alan Modra <amodra@gmail.com>
11618 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
11621 2017-08-16 Alan Modra <amodra@gmail.com>
11623 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
11624 (save_reg_p): ..into this. Update all callers.
11625 (first_reg_to_save): Simplify.
11627 2017-08-16 Alan Modra <amodra@gmail.com>
11629 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
11632 2017-08-15 Joseph Myers <joseph@codesourcery.com>
11636 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
11637 constant count if that count is less than 32.
11639 2017-08-15 Nathan Sidwell <nathan@acm.org>
11641 * gcc.c (execute): Emit friendlier message if inferior is killed
11642 by an external cause.
11644 2017-08-15 Richard Biener <rguenther@suse.de>
11646 PR tree-optimization/81790
11647 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
11648 CONSTRUCTORs from simplifying and VN.
11650 2017-08-14 Martin Sebor <msebor@redhat.com>
11652 * builtin-attrs.def: Add comments.
11654 2017-08-14 Martin Sebor <msebor@redhat.com>
11657 * doc/extend.texi (attribute nonstring): Document new attribute.
11659 2017-08-14 Martin Sebor <msebor@redhat.com>
11662 * tree-diagnostic.c (default_tree_printer): Handle %G.
11663 * gimple-pretty-print.h (percent_G_format): Declare new function.
11664 * gimple-pretty-print.c (percent_G_format): Define.
11665 * tree-pretty-print.c (percent_K_format): Add argument.
11667 2017-08-14 Martin Sebor <msebor@redhat.com>
11669 PR translation/79998
11670 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
11671 Remove a stray space.
11673 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
11676 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
11677 (*iordi_1_bts): Ditto.
11678 (*xordi_1_btc): Ditto.
11680 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11683 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
11684 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11686 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
11687 quoted strings, and make more translator-friendly.
11688 (darwin_rs6000_override_options): Likewise.
11689 (rs6000_option_override_internal): Likewise.
11690 (rs6000_return_in_memory): Fix overlong line.
11691 (init_cmulative_args): Use quoted strings, and make more
11692 translator-friendly.
11693 (rs6000_pass_by_reference): Fix overlong line.
11694 (def_builtin): Use quoted strings.
11695 (altivec_expand_predicate_builtin): Use quoted strings, and make
11696 more translator-friendly.
11697 (htm_expand_builtin): Use quoted strings.
11698 (cpu_expand_builtin): Use quoted strings, and make more
11699 translator-friendly.
11700 (altivec_expand_builtin): Likewise.
11701 (paired_expand_predicate_builtin): Likewise.
11702 (rs6000_invalid_builtin): Likewise.
11703 (builtin_function_type): Use quoted strings.
11704 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
11705 more translator-friendly.
11706 (rs6000_trampoline_init): Likewise.
11707 (rs6000_handle_altivec_attribute): Likewise.
11708 (rs6000_inner_target_options): Use quoted strings.
11709 (rs6000_disable_incompatible_switches): Likewise.
11710 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
11711 strings, and make more translator-friendly.
11712 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
11714 2017-08-14 Bin Cheng <bin.cheng@arm.com>
11716 PR tree-optimization/81799
11717 * tree-loop-distribution.c (version_loop_by_alias_check): Force
11718 cond_expr to simple gimple operand.
11720 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
11722 PR middle-end/46932
11723 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
11725 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
11729 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
11730 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
11731 TARGET_GASISR_PROLOGUES.
11732 * config/avr/avr.c (avr_option_override): Same.
11733 (avr_pass_pre_proep::execute): Same.
11735 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
11738 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
11739 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
11741 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
11743 * config/i386/i386.md (*load_tp_<mode>): Redefine as
11744 define_insn_and_split. Split to a memory load from 0 in
11745 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
11746 using PTR mode iterator.
11747 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
11748 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
11749 (*add_tp_<mode>): Redefine as define_insn_and_split.
11750 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
11751 address space. Merge with *add_tp_x32 using PTR mode iterator.
11752 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
11753 Split to an add with a memory load from 0 in
11754 DEFAULT_TLS_SEG_REG address space.
11756 2017-08-12 Andrew Pinski <apinski@cavium.com>
11758 * config/aarch64/aarch64-option-extensions.def (rdma):
11759 Fix feature string to what Linux prints out in /proc/cpuinfo.
11761 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
11764 * dwarf2out.c (modified_type_die): For C typedef types that have
11765 an ultimate origin, process the ultimate origin instead of the
11767 (gen_typedef_die): Assert that input DECLs have no ultimate
11769 (gen_type_die_with_usage): For typedef variants that have an
11770 ultimate origin, just call gen_decl_die on the original DECL.
11771 (process_scope_var): Avoid creating DIEs for local typedefs and
11772 concrete static variables.
11774 2017-08-12 Alan Modra <amodra@gmail.com>
11778 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
11779 match gnu-user.h startfile.
11780 (ENDFILE_LINUX_SPEC): Similarly.
11782 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
11785 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
11787 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
11789 2017-08-11 Tamar Christina <tamar.christina@arm.com>
11790 * config/aarch64/aarch64.md (mov<mode>): Change.
11791 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11792 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
11793 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
11795 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
11797 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
11798 for storage order barriers.
11800 2017-08-11 Martin Liska <mliska@suse.cz>
11803 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
11804 variables of void type.
11806 2017-08-11 Martin Liska <mliska@suse.cz>
11808 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
11809 TARGET_SUPPORTS_ALIASES.
11810 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
11811 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
11812 (optimize_weakref): Likewise.
11813 * symtab.c (symtab_node::noninterposable_alias): Likewise.
11814 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
11815 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
11817 2017-08-11 Martin Liska <mliska@suse.cz>
11820 * config/i386/i386.c (make_resolver_func): Do complete
11821 refactoring of the function.
11823 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
11826 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
11827 * config/i386/i386.c (ix86_stack_protect_guard): Use
11828 ix86_stack_protect_guard_symbol_str to generate varible declaration.
11829 * doc/invoke.texi (x86 Options): Document
11830 -mstack-protector-guard-symbol= option.
11832 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
11834 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
11835 * config/i386/i386.c (ix86_split_stack_guard): New function.
11836 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
11837 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
11838 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
11839 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
11840 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
11841 (split_stack_space_check): Call ix86_split_stack_guard.
11843 2017-08-10 Martin Sebor <msebor@redhat.com>
11845 * print-tree.c (print_node): Print location using the established
11847 Replace spaces with colons.
11848 (debug_raw, debug): Ditto.
11850 2017-08-10 Martin Sebor <msebor@redhat.com>
11853 * pretty-print.c (pp_format): Correct the handling of %s precision.
11855 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
11858 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11860 (ix86_finalize_stack_frame_flags): This. Also clear
11861 frame_pointer_needed if -fno-omit-frame-pointer is used without
11863 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11864 with ix86_finalize_stack_frame_flags.
11865 (ix86_expand_epilogue): Likewise.
11866 (ix86_expand_split_stack_prologue): Likewise.
11867 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
11869 2017-08-10 Martin Liska <mliska@suse.cz>
11872 * c-attribs.c (handle_target_attribute):
11873 Report warning for an empty string argument of target attribute.
11875 2017-08-09 Jakub Jelinek <jakub@redhat.com>
11878 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
11880 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
11881 or DECL_NONLOCAL labels.
11882 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
11883 or DECL_NONLOCAL labels here.
11885 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
11887 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
11888 to indicate when early gimple folding has been disabled.
11889 (rs6000_gimple_fold_builtin): Add debug content.
11890 (rs6000_invalid_builtin): Fix whitespace.
11891 (rs6000_expand_builtin): Fix whitespace.
11892 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
11894 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
11897 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
11898 SAVE_MULTIPLE if not all the registers that saves, should be saved.
11900 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
11902 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
11903 (qdf24xx): Likewise.
11904 * config/aarch64/aarch64.md: Include falkor.md.
11905 * config/aarch64/falkor.md: New.
11907 2017-08-09 Marek Polacek <polacek@redhat.com>
11910 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
11911 * diagnostic.c (emit_diagnostic): Add a comment.
11912 (emit_diagnostic_valist): New function.
11914 2017-08-09 Marek Polacek <polacek@redhat.com>
11917 * input.c (make_location): New overload.
11918 * input.h (make_location): Declare.
11920 2017-08-08 Alan Modra <amodra@gmail.com>
11921 H.J. Lu <hongjiu.lu@intel.com>
11924 * gcc.c (NO_PIE_SPEC): Delete.
11925 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
11927 (LINK_PIE_SPEC): ..to here.
11928 (LINK_COMMAND_SPEC): Support -no-pie.
11929 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
11930 chain of crtbegin*.o selection, update for PIE_SPEC changes and
11932 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
11933 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
11934 (ENDFILE_CRTEND_SPEC): Similarly.
11936 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
11939 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
11940 (mstack-protector-guard-offset=): Ditto.
11941 * config/i386/i386.c (ix86_option_override): Handle
11942 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
11944 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
11945 ix86_stack_protect_guard_offset variables.
11946 (TARGET_STACK_PROTECT_GUARD): Always define.
11947 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
11948 and -mstack-protector-guard-offset= options.
11950 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11952 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
11953 boundary case for the last candidate.
11955 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11957 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
11958 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
11960 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11962 PR middle-end/19706
11963 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
11964 * config/aarch64/aarch64-builtins.c
11965 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
11966 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
11967 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
11969 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11970 Andrew Pinski <pinskia@gmail.com>
11972 PR middle-end/19706
11973 * internal-fn.def (XORSIGN): New.
11974 * optabs.def (xorsign_optab): New.
11975 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
11976 (convert_expand_mult_copysign): New.
11977 (pass_optimize_widening_mul::execute): Call
11978 convert_expand_mult_copysign.
11980 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11982 PR tree-optimization/81354
11983 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
11984 Insert on edges rather than explicitly creating landing pads.
11985 (analyze_candidates_and_replace): Commit edge inserts.
11987 2017-08-08 Richard Biener <rguenther@suse.de>
11989 PR middle-end/81719
11990 * tree-ssa-loop-niter.c: Include tree-dfa.h.
11991 (expand_simple_operations): Also look through ADDR_EXPRs with
11992 MEM_REF bases treating them as POINTER_PLUS_EXPR.
11994 2017-08-08 Richard Biener <rguenther@suse.de>
11996 PR tree-optimization/81723
11997 * tree-vect-slp.c (struct bst_traits): New hash traits.
11998 (bst_fail): New global.
11999 (vect_build_slp_tree_2): New worker, split out from ...
12000 (vect_build_slp_tree): ... this now wrapping it with using
12001 bst_fail set to cache SLP tree build fails. Properly handle
12003 (vect_analyze_slp_instance): Allocate and free bst_fail.
12005 2017-08-08 Martin Liska <mliska@suse.cz>
12008 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
12009 LABEL_DECLs that can be from a different function.
12011 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12013 PR tree-optimization/81744
12014 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
12015 loop's number of iterations.
12017 2017-08-08 Martin Liska <mliska@suse.cz>
12019 * asan.c: Include header files.
12020 * attribs.c (build_decl_attribute_variant): New function moved
12022 (build_type_attribute_qual_variant): Likewise.
12023 (cmp_attrib_identifiers): Likewise.
12024 (simple_cst_list_equal): Likewise.
12025 (omp_declare_simd_clauses_equal): Likewise.
12026 (attribute_value_equal): Likewise.
12027 (comp_type_attributes): Likewise.
12028 (build_type_attribute_variant): Likewise.
12029 (lookup_ident_attribute): Likewise.
12030 (remove_attribute): Likewise.
12031 (merge_attributes): Likewise.
12032 (merge_type_attributes): Likewise.
12033 (merge_decl_attributes): Likewise.
12034 (merge_dllimport_decl_attributes): Likewise.
12035 (handle_dll_attribute): Likewise.
12036 (attribute_list_equal): Likewise.
12037 (attribute_list_contained): Likewise.
12038 * attribs.h (lookup_attribute): New function moved from tree.[ch].
12039 (lookup_attribute_by_prefix): Likewise.
12040 * bb-reorder.c: Include header files.
12041 * builtins.c: Likewise.
12042 * calls.c: Likewise.
12043 * cfgexpand.c: Likewise.
12044 * cgraph.c: Likewise.
12045 * cgraphunit.c: Likewise.
12046 * convert.c: Likewise.
12047 * dwarf2out.c: Likewise.
12048 * final.c: Likewise.
12049 * fold-const.c: Likewise.
12050 * function.c: Likewise.
12051 * gimple-expr.c: Likewise.
12052 * gimple-fold.c: Likewise.
12053 * gimple-pretty-print.c: Likewise.
12054 * gimple.c: Likewise.
12055 * gimplify.c: Likewise.
12056 * hsa-common.c: Likewise.
12057 * hsa-gen.c: Likewise.
12058 * internal-fn.c: Likewise.
12059 * ipa-chkp.c: Likewise.
12060 * ipa-cp.c: Likewise.
12061 * ipa-devirt.c: Likewise.
12062 * ipa-fnsummary.c: Likewise.
12063 * ipa-inline.c: Likewise.
12064 * ipa-visibility.c: Likewise.
12066 * lto-cgraph.c: Likewise.
12067 * omp-expand.c: Likewise.
12068 * omp-general.c: Likewise.
12069 * omp-low.c: Likewise.
12070 * omp-offload.c: Likewise.
12071 * omp-simd-clone.c: Likewise.
12072 * opts-global.c: Likewise.
12073 * passes.c: Likewise.
12074 * predict.c: Likewise.
12075 * sancov.c: Likewise.
12076 * sanopt.c: Likewise.
12077 * symtab.c: Likewise.
12078 * toplev.c: Likewise.
12079 * trans-mem.c: Likewise.
12080 * tree-chkp.c: Likewise.
12081 * tree-eh.c: Likewise.
12082 * tree-into-ssa.c: Likewise.
12083 * tree-object-size.c: Likewise.
12084 * tree-parloops.c: Likewise.
12085 * tree-profile.c: Likewise.
12086 * tree-ssa-ccp.c: Likewise.
12087 * tree-ssa-live.c: Likewise.
12088 * tree-ssa-loop.c: Likewise.
12089 * tree-ssa-sccvn.c: Likewise.
12090 * tree-ssa-structalias.c: Likewise.
12091 * tree-ssa.c: Likewise.
12092 * tree-streamer-in.c: Likewise.
12093 * tree-vectorizer.c: Likewise.
12094 * tree-vrp.c: Likewise.
12095 * tsan.c: Likewise.
12096 * ubsan.c: Likewise.
12097 * varasm.c: Likewise.
12098 * varpool.c: Likewise.
12099 * tree.c: Remove functions moved to attribs.[ch].
12100 * tree.h: Likewise.
12101 * config/aarch64/aarch64.c: Add attrs.h header file.
12102 * config/alpha/alpha.c: Likewise.
12103 * config/arc/arc.c: Likewise.
12104 * config/arm/arm.c: Likewise.
12105 * config/avr/avr.c: Likewise.
12106 * config/bfin/bfin.c: Likewise.
12107 * config/c6x/c6x.c: Likewise.
12108 * config/cr16/cr16.c: Likewise.
12109 * config/cris/cris.c: Likewise.
12110 * config/darwin.c: Likewise.
12111 * config/epiphany/epiphany.c: Likewise.
12112 * config/fr30/fr30.c: Likewise.
12113 * config/frv/frv.c: Likewise.
12114 * config/ft32/ft32.c: Likewise.
12115 * config/h8300/h8300.c: Likewise.
12116 * config/i386/winnt.c: Likewise.
12117 * config/ia64/ia64.c: Likewise.
12118 * config/iq2000/iq2000.c: Likewise.
12119 * config/lm32/lm32.c: Likewise.
12120 * config/m32c/m32c.c: Likewise.
12121 * config/m32r/m32r.c: Likewise.
12122 * config/m68k/m68k.c: Likewise.
12123 * config/mcore/mcore.c: Likewise.
12124 * config/microblaze/microblaze.c: Likewise.
12125 * config/mips/mips.c: Likewise.
12126 * config/mmix/mmix.c: Likewise.
12127 * config/mn10300/mn10300.c: Likewise.
12128 * config/moxie/moxie.c: Likewise.
12129 * config/msp430/msp430.c: Likewise.
12130 * config/nds32/nds32-isr.c: Likewise.
12131 * config/nds32/nds32.c: Likewise.
12132 * config/nios2/nios2.c: Likewise.
12133 * config/nvptx/nvptx.c: Likewise.
12134 * config/pa/pa.c: Likewise.
12135 * config/pdp11/pdp11.c: Likewise.
12136 * config/powerpcspe/powerpcspe.c: Likewise.
12137 * config/riscv/riscv.c: Likewise.
12138 * config/rl78/rl78.c: Likewise.
12139 * config/rx/rx.c: Likewise.
12140 * config/s390/s390.c: Likewise.
12141 * config/sh/sh.c: Likewise.
12142 * config/sol2.c: Likewise.
12143 * config/sparc/sparc.c: Likewise.
12144 * config/spu/spu.c: Likewise.
12145 * config/stormy16/stormy16.c: Likewise.
12146 * config/tilegx/tilegx.c: Likewise.
12147 * config/tilepro/tilepro.c: Likewise.
12148 * config/v850/v850.c: Likewise.
12149 * config/vax/vax.c: Likewise.
12150 * config/visium/visium.c: Likewise.
12151 * config/xtensa/xtensa.c: Likewise.
12153 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12156 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
12157 constraints since the -mupper-regs-* switches have been
12159 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
12160 into a vector from a double word element that was extracted from
12161 another vector, and eliminate extra XXPERMDI instructions.
12162 (vsx_concat_<mode>_2): Likewise.
12163 (vsx_concat_<mode>_3): Likewise.
12164 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
12165 concat to allow optimizing inserts from previous extracts.
12167 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
12169 * config/i386/i386.c (ix86_stack_protect_guard): Generate
12170 memory reference to a SSP offset in TLS address space.
12171 (ix86_print_operand) <case '@'>: Remove.
12172 (ix86_print_operand_punct_valid_p): Remove '@' code.
12173 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
12174 UNSPEC_SP_TLS_TEST.
12175 (stack_tls_protect_set_<mode>): Remove.
12176 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
12177 (stack_tls_protect_test_<mode>): Remove.
12178 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
12180 2017-08-07 Olivier Hainque <hainque@adacore.com>
12183 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
12185 2017-08-07 Douglas Rupp <rupp@adacore.com>
12187 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
12188 variable was referenced as multidir in command.
12190 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12193 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
12196 2017-08-07 Martin Liska <mliska@suse.cz>
12198 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
12199 * config/rl78/rl78.c: Add include of attribs.h.
12200 * config/sh/sh.c: Likewise.
12201 * config/v850/v850.c: Likewise.
12203 2017-08-07 Tom de Vries <tom@codesourcery.com>
12205 PR middle-end/78266
12206 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
12208 2017-08-07 Martin Liska <mliska@suse.cz>
12210 * config/mips/mips.c: Include attribs.h.
12212 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
12215 * doc/invoke.texi: Document change in behvaior for -Ofast for
12218 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
12220 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
12222 (aarch64_pop_regs): Likewise.
12223 (aarch64_gen_load_pair): Likewise.
12224 (aarch64_save_callee_saves): Likewise.
12225 (aarch64_restore_callee_saves): Likewise.
12227 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12229 * config/i386/i386.c: Revert the last change.
12231 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12234 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12236 (ix86_finalize_stack_frame_flags): This. Also clear
12237 frame_pointer_needed if -fno-omit-frame-pointer is used without
12239 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12240 with ix86_finalize_stack_frame_flags.
12241 (ix86_expand_epilogue): Likewise.
12242 (ix86_expand_split_stack_prologue): Likewise.
12244 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12247 * config/i386/i386.c (get_builtin_code_for_version): Set priority
12248 to P_AES for Westmere.
12250 2017-08-07 Jonathan Yong <10walls@gmail.com>
12252 * config/i386/mingw.opt (fset-stack-executable): Removed.
12253 * config/i386/cygming.opt (fset-stack-executable): Moved
12255 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
12257 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
12259 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
12261 2017-08-07 Marek Polacek <polacek@redhat.com>
12263 PR middle-end/81737
12264 * fold-const.c (fold_indirect_ref_1): Check type_domain.
12266 2017-08-07 Martin Liska <mliska@suse.cz>
12268 * attribs.h (canonicalize_attr_name): New function.
12269 (cmp_attribs): Move from c-format.c and adjusted.
12270 (is_attribute_p): Moved from tree.h.
12271 * tree-inline.c: Add new includes.
12272 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
12273 (private_is_attribute_p): Remove.
12274 (private_lookup_attribute): Likewise.
12275 (private_lookup_attribute_by_prefix): Simplify.
12276 (remove_attribute): Use is_attribute_p.
12277 * tree.h: Remove removed declarations.
12279 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12281 PR middle-end/81698
12282 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
12283 instead of computing it in the function. Formatting fix.
12284 (expand_case): Don't rely on default_edge being the first edge,
12285 clear it if removing it, pass default_edge to
12286 emit_case_dispatch_table.
12287 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
12290 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
12292 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
12293 insn in the function, emit NOP after the insn.
12295 2017-08-06 Tom de Vries <tom@codesourcery.com>
12297 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
12300 2017-08-06 Tom de Vries <tom@codesourcery.com>
12302 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
12305 2017-08-04 Yury Gribov <tetra2005@gmail.com>
12307 PR tree-optimization/57371
12308 * match.pd: New pattern.
12310 2017-08-04 Marek Polacek <polacek@redhat.com>
12312 PR middle-end/81695
12313 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
12314 perform the computation in offset_int.
12316 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12318 PR tree-optimization/81136
12319 * tree-vectorizer.h: Include tree-hash-traits.h.
12320 (vec_base_alignments): New typedef.
12321 (vec_info): Add a base_alignments field.
12322 (vect_record_base_alignments): Declare.
12323 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
12325 (DR_IS_CONDITIONAL_IN_STMT): New macro.
12326 (create_data_ref): Add an is_conditional_in_stmt argument.
12327 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
12328 the is_conditional_in_stmt field.
12329 (data_ref_loc): Add an is_conditional_in_stmt field.
12330 (get_references_in_stmt): Set the is_conditional_in_stmt field.
12331 (find_data_references_in_stmt): Update call to create_data_ref.
12332 (graphite_find_data_references_in_stmt): Likewise.
12333 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
12334 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
12335 (vect_record_base_alignment): New function.
12336 (vect_record_base_alignments): Likewise.
12337 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
12338 for nested statements even if we fail to compute a misalignment.
12339 Use pooled base alignments for unconditional references.
12340 (vect_find_same_alignment_drs): Compare base addresses instead
12342 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
12343 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
12345 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12347 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
12348 Add an explicit name for the enum. Use auto_vec for slp_instances
12349 and grouped_stores.
12350 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
12352 (_bb_vec_info): Add a constructor and destructor.
12353 (vinfo_for_stmt): Return NULL for uids of -1 as well.
12354 (destroy_loop_vec_info): Delete.
12355 (vect_destroy_datarefs): Likewise.
12356 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
12357 (vec_info::vec_info): New function.
12358 (vec_info::~vec_info): Likewise.
12359 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
12360 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
12361 destroy_loop_vec_info.
12362 * tree-vect-loop.c (new_loop_vec_info): Replace with...
12363 (_loop_vec_info::_loop_vec_info): ...this.
12364 (destroy_loop_vec_info): Replace with...
12365 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
12366 the stmt_vec_infos. Leave handling of vec_info information to its
12367 destructor. Remove explicit vector releases.
12368 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
12369 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
12370 * tree-vect-slp.c (new_bb_vec_info): Replace with...
12371 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
12372 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
12373 (destroy_bb_vec_info): Replace with...
12374 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
12375 information to its destructor.
12376 (vect_slp_analyze_bb_1): Use new and delete instead of
12377 new_bb_vec_info and destroy_bb_vec_info.
12378 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
12381 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12383 * tree-data-ref.h (subscript): Add access_fn field.
12384 (data_dependence_relation): Add could_be_independent_p.
12385 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
12386 (same_access_functions): Move to tree-data-ref.c.
12387 * tree-data-ref.c (ref_contains_union_access_p): New function.
12388 (access_fn_component_p): Likewise.
12389 (access_fn_components_comparable_p): Likewise.
12390 (dr_analyze_indices): Add a reference to access_fn_component_p.
12391 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
12393 (constant_access_functions): Likewise.
12394 (add_other_self_distances): Likewise.
12395 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
12396 (initialize_data_dependence_relation): Use XCNEW and remove
12397 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
12398 of access functions that have the same type. Allow the
12399 subsequence to end with different bases in some circumstances.
12400 Record the chosen access functions in SUB_ACCESS_FN.
12401 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
12402 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
12403 (subscript_dependence_tester_1): Likewise dra and drb.
12404 (build_classic_dist_vector): Update calls accordingly.
12405 (subscript_dependence_tester): Likewise.
12406 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
12407 DDR_COULD_BE_INDEPENDENT_P.
12408 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
12409 comp_alias_ddrs instead of may_alias_ddrs.
12410 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
12412 (vect_analyze_data_ref_dependence): Use it if
12413 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
12414 distance vectors if that fails.
12415 (dependence_distance_ge_vf): New function.
12416 (vect_prune_runtime_alias_test_list): Use it. Don't clear
12417 LOOP_VINFO_MAY_ALIAS_DDRS.
12419 2017-08-04 Richard Biener <rguenther@suse.de>
12421 PR middle-end/81705
12422 * fold-const.c (fold_binary_loc): Properly restrict
12423 minus_var0 && minus_var1 case when associating undefined overflow
12426 2017-08-04 Tom de Vries <tom@codesourcery.com>
12428 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
12430 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12432 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12433 Don't start diagnostic messages with a capital letter.
12434 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12436 (rs6000_invalid_builtin): Likewise.
12437 (rs6000_trampoline_init): Likewise.
12439 2017-08-03 Jakub Jelinek <jakub@redhat.com>
12442 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
12443 after setting changeable df flags.
12445 2017-08-03 Richard Biener <rguenther@suse.de>
12447 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
12448 up if the use is in USE - X.
12450 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
12452 * toplev.c (dumpfile.h): New include.
12453 (internal_error_reentered): New static function. Use it...
12454 (internal_error_function): ...here to handle reentered internal_error.
12456 2017-08-03 Richard Biener <rguenther@suse.de>
12458 PR middle-end/81148
12459 * fold-const.c (split_tree): Add minus_var and minus_con
12460 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
12461 here but always use minus_*.
12462 (associate_trees): Assert we never associate with MINUS_EXPR
12463 and NULL first operand. Do not recurse for PLUS_EXPR operands
12464 when associating as MINUS_EXPR either.
12465 (fold_binary_loc): Track minus_var and minus_con.
12467 2017-08-03 Tom de Vries <tom@codesourcery.com>
12470 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
12471 ACCEL_COMPILER, apply finish_options on
12472 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
12474 2017-08-03 Tom de Vries <tom@codesourcery.com>
12477 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
12478 function_entry_patch_area_size > 0.
12480 2017-08-03 Jakub Jelinek <jakub@redhat.com>
12483 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
12484 instead of 10??LU, perform unit multiplication in wide_int,
12485 don't change alloc_object_size_limit if the limit is larger
12488 PR tree-optimization/81655
12489 PR tree-optimization/81588
12490 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
12491 the case when ranges[i].low and high are 1 for unsigned type with
12494 PR middle-end/81052
12495 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
12496 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
12498 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12500 * tree-vrp.h: Add include guard.
12502 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
12505 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
12506 (ud2): New insn pattern.
12507 * config/i386/i386.c (ix86_expand_epilogue):
12508 For naked functions, generate ud2 instead of trap insn.
12510 2017-08-02 Marek Polacek <polacek@redhat.com>
12513 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
12515 2017-08-02 Tom de Vries <tom@codesourcery.com>
12516 Cesar Philippidis <cesar@codesourcery.com>
12518 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
12519 Add missing edge probabilities.
12521 2017-08-02 Tamar Christina <tamar.christina@arm.com>
12523 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
12524 Correct endianness.
12526 2017-08-02 Jakub Jelinek <jakub@redhat.com>
12528 PR middle-end/79499
12529 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
12530 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
12531 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
12532 prologue_seq sequences - if any.
12534 2017-08-02 Richard Biener <rguenther@suse.de>
12536 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
12537 via vectors if supported, integer extracts via punning if supported
12538 or otherwise vector extracts.
12540 2017-08-02 Richard Biener <rguenther@suse.de>
12542 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
12544 (bitmap_insert_into_set): ... this.
12546 2017-08-02 Richard Biener <rguenther@suse.de>
12548 PR tree-optimization/81633
12550 2015-08-17 Alan Hayward <alan.hayward@arm.com>
12552 PR tree-optimization/71752
12553 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
12555 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
12557 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
12558 (machine_function::call_ms2sysv_pad_out): Remove field.
12559 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
12560 (ix86_compute_frame_layout): Likewise.
12562 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
12565 * config/i386/i386.c (ix86_set_func_type): Disallow naked
12566 attribute with interrupt attribute.
12568 2017-08-01 Andrew Pinski <apinski@cavium.com>
12570 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
12571 BIT_INSERT_EXPR's operand 1
12572 to see if the types precision matches.
12574 2017-08-01 Martin Liska <mliska@suse.cz>
12576 PR middle-end/70140
12577 * builtins.c (expand_builtin_memcpy_args): Remove.
12578 (expand_builtin_memcpy): Call newly added function
12579 expand_builtin_memory_copy_args.
12580 (expand_builtin_memcpy_with_bounds): Likewise.
12581 (expand_builtin_mempcpy): Remove last argument.
12582 (expand_builtin_mempcpy_with_bounds): Likewise.
12583 (expand_builtin_memory_copy_args): New function created from
12584 expand_builtin_mempcpy_args with small modifications.
12585 (expand_builtin_mempcpy_args): Remove.
12586 (expand_builtin_stpcpy): Remove unused argument.
12587 (expand_builtin): Likewise.
12588 (expand_builtin_with_bounds): Likewise.
12590 2017-08-01 Martin Liska <mliska@suse.cz>
12593 Make mempcpy more optimal (PR middle-end/70140).
12595 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12598 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12599 __builtin_vec_cmpne verify both arguments are compatible vectors
12600 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
12601 verify arg1_type is a pointer or array type. For __builtin_vec_st,
12602 move computation of aligned to after checking the argument types.
12606 * config/rs6000/vsx.md (vextract_fp_from_shorth,
12607 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
12610 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
12611 Doug Rupp <rupp@adacore.com>
12612 Olivier Hainque <hainque@adacore.com>
12614 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
12615 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
12617 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
12618 for TARGET_OS_CPP_BUILTIN.
12619 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
12620 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
12622 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
12623 passing required abi options to the assembler for EABI configurations.
12624 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
12625 of .text.hot and .text.unlikely sections for kernel modules when
12626 using ARM style exceptions.
12627 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
12628 options. Add EXTRA_CC1_SPEC.
12629 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
12631 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
12633 (ARM_TARGET2_DWARF_FORMAT): Define.
12634 * config/arm/t-vxworks: Adjust multilib control to removal of the
12635 Diab command line options.
12637 2017-08-01 Martin Liska <mliska@suse.cz>
12639 PR gcov-profile/81561
12640 * gcov.c (unblock): Make unblocking safe as we need to preserve
12641 index correspondence of blocks and block_lists.
12643 2017-08-01 Richard Biener <rguenther@suse.de>
12645 PR tree-optimization/81181
12646 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
12647 (compute_antic): ... end of iteration here.
12649 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
12651 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
12652 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
12653 (ftree-slp-vectorize): Likewise.
12654 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
12655 can no longer be set independent of flag_tree_loop_vectorize.
12656 * omp-general.c (emp_max_vf): Likewise.
12657 * opts.c (enable_fdo_optimizations): Remove references to
12658 flag_tree_vectorize, these are now implicit.
12659 (common_handle_option): Remove handling for OPT_ftree_vectorize,
12660 and leave it for the options machinery.
12662 2017-08-01 Martin Liska <mliska@suse.cz>
12664 PR middle-end/70140
12665 * builtins.c (expand_builtin_memcpy_args): Remove.
12666 (expand_builtin_memcpy): Call newly added function
12667 expand_builtin_memory_copy_args.
12668 (expand_builtin_memcpy_with_bounds): Likewise.
12669 (expand_builtin_mempcpy): Remove last argument.
12670 (expand_builtin_mempcpy_with_bounds): Likewise.
12671 (expand_builtin_memory_copy_args): New function created from
12672 expand_builtin_mempcpy_args with small modifications.
12673 (expand_builtin_mempcpy_args): Remove.
12674 (expand_builtin_stpcpy): Remove unused argument.
12675 (expand_builtin): Likewise.
12676 (expand_builtin_with_bounds): Likewise.
12678 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
12681 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
12682 print "ds:" only for immediates in generic address space.
12684 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
12687 * config/i386/i386.c (ix86_funciton_naked): New prototype.
12688 (ix86_function_ok_for_sibcall): Return false for naked functions.
12690 2017-08-01 Richard Biener <rguenther@suse.de>
12692 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
12693 (compute_antic): Seed worklist with exit block predecessors.
12694 * cfganal.c (dfs_find_deadend): For a cycle return the source
12695 of the edge closing it.
12697 2017-08-01 Tamar Christina <tamar.christina@arm.com>
12699 * config/aarch64/aarch64.c
12700 (aarch64_can_const_movi_rtx_p): Move 0 check.
12702 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12704 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
12705 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
12707 * match.pd: Ditto in address comparison pattern.
12709 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12711 PR tree-optimization/81627
12712 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
12713 closed ssa form for store-store chain.
12715 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12717 PR tree-optimization/81620
12718 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
12719 for store-store chain.
12721 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12723 PR tree-optimization/81588
12724 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
12725 ranges[i].in_p, invert comparison code ccode. For >/>=,
12726 swap rhs1 and rhs2 and comparison code unconditionally,
12727 for </<= don't do that. Don't swap rhs1/rhs2 again if
12728 ranges[i].in_p, instead invert comparison code ccode if
12729 opcode or oe->rank is BIT_IOR_EXPR.
12732 * optabs.def (vec_extract_optab, vec_init_optab): Change from
12733 a direct optab to conversion optab.
12734 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
12735 with GET_MODE_INNER as last argument instead of optab_handler.
12736 * expmed.c (extract_bit_field_1): Likewise. Use vector from
12737 vector extraction if possible and optab is available.
12738 * expr.c (store_constructor): Use convert_optab_handler instead
12739 of optab_handler. Use vector initialization from smaller
12740 vectors if possible and optab is available.
12741 * tree-vect-stmts.c (vectorizable_load): Likewise.
12742 * doc/md.texi (vec_extract, vec_init): Document that the optabs
12743 now have two modes.
12744 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
12745 of vec_init from half-sized vectors with the same element mode.
12746 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
12747 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
12748 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
12749 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
12750 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
12751 after mode in gen_vec_extract* calls.
12752 (vec_extract<mode>): Renamed to ...
12753 (vec_extract<mode><ssescalarmodelower>): ... this.
12754 (vec_extract<mode><ssehalfvecmodelower>): New expander.
12755 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
12756 element mode after mode in gen_vec_init* calls.
12757 (VEC_INIT_HALF_MODE): New mode iterator.
12758 (vec_init<mode>): Renamed to ...
12759 (vec_init<mode><ssescalarmodelower>): ... this.
12760 (vec_init<mode><ssehalfvecmodelower>): New expander.
12761 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
12762 (vec_extractv2sfsf): ... this.
12763 (vec_initv2sf): Renamed to ...
12764 (vec_initv2sfsf): ... this.
12765 (vec_extractv2si): Renamed to ...
12766 (vec_extractv2sisi): ... this.
12767 (vec_initv2si): Renamed to ...
12768 (vec_initv2sisi): ... this.
12769 (vec_extractv4hi): Renamed to ...
12770 (vec_extractv4hihi): ... this.
12771 (vec_initv4hi): Renamed to ...
12772 (vec_initv4hihi): ... this.
12773 (vec_extractv8qi): Renamed to ...
12774 (vec_extractv8qiqi): ... this.
12775 (vec_initv8qi): Renamed to ...
12776 (vec_initv8qiqi): ... this.
12777 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
12778 (vec_init<mode>): Renamed to ...
12779 (vec_init<mode><VEC_base_l>): ... this.
12780 (vec_extract<mode>): Renamed to ...
12781 (vec_extract<mode><VEC_base_l>): ... this.
12782 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
12783 (vec_initv2sfsf): ... this.
12784 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
12785 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
12786 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
12787 element mode after mode in gen_vec_init* calls.
12788 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
12789 (vec_init<mode><Vel>): ... this.
12790 (vec_extract<mode>): Renamed to ...
12791 (vec_extract<mode><Vel>): ... this.
12792 * config/aarch64/iterators.md (Vel): New mode attribute.
12793 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
12794 Add element mode after mode in gen_vec_extract* calls.
12795 * config/s390/vector.md (non_vec_l): New mode attribute.
12796 (vec_extract<mode>): Renamed to ...
12797 (vec_extract<mode><non_vec_l>): ... this.
12798 (vec_init<mode>): Renamed to ...
12799 (vec_init<mode><non_vec_l>): ... this.
12800 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
12801 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
12803 * config/arm/iterators.md (V_elem_l): New mode attribute.
12804 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
12805 (vec_extract<mode><V_elem_l>): ... this.
12806 (vec_extractv2di): Renamed to ...
12807 (vec_extractv2didi): ... this.
12808 (vec_init<mode>): Renamed to ...
12809 (vec_init<mode><V_elem_l>): ... this.
12810 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
12811 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
12812 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
12813 Add element mode after gen_vec_extract* calls.
12814 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
12815 (vec_init<mode><unitmode>): ... this.
12816 (vec_extract<mode>): Renamed to ...
12817 (vec_extract<mode><unitmode>): ... this.
12818 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
12819 (vec_init<mode><unitmode>): ... this.
12820 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
12821 (vec_initv2sfsf): ... this.
12822 (vec_extractv2sf): Renamed to ...
12823 (vec_extractv2sfsf): ... this.
12824 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
12825 Add element mode after gen_vec_extract* calls.
12826 * config/mips/mips.md (unitmode): New mode iterator.
12827 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
12828 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
12829 * config/spu/spu.md (inner_l): New mode attribute.
12830 (vec_init<mode>): Renamed to ...
12831 (vec_init<mode><inner_l>): ... this.
12832 (vec_extract<mode>): Renamed to ...
12833 (vec_extract<mode><inner_l>): ... this.
12834 * config/sparc/sparc.md (veltmode): New mode iterator.
12835 (vec_init<VMALL:mode>): Renamed to ...
12836 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
12837 * config/ia64/vect.md (vec_initv2si): Renamed to ...
12838 (vec_initv2sisi): ... this.
12839 (vec_initv2sf): Renamed to ...
12840 (vec_initv2sfsf): ... this.
12841 (vec_extractv2sf): Renamed to ...
12842 (vec_extractv2sfsf): ... this.
12843 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
12844 (vec_init<mode>): Renamed to ...
12845 (vec_init<mode><VEC_base_l>): ... this.
12846 (vec_extract<mode>): Renamed to ...
12847 (vec_extract<mode><VEC_base_l>): ... this.
12848 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
12849 (vec_initv2sfsf): ... this.
12850 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
12851 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
12852 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
12853 gen_vec_init* calls.
12855 2017-08-01 Richard Biener <rguenther@suse.de>
12857 PR tree-optimization/81297
12858 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
12859 TREE_OVERFLOW from INTEGER_CSTs.
12861 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
12863 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
12865 2017-07-31 Carl Love <cel@us.ibm.com>
12867 * config/rs6000/rs6000-c: Add support for built-in functions
12868 vector signed char vec_xl_be (signed long long, signed char *);
12869 vector unsigned char vec_xl_be (signed long long, unsigned char *);
12870 vector signed int vec_xl_be (signed long long, signed int *);
12871 vector unsigned int vec_xl_be (signed long long, unsigned int *);
12872 vector signed long long vec_xl_be (signed long long, signed long long *);
12873 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
12874 vector signed short vec_xl_be (signed long long, signed short *);
12875 vector unsigned short vec_xl_be (signed long long, unsigned short *);
12876 vector double vec_xl_be (signed long long, double *);
12877 vector float vec_xl_be (signed long long, float *);
12878 * config/rs6000/altivec.h (vec_xl_be): Add #define.
12879 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
12880 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
12882 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
12883 (altivec_expand_builtin): Add switch statement to call
12884 altivec_expand_xl_be for each builtin.
12885 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
12886 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
12887 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
12888 __builtin_vsx_le_be_v16qi.
12889 * doc/extend.texi: Update the built-in documentation file for the
12890 new built-in functions.
12892 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
12895 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
12897 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12899 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12901 * config.gcc: Add z14.
12902 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
12903 CPU model numbers for z13s and z14.
12904 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
12906 * config/s390/s390-opts.h (enum processor_type): Rename
12907 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
12908 * config/s390/s390.c (processor_table): Add field for CPU name to
12909 be passed to Binutils.
12910 (s390_asm_output_machine_for_arch): Use the new field in
12911 processor_table for Binutils.
12912 (s390_expand_builtin): Replace arch12 with z14.
12913 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
12914 (s390_get_sched_attrmask): Likewise.
12915 (s390_get_unit_mask): Likewise.
12916 * config/s390/s390.opt: Add z14 to processor_type enum.
12918 2017-07-31 Martin Jambor <mjambor@suse.cz>
12921 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
12922 regardless of optimization level.
12924 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
12925 Martin Liska <mliska@suse.cz>
12927 * predict.def: Remove old comment and adjust probability.
12928 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
12929 PREDICT statements.
12931 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
12934 * config/i386/i386.c (ix86_function_naked): New function.
12935 (ix86_can_use_return_insn_p): Return false for naked functions.
12936 (ix86_expand_prologue): Skip prologue for naked functions.
12937 (ix86_expand_epilogue): Skip epilogue for naked functions
12938 and emit trap instruction.
12939 (ix86_warn_func_return): New function.
12940 (ix86_attribute_table): Add "naked" attribute specification.
12941 (TARGET_WARN_FUNC_RETURN): Define.
12942 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
12944 2017-07-31 Martin Liska <mliska@suse.cz>
12946 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
12947 (dump_gimple_bb_header): Always dump BB info.
12948 (pp_cfg_jump): Do not append info about BB when dumping a jump.
12950 2017-07-31 Martin Liska <mliska@suse.cz>
12953 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
12954 also with current_function_decl non-null equality.
12956 2017-07-31 Jakub Jelinek <jakub@redhat.com>
12959 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
12960 change type to the element type, instead add eltype variable and
12961 use it where we are interested in the element type.
12963 PR tree-optimization/81603
12964 * ipa-polymorphic-call.c
12965 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
12966 offset arithmetic in offset_int, bail out if the resulting bit offset
12967 doesn't fit into shwi.
12969 2017-07-31 Martin Liska <mliska@suse.cz>
12971 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
12972 (gimplify_save_expr): Fix comment.
12974 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
12977 * config/i386/i386.c (ix86_function_arg): Update arguments for
12979 (ix86_compute_frame_layout): Set the initial stack offset to
12980 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
12981 INCOMING_FRAME_SP_OFFSET.
12982 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
12983 stack before exception handler returns.
12984 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
12985 the 'ERROR_CODE' for exception handler.
12987 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
12989 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
12990 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
12991 (ASM_OUTPUT_REG_POP): Ditto.
12992 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
12993 instead of asm_fprintf to output pure string.
12995 2017-07-29 Jakub Jelinek <jakub@redhat.com>
12997 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
12998 to imported_module_or_decl hook.
12999 (debug_nothing_tree_tree_tree_bool): Remove.
13000 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
13001 * debug.c (do_nothing_debug_hooks): Use
13002 debug_nothing_tree_tree_tree_bool_bool instead of
13003 debug_nothing_tree_tree_tree_bool.
13004 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13005 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
13006 * sdbout.c (sdb_debug_hooks): Likewise.
13007 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
13008 (gen_namespace_die): Add DW_AT_export_symbols attribute if
13010 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
13011 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
13012 attribute, don't add anything.
13014 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13016 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
13017 (fold_build2_stat_loc): Likewise.
13018 (fold_build3_stat_loc): Likewise.
13019 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
13020 (fold_build1_loc): Remove macro.
13021 (fold_build2_loc): Likewise.
13022 (fold_build3_loc): Likewise.
13024 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13026 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
13027 (gimple_build_debug_bind_source_stat): Likewise.
13028 * gimple.h (gimple_build_debug_bind): Remove macro.
13029 (gimple_build_debug_bind_source): Likewise.
13031 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13033 * bitmap.c (bitmap_alloc): Adjust.
13034 (bitmap_gc_alloc): Likewise.
13035 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
13037 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13039 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
13040 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
13041 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
13042 (bitmap_gc_alloc_stat): Likewise.
13043 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
13045 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13047 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
13048 * rtl.h (shallow_copy_rtx): Remove macro.
13050 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13052 * emit-rtl.c (gen_raw_REG): Adjust.
13053 * gengenrtl.c (gendef): Likewise.
13054 * rtl.c (rtx_alloc_stat): Remove _stat from name.
13055 * rtl.h (rtx_alloc): Remove macro.
13057 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13059 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
13060 (build_tree_list_stat): Likewise.
13061 * tree.h (build_tree_list): Remove macro.
13062 (build_tree_list_vec): Likewise.
13064 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13066 * tree.c (make_vector_stat): Remove _stat from name.
13067 (build_vector_stat): Likewise.
13068 * tree.h (make_vector_stat): Remove macro.
13069 (build_vector_stat): Likewise.
13071 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13073 * tree.h (build_var_debug_value): Remove prototype.
13075 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13077 * tree.c (tree_cons_stat): Remove _stat from name.
13078 * tree.h (tree_cons): Remove macro.
13080 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13082 * tree.c (build_vl_exp_stat): Remove _stat from name.
13083 * tree.h (build_vl_exp): Remove macro.
13085 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13087 * tree.c (build_decl_stat): Remove _stat from name.
13088 * tree.h (build_decl): Remove macro.
13090 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13092 * gimple.c (gimple_build_with_ops_stat): Adjust.
13093 (gimple_alloc_stat): Remove _stat from name.
13094 * gimple.h (gimple_alloc): Remove macro.
13096 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13098 * tree.c (make_tree_vec_stat): Remove _stat from name.
13099 (grow_tree_vec_stat): Likewise.
13100 * tree.h (make_tree_vec_stat): Adjust prototype.
13101 (grow_tree_vec_stat): Likewise.
13102 (make_tree_vec): Remove macro.
13103 (grow_tree_vec): Likewise.
13105 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13107 * fold-const.c (fold_build1_stat_loc): Adjust.
13108 (fold_build2_stat_loc): Likewise.
13109 (fold_build3_stat_loc): Likewise.
13110 * tree.c (build0_stat): Remove _stat from name.
13111 (build1_stat): Likewise.
13112 (build2_stat): Likewise.
13113 (build3_stat): Likewise.
13114 (build4_stat): Likewise.
13115 (build5_stat): Likewise.
13116 * tree.h (build1_loc): Remove macro, and rename _stat function
13118 (build2_loc): Likewise.
13119 (build3_loc): Likewise.
13120 (build4_loc): Likewise.
13121 (build5_loc): Likewise.
13123 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13125 * tree.c (make_int_cst_stat): Remove _stat from name.
13126 * tree.h (make_int_cst_stat): Adjust prototype.
13127 (make_int_cst): Remove macro.
13129 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13131 * tree.c (make_tre_binfo_stat): Remove _stat from name.
13132 * tree.h (make_tree_binfo_stat): Adjust prototype.
13133 (make_tree_binfo): Remove.
13135 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13137 * tree.c (copy_node_stat): Rename to copy_node.
13138 (build_distinct_type_copy): Adjust.
13139 * tree.h (copy_node_stat): Adjust prototype.
13140 (copy_node): Remove macro.
13142 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13144 * tree.c (make_node_stat): rename to make_node.
13145 (build_tree_list_stat): Adjust.
13146 (build0_stat): Likewise.
13147 (build2_stat): Likewise.
13148 (build3_stat): Likewise.
13149 (build4_stat): Likewise.
13150 (build5_stat): Likewise.
13151 (build_decl_stat): Likewise.
13152 * tree.h (make_node_stat): Adjust prototype.
13153 (make_node): remove macro.
13155 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
13157 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
13158 (PPC_FEATURE2_SCV): Likewise.
13159 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
13161 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13163 * config/aarch64/aarch64.c
13164 (aarch64_internal_mov_immediate): Add new special pattern.
13165 * config/aarch64/aarch64.md (*movdi_aarch64):
13166 Add reg/32bit const mov case.
13168 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13169 Richard Sandiford <richard.sandiford@linaro.org>
13171 * config/aarch64/aarch64.md (mov<mode>): Generalize.
13172 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13173 Add integer and movi cases.
13174 (movi-split-hf-df-sf split, fp16): New.
13175 (enabled): Added TARGET_FP_F16INST.
13176 * config/aarch64/iterators.md (GPF_HF): New.
13177 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
13179 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13181 * config/aarch64/aarch64.c
13182 (aarch64_simd_container_mode): Add prototype.
13183 (aarch64_expand_mov_immediate): Add HI support.
13184 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
13185 (aarch64_can_const_movi_rtx_p): New.
13186 (aarch64_preferred_reload_class):
13187 Remove restrictions of using FP registers for certain SIMD operations.
13188 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
13189 (aarch64_valid_floating_const): Add integer move validation.
13190 (aarch64_simd_imm_scalar_p): Remove.
13191 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
13192 (aarch64_legitimate_constant_p): Expand list of supported cases.
13193 * config/aarch64/aarch64-protos.h
13194 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
13195 (aarch64_reinterpret_float_as_int): New.
13196 (aarch64_simd_imm_scalar_p): Remove.
13197 * config/aarch64/constraints.md (Uvi): New.
13198 (Dd): Split into Ds and new Dd.
13199 * config/aarch64/aarch64.md (*movsi_aarch64):
13201 (*movdi_aarch64): Add SIMD mov case.
13203 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13205 * tree-predcom.c: (struct chain): Handle store-store chain in which
13206 stores for elimination only store loop invariant values.
13207 (execute_pred_commoning_chain): Ditto.
13208 (prepare_initializers_chain_store_elim): Ditto.
13209 (prepare_finalizers): Ditto.
13210 (is_inv_store_elimination_chain): New function.
13211 (initialize_root_vars_store_elim_1): New function.
13213 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13215 * tree-predcom.c: Revise general description of the pass.
13216 (enum chain_type): New enum type for store elimination.
13217 (struct chain): New field supporting store elimination.
13218 (struct component): Ditto.
13219 (dump_chain): Dump store-stores chain.
13220 (release_chain): Release resources.
13221 (split_data_refs_to_components): Compute and create component
13222 contains only stores for elimination.
13223 (get_chain_last_ref_at): New function.
13224 (make_invariant_chain): Initialization.
13225 (make_rooted_chain): Specify chain type in parameter and record it.
13226 (add_looparound_copies): Skip for store-stores chain.
13227 (determine_roots_comp): Compute type of chain and pass it to
13229 (initialize_root_vars_store_elim_2): New function.
13230 (finalize_eliminated_stores): New function.
13231 (remove_stmt): Handle store for elimination.
13232 (execute_pred_commoning_chain): Execute predictive commoning on
13233 store-store chains.
13234 (determine_unroll_factor): Skip unroll for store-stores chain.
13235 (prepare_initializers_chain_store_elim): New function.
13236 (prepare_initializers_chain): Hanlde store-store chain.
13237 (prepare_finalizers_chain, prepare_finalizers): New function.
13238 (tree_predictive_commoning_loop): Return integer value indicating
13239 if loop is unrolled or lcssa form is corrupted.
13240 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
13242 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13244 * tree-predcom.c (initialize_root): Delete.
13245 (execute_pred_commoning_chain): Initialize root vars and replace
13246 reference of non-combined chain directly, rather than call above
13249 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13251 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
13252 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
13254 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13256 * tree-predcom.c (struct chain): New field init_seq.
13257 (release_chain): Release init_seq.
13258 (prepare_initializers_chain): Record intialization stmts in above
13260 (insert_init_seqs): New function.
13261 (tree_predictive_commoning_loop): Call insert_init_seqs.
13263 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13265 * tree-predcom.c (determine_roots_comp): Skip trivial components.
13267 2017-07-28 Richard Biener <rguenther@suse.de>
13269 * match.pd: Remove superfluous :c.
13270 * genmatch.c (simplify::id): Add member.
13271 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
13273 (current_id): New global.
13274 (dt_node::parent): Move from ...
13275 (dt_operand::parent): ... here. Add for_id member.
13276 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
13277 (decision_tree::find_node): Relax order requirement when
13278 merging DT_TRUE nodes to ones inbetween the current simplify
13279 and the one we try to merge with. Add diagnostic whenever
13280 we need to enforce pattern order by not merging.
13281 (decision_tree::insert): Set current_id.
13282 (decision_tree::print_node): Dump parent node and for_id.
13283 (parser::last_id): Add member.
13284 (parser::push_simplify): Assign unique id.
13285 (parser::parser): Initialize last_id.
13287 2017-07-28 Martin Liska <mliska@suse.cz>
13290 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
13291 gimple_build_debug_bind.
13293 2017-07-28 Richard Biener <rguenther@suse.de>
13295 PR tree-optimization/81502
13296 * match.pd: Add pattern combining BIT_INSERT_EXPR with
13298 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
13300 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
13301 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
13302 for BIT_FIELD_REF args.
13303 * fold-const.c (make_bit_field_ref): Likewise.
13304 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
13306 2017-07-28 Jakub Jelinek <jakub@redhat.com>
13309 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
13310 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
13311 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
13312 Or it into SANITIZER_UNDEFINED.
13313 * ubsan.c: Include gimple-fold.h and varasm.h.
13314 (ubsan_expand_ptr_ifn): New function.
13315 (instrument_pointer_overflow): New function.
13316 (maybe_instrument_pointer_overflow): New function.
13317 (instrument_object_size): Formatting fix.
13318 (pass_ubsan::execute): Call instrument_pointer_overflow
13319 and maybe_instrument_pointer_overflow.
13320 * internal-fn.c (expand_UBSAN_PTR): New function.
13321 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
13322 * sanitizer.def (__ubsan_handle_pointer_overflow,
13323 __ubsan_handle_pointer_overflow_abort): New builtins.
13324 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
13325 * internal-fn.def (UBSAN_PTR): New internal function.
13326 * opts.c (sanitizer_opts): Add pointer-overflow.
13327 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
13328 * fold-const.c (build_range_check): Compute pointer range check in
13329 integral type if pointer arithmetics would be needed. Formatting
13332 2017-07-28 Martin Liska <mliska@suse.cz>
13335 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
13336 parameters that are of a variable-length.
13338 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
13340 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
13342 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
13343 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
13344 (CRT_CALL_STATIC_FUNCTION): Likewise.
13345 (ASM_DEFAULT_SPEC): New define.
13346 (ASM_SPEC32): Likewise.
13347 (ASM_SPEC64): Likewise.
13348 (ASM_SPEC_COMMON): Likewise.
13349 (ASM_SPEC): Likewise.
13350 (INVALID_64BIT): Likewise.
13351 (LINK_OS_DEFAULT_SPEC): Likewise.
13352 (LINK_OS_SPEC32): Likewise.
13353 (LINK_OS_SPEC64): Likewise.
13354 (POWERPC_LINUX): Likewise.
13355 (PTRDIFF_TYPE): Likewise.
13356 (RESTORE_FP_PREFIX): Likewise.
13357 (RESTORE_FP_SUFFIX): Likewise.
13358 (SAVE_FP_PREFIX): Likewise.
13359 (SAVE_FP_SUFFIX): Likewise.
13360 (SIZE_TYPE): Likewise.
13361 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
13362 (TARGET_64BIT): Likewise.
13363 (TARGET_64BIT): Likewise.
13364 (TARGET_AIX): Likewise.
13365 (WCHAR_TYPE_SIZE): Likewise.
13366 (WCHAR_TYPE): Undefine.
13367 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
13368 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
13369 (CPP_OS_RTEMS_SPEC): Delete.
13370 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
13371 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
13373 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
13375 2017-07-28 Jakub Jelinek <jakub@redhat.com>
13377 PR tree-optimization/81578
13378 * tree-parloops.c (build_new_reduction): Bail out if
13379 reduction_code isn't one of the standard OpenMP reductions.
13380 Move the details printing after that decision.
13382 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13384 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
13385 related to reload_in_progress.
13386 (splat_input_operand): Likewise.
13387 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
13389 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
13391 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
13392 (TARGET_INSTANTIATE_DECLS): Likewise.
13393 (legitimate_indexed_address_p): Delete reload_in_progress code.
13394 (rs6000_debug_legitimate_address_p): Likewise.
13395 (rs6000_eliminate_indexed_memrefs): Likewise.
13396 (rs6000_emit_le_vsx_store): Likewise.
13397 (rs6000_emit_move_si_sf_subreg): Likewise.
13398 (rs6000_emit_move): Likewise.
13399 (register_to_reg_type): Likewise.
13400 (rs6000_pre_atomic_barrier): Likewise.
13401 (rs6000_machopic_legitimize_pic_address): Likewise.
13402 (rs6000_allocate_stack_temp): Likewise.
13403 (rs6000_address_for_fpconvert): Likewise.
13404 (rs6000_address_for_altivec): Likewise.
13405 (rs6000_secondary_memory_needed_rtx): Delete function.
13406 (rs6000_check_sdmode): Likewise.
13407 (rs6000_alloc_sdmode_stack_slot): Likewise.
13408 (rs6000_instantiate_decls): Likewise.
13409 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
13410 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
13411 Delete reload_in_progress.
13412 (*vec_reload_and_plus_<mptrsize>): Likewise.
13413 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
13414 (vsx_div_v2di): Likewise.
13415 (vsx_udiv_v2di): Likewise.
13417 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13419 * config/rs6000/rs6000.opt (mlra): Replace with stub.
13420 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
13421 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
13422 (rs6000_debug_reg_global): Delete print of LRA status.
13423 (rs6000_option_override_internal): Delete dead LRA related code.
13424 (rs6000_lra_p): Delete function.
13425 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
13427 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
13429 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
13430 * config/riscv/rtems.h: New file.
13432 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13433 Sudakshina Das <sudi.das@arm.com>
13435 * config/aarch64/aarch64.md
13436 (define_split for and<mode>3nr_compare): Move
13437 non aarch64_logical_operand to a register.
13438 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
13439 register immediate operand to a register.
13440 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
13442 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13444 PR middle-end/81564
13445 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
13447 2017-07-27 Richard Biener <rguenther@suse.de>
13449 PR tree-optimization/81573
13450 PR tree-optimization/81494
13451 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
13452 multi defuse cycle case.
13454 2017-07-27 Richard Biener <rguenther@suse.de>
13456 PR tree-optimization/81571
13457 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
13460 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
13462 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
13463 earlier and only if MASK_FPU is set. Adjust formatting.
13465 2017-07-27 Martin Liska <mliska@suse.cz>
13467 * opt-functions.awk: Add validation of value of Init.
13468 * optc-gen.awk: Pass new argument.
13470 2017-07-27 Martin Liska <mliska@suse.cz>
13472 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
13473 Fix wrong condition.
13475 2017-07-27 Martin Liska <mliska@suse.cz>
13477 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
13478 BBs and edges seen by autoFDO.
13480 2017-07-27 Richard Biener <rguenther@suse.de>
13482 PR tree-optimization/81502
13483 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
13484 with incompatible but same sized type.
13485 (execute_update_addresses_taken): Likewise.
13487 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
13489 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
13490 flag_tree_loop_vectorize rather than flag_tree_vectorize.
13492 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13495 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
13496 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
13497 Change s_operand to memory_operand.
13499 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
13501 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
13502 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
13503 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
13504 Emit instructions rather than returning an expression. Handle TFmode
13505 and KFmode by casting to TImode.
13506 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
13507 (rs6000_emit_le_vsx_store): Likewise.
13508 * config/rs6000/vsx.md (VSX_TI): New iterator.
13509 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
13510 (*vsx_le_undo_permute_<mode>): Likewise.
13511 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
13512 emit the split sequence.
13513 (*vsx_le_perm_store_<mode>): Likewise.
13515 2017-07-27 Jakub Jelinek <jakub@redhat.com>
13517 PR tree-optimization/81555
13518 PR tree-optimization/81556
13519 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
13520 if true, force CHANGED for the recursive invocation.
13521 (reassociate_bb): Remember original length of ops array, pass
13522 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
13524 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
13525 attributes for noipa attribute. For naked attribute use
13526 lookup_attribute first before lookup_attribute_spec.
13527 * final.c (rest_of_handle_final): Disable IPA RA for functions with
13529 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
13530 for functions with noipa attribute.
13531 (cgraph_externally_visible_p): Return true for functions with noipa
13533 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
13534 for functions with noipa attribute.
13535 * doc/extend.texi: Document noipa function attribute.
13536 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
13537 also for functions with noipa attribute.
13538 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
13540 2017-07-26 Andrew Pinski <apinski@cavium.com>
13542 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
13543 vec_unalign_load_cost and vec_unalign_store_cost.
13545 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13547 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
13548 -mvsx-small-integer option.
13549 (ISA_3_0_MASKS_IEEE): Likewise.
13550 (OTHER_VSX_VECTOR_MASKS): Likewise.
13551 (POWERPC_MASKS): Likewise.
13552 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
13553 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
13554 code, only testing for DImode being allowed in non-VSX floating
13556 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
13557 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
13559 (rs6000_option_override_internal): Delete -mvsx-small-integer.
13560 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
13561 TARGET_P8_VECTOR test.
13562 (rs6000_secondary_reload_simple_move): Likewise.
13563 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
13564 since TARGET_P9_VECTOR was already tested.
13565 (rs6000_opt_masks): Remove -mvsx-small-integer.
13566 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
13567 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13569 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
13570 test for TARGET_VEXTRACTUB was used, and that uses
13572 (p9 extract splitter): Likewise.
13573 (vsx_extract_<mode>_di_p9): Likewise.
13574 (vsx_extract_<mode>_store_p9): Likewise.
13575 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
13576 for TARGET_P9_VECTOR was used. Delete code that is now dead with
13577 the elimination of TARGET_VSX_SMALL_INTEGER.
13578 (vsx_extract_<mode>_p8): Likewise.
13579 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
13580 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
13581 (vsx_set_<mode>_p9): Likewise.
13582 (vsx_set_v4sf_p9): Likewise.
13583 (vsx_set_v4sf_p9_zero): Likewise.
13584 (vsx_insert_extract_v4sf_p9): Likewise.
13585 (vsx_insert_extract_v4sf_p9_2): Likewise.
13586 * config/rs6000/rs6000.md (sign extend splitter): Change
13587 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
13588 (floatsi<mode>2_lfiwax_mem): Likewise.
13589 (floatunssi<mode>2_lfiwzx_mem): Likewise.
13590 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
13591 since a test for TARGET_P9_VECTOR was used.
13592 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13593 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
13594 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13595 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13596 TARGET_P8_VECTOR test.
13597 (fix_trunc<mode>si2_stfiwx): Likewise.
13598 (fix_trunc<mode>si2_internal): Likewise.
13599 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
13600 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13602 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13603 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13604 TARGET_P8_VECTOR test.
13605 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13606 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
13607 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13609 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13610 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
13611 since a test for TARGET_P9_VECTOR was used.
13612 (splitter for loading small constants): Likewise.
13614 2017-07-26 Andrew Pinski <apinski@cavium.com>
13616 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
13619 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
13622 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
13623 (fp_valid_at): Likewise.
13625 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13627 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
13628 (qdf24xx_addrcost_table): Likewise.
13629 (cortexa57_tunings): Update to use generic_branch_cost.
13630 (cortexa72_tunings): Likewise.
13631 (cortexa73_tunings): Likewise.
13632 (qdf24xx_tunings): Likewise.
13634 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13636 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
13637 (thunderx2t99_branch_cost): Likewise.
13638 (cortexa35_tunings): Update to use generic_branch_cost.
13639 (cortexa53_tunings): Likewise.
13640 (cortexa57_tunings): Likewise.
13641 (cortexa72_tunings): Likewise.
13642 (cortexa73_tunings): Likewise.
13643 (thunderx2t99_tunings): Likewise.
13645 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13647 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
13648 (sparc_option_override): Honour MASK_FSMULD.
13649 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
13650 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
13651 * config/sparc/sparc.opt (mfsmuld): New option.
13652 * doc/invoke.texi (mfsmuld): Document option.
13654 2017-07-26 Marek Polacek <polacek@redhat.com>
13656 PR middle-end/70992
13657 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
13659 2017-07-26 Richard Biener <rguenther@suse.de>
13661 * gimple-match-head.c (do_valueize): Return OP if valueize
13663 (get_def): New helper to get at the def stmt of a SSA name
13664 if valueize allows.
13665 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
13666 do_valueize to get at the def stmt.
13667 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
13669 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
13671 PR middle-end/46932
13672 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
13674 2017-07-26 Martin Liska <mliska@suse.cz>
13677 * function.c (expand_function_start): Make expansion of
13678 nonlocal_goto_save_area after parm_birth_insn.
13680 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13682 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
13683 from all CPU target flags enable members.
13685 2017-07-26 Richard Biener <rguenther@suse.de>
13687 * genmatch.c (dt_simplify::gen): Make iterator vars const.
13688 (decision_tree::gen): Make 'type' const.
13689 (write_predicate): Likewise.
13691 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
13693 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
13694 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
13695 (rs6000_option_override_internal): Likewise.
13696 (rs6000_expand_vector_set): Likewise.
13697 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
13698 (TARGET_UPPER_REGS_SF): Likewise.
13699 (TARGET_UPPER_REGS_DI): Likewise.
13700 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
13701 (TARGET_DIRECT_MOVE_64BIT): Likewise.
13702 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
13703 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13704 (Splitters for DI constants in Altivec registers): Likewise.
13705 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
13706 (vsx_set_v4sf_p9): Likewise.
13707 (vsx_set_v4sf_p9_zero): Likewise.
13708 (vsx_insert_extract_v4sf_p9): Likewise.
13709 (vsx_insert_extract_v4sf_p9_2): Likewise.
13711 2017-07-25 Carl Love <cel@us.ibm.com>
13713 * doc/extend.texi: Update the built-in documentation file for the
13714 existing built-in functions
13715 vector signed char vec_cnttz (vector signed char);
13716 vector unsigned char vec_cnttz (vector unsigned char);
13717 vector signed short vec_cnttz (vector signed short);
13718 vector unsigned short vec_cnttz (vector unsigned short);
13719 vector signed int vec_cnttz (vector signed int);
13720 vector unsigned int vec_cnttz (vector unsigned int);
13721 vector signed long long vec_cnttz (vector signed long long);
13722 vector unsigned long long vec_cnttz (vector unsigned long long);
13724 2017-07-25 Andrew Pinski <apinski@cavium.com>
13726 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
13727 accesses where the use is for the first operand of a BIT_INSERT.
13729 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
13732 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
13733 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
13735 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
13737 * config/i386/gstabs.h: Delete.
13738 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
13740 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
13742 * config/i386/i386.c (ix86_decompose_address): Do not check for
13743 register RTX when looking at index_reg or base_reg.
13744 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
13746 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
13748 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
13749 to update EH info here.
13751 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13753 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
13755 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13757 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
13759 2017-07-25 Torsten Duwe <duwe@suse.de>
13761 * common.opt: Introduce -fpatchable-function-entry
13762 command line option, and its variables function_entry_patch_area_size
13763 and function_entry_patch_area_start.
13764 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
13765 including a two-value parser.
13766 * target.def (print_patchable_function_entry): New target hook.
13767 * targhooks.h (default_print_patchable_function_entry): New function.
13768 * targhooks.c (default_print_patchable_function_entry): Likewise.
13769 * toplev.c (process_options): Switch off IPA-RA if
13770 patchable function entries are being generated.
13771 * varasm.c (assemble_start_function): Look at the
13772 patchable-function-entry command line switch and current
13773 function attributes and maybe generate NOP instructions by
13774 calling the print_patchable_function_entry hook.
13775 * doc/extend.texi: Document patchable_function_entry attribute.
13776 * doc/invoke.texi: Document -fpatchable_function_entry
13777 command line option.
13778 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
13780 * doc/tm.texi: Re-generate.
13782 2017-07-25 Jakub Jelinek <jakub@redhat.com>
13785 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
13786 TARGET_AVX512DQ rather than TARGET_AVX512BW.
13788 2017-07-25 Tamar Christina <tamar.christina@arm.com>
13790 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
13792 2017-07-25 Richard Biener <rguenther@suse.de>
13794 PR tree-optimization/81455
13795 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
13796 not walk in cycles when looking for guards.
13798 2017-07-25 Richard Biener <rguenther@suse.de>
13800 PR tree-optimization/81529
13801 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
13802 when optimizing backedge uses.
13804 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
13806 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
13808 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
13809 to dl_section_ref. On AIX, append an expression to subtract
13810 the size of the section length to dl_section_ref.
13812 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
13814 * configure.ac: If any of the config.* scripts fail, exit 1.
13815 * configure: Regenerate.
13817 2017-07-25 Richard Biener <rguenther@suse.de>
13819 PR middle-end/81546
13820 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
13821 of immediate uses, be more verbose on errors.
13823 2017-07-25 Richard Biener <rguenther@suse.de>
13825 PR tree-optimization/81510
13826 * tree-vect-loop.c (vect_is_simple_reduction): When the
13827 reduction stmt is not inside the loop bail out.
13829 2017-07-25 Richard Biener <rguenther@suse.de>
13831 PR tree-optimization/81303
13832 * tree-vect-loop-manip.c (vect_loop_versioning): Build
13833 profitability check against LOOP_VINFO_NITERSM1.
13835 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13837 * domwalk.c (cmp_bb_postorder): Simplify.
13838 (sort_bbs_postorder): New function. Use it...
13839 (dom_walker::walk): ...here to optimize common cases.
13841 2017-07-25 Martin Liska <mliska@suse.cz>
13844 * ipa-visibility.c (function_and_variable_visibility): Make the
13845 redirection just on target that supports aliasing.
13846 Fix GNU coding style.
13848 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
13851 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
13853 * config/arm/rtems.h: Likewise.
13854 * config/bfin/rtems.h: Likewise.
13855 * config/i386/rtemself.h: Likewise.
13856 * config/lm32/rtems.h: Likewise.
13857 * config/m32c/rtems.h: Likewise.
13858 * config/m68k/rtemself.h: Likewise.
13859 * config/microblaze/rtems.h: Likewise.
13860 * config/mips/rtems.h: Likewise.
13861 * config/moxie/rtems.h: Likewise.
13862 * config/nios2/rtems.h: Likewise.
13863 * config/powerpcspe/rtems.h: Likewise.
13864 * config/rs6000/rtems.h: Likewise.
13865 * config/rtems.h: Likewise.
13866 * config/sh/rtems.h: Likewise.
13867 * config/sh/rtemself.h: Likewise.
13868 * config/sparc/rtemself.h: Likewise.
13870 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
13873 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
13874 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
13875 * tree-ssa-structalias.c (alias_get_name): Same.
13877 2017-07-25 Bin Cheng <bin.cheng@arm.com>
13880 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
13881 instructions if no du chain is found.
13883 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
13885 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
13887 2017-07-25 Richard Biener <rguenther@suse.de>
13889 PR middle-end/81505
13890 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
13893 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
13895 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
13896 upper-regs options.
13897 (ISA_2_7_MASKS_SERVER): Likewise.
13898 (ISA_3_0_MASKS_IEEE): Likewise.
13899 (OTHER_P8_VECTOR_MASKS): Likewise.
13900 (OTHER_VSX_VECTOR_MASKS): Likewise.
13901 (POWERPC_MASKS): Likewise.
13902 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
13903 duplicate list of options.
13904 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
13905 explicit -mupper-regs options.
13906 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
13907 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
13908 alias for -mupper-regs-df.
13909 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13910 (rs6000_init_hard_regno_mode_ok): Likewise.
13911 (rs6000_option_override_internal): Likewise.
13912 (rs6000_opt_masks): Likewise.
13913 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
13914 options in terms of whether -mvsx or -mpower8-vector was used.
13915 (TARGET_UPPER_REGS_DI): Likewise.
13916 (TARGET_UPPER_REGS_SF): Likewise.
13917 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
13918 -mupper-regs-* options.
13920 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
13922 * passes.c (emergency_dump_function): Print some empty lines and a
13923 header before the RTL dump.
13925 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
13927 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
13929 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
13932 * config/aarch64/aarch64.c (aarch64_classify_symbol):
13933 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
13935 2017-07-24 Carl Love <cel@us.ibm.com>
13937 * config/rs6000/rs6000-c.c: Add support for built-in functions
13938 vector float vec_extract_fp32_from_shorth (vector unsigned short);
13939 vector float vec_extract_fp32_from_shortl (vector unsigned short);
13940 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
13941 vec_extract_fp_from_shortl): Add defines for the two builtins.
13942 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
13943 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
13945 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
13946 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
13947 * doc/extend.texi: Update the built-in documentation file for the
13948 new built-in function.
13950 2017-07-24 Jakub Jelinek <jakub@redhat.com>
13953 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
13955 * doc/generic.texi: Likewise.
13956 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
13957 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
13959 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
13961 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
13962 (aarch64_mls_elt_merge<mode>): Likewise.
13964 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
13966 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13967 having __cxa_atexit.
13969 2017-07-23 Michael Collison <michael.collison@arm.com>
13971 * config/arm/arm.c (arm_option_override): Deprecate
13972 use of -mstructure-size-boundary.
13973 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
13974 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
13976 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13979 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13980 Reduce cost estimate for direct moves.
13982 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
13985 * config/i386/i386.c (ix86_option_override_internal): Disable
13986 BMI, BMI2 and TBM instructions for -m16.
13988 2017-07-21 Carl Love <cel@us.ibm.com>
13990 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13991 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13992 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13993 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13994 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13995 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13996 VMULOSW): New enum "unspec" values.
13997 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13998 altivec_vmulosw): New patterns.
13999 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14000 VMULOSW): Add definitions.
14002 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
14004 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
14005 (qdf24xx): Likewise.
14006 * config/aarch64/aarch64-options-extensions.def (rdma); New.
14007 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
14008 (AARCH64_FL_V8_1): Renumber.
14009 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
14010 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
14011 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
14012 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
14013 rdma to feature modifiers list.
14015 2017-07-21 Yury Gribov <tetra2005@gmail.com>
14017 PR middle-end/56727
14018 * ipa-visibility (function_and_variable_visibility): Convert
14019 recursive PLT call to direct call if appropriate.
14021 2017-07-21 Andrew Pinski <apinski@cavium.com>
14023 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
14024 operand 1 to see if the types precision matches.
14025 * fold-const.c (operand_equal_p): Likewise.
14027 2017-07-21 Richard Biener <rguenther@suse.de>
14029 PR tree-optimization/81303
14030 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
14031 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
14032 (vect_peeling_hash_get_lowest_cost): Adjust.
14033 (vect_enhance_data_refs_alignment): Likewise. Use
14034 vect_get_peeling_costs_all_drs to compute the penalty for no
14035 peeling to match up costs.
14037 2017-07-21 Richard Biener <rguenther@suse.de>
14039 PR tree-optimization/81500
14040 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
14041 we didn't identify a reduction path.
14043 2017-07-21 Tom de Vries <tom@codesourcery.com>
14044 Cesar Philippidis <cesar@codesourcery.com>
14046 PR gcov-profile/81442
14047 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
14050 2017-07-21 Tom de Vries <tom@codesourcery.com>
14053 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
14055 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
14056 nvptx_override_options_after_change.
14058 2017-07-21 Ulrich Drepper <drepper@redhat.com>
14060 * dwarf2out.c (output_file_names): Avoid double testing for
14061 dwarf_version >= 5.
14063 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
14065 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
14067 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14069 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
14071 (try_crossjump_to_edge): Do not punt on partitioned functions.
14073 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14075 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
14076 Put all BBs reachable only via paths crossing cold region to cold
14078 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
14080 2016-07-21 Richard Biener <rguenther@suse.de>
14082 PR tree-optimization/81303
14083 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
14084 into account prologue and epilogue iterations when raising
14085 min_profitable_iters to sth at least covering one vector iteration.
14087 2017-07-21 Tamar Christina <tamar.christina@arm.com>
14089 * config/arm/arm.c (arm_test_cpu_arch_dat):
14092 2017-07-20 Nathan Sidwell <nathan@acm.org>
14094 Remove TYPE_METHODS.
14095 * tree.h (TYPE_METHODS): Delete.
14096 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
14097 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
14098 (dbxout_type_methods): Scan TYPE_FIELDS.
14099 (dbxout_type): Don't check TYPE_METHODS here.
14100 * function.c (use_register_for_decl): Always ignore register for
14101 class types when not optimizing.
14102 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
14103 * tree.c (free_lang_data_in_type): Stitch out member functions and
14104 templates from TYPE_FIELDS.
14105 (build_distinct_type_copy, verify_type_variant,
14106 verify_type): Member fns are on TYPE_FIELDS.
14107 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
14108 * tree-pretty-print.c (dump_generic_node): Likewise.
14110 2017-07-20 Jakub Jelinek <jakub@redhat.com>
14113 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
14114 V2TImode and V4TImode.
14115 (ix86_expand_vector_extract): Likewise.
14116 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
14117 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
14118 (ssescalarmode): Handle V4TImode and V2TImode.
14119 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
14120 (*vec_extractv2ti, *vec_extractv4ti): New insns.
14121 (VEXTRACTI128_MODE): New mode iterator.
14122 (splitter for *vec_extractv?ti first element): New.
14123 (VEC_INIT_MODE): New mode iterator.
14124 (vec_init<mode>): Consolidate 3 expanders into one using
14125 VEC_INIT_MODE mode iterator.
14127 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14129 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
14130 non_spilled_static_chain_regno_p.
14132 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14134 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
14136 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
14138 * bb-reorder.c (connect_traces): Allow copying of blocks within
14141 2017-07-20 Richard Biener <rguenther@suse.de>
14143 * gimple.h (gimple_phi_result): Add gphi * overload.
14144 (gimple_phi_result_ptr): Likewise.
14145 (gimple_phi_arg): Likewise. Adjust index assert to only
14146 allow actual argument accesses rather than all slots available
14148 (gimple_phi_arg_def): Add gphi * overload.
14149 * tree-phinodes.c (make_phi_node): Initialize only actual
14151 (resize_phi_node): Clear memory not covered by old node,
14152 do not initialize excess argument slots.
14153 (reserve_phi_args_for_new_edge): Initialize new argument slot
14156 2017-07-20 Bin Cheng <bin.cheng@arm.com>
14158 PR tree-optimization/81388
14160 2016-07-21 Bin Cheng <bin.cheng@arm.com>
14162 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
14163 by removing computation of may_be_zero.
14165 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14166 Tom de Vries <tom@codesourcery.com>
14168 PR middle-end/81030
14169 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
14170 when gimple level profile disagrees with what RTL expander did.
14172 2017-07-20 Richard Biener <rguenther@suse.de>
14174 PR tree-optimization/61171
14175 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
14176 (vect_analyze_stmt): Add slp instance parameter.
14177 (vectorizable_reduction): Likewise.
14178 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
14179 (vect_is_simple_reduction): Deal with chains not detected
14180 as SLP reduction chain, specifically not properly associated
14181 chains containing a mix of plus/minus.
14182 (get_reduction_op): Remove.
14183 (get_initial_defs_for_reduction): Simplify, pass in whether
14184 this is a reduction chain, pass in the SLP node for the PHIs.
14185 (vect_create_epilog_for_reduction): Get the SLP instance as
14187 (vectorizable_reduction): Get the SLP instance as arg.
14188 During analysis remember the SLP node with the PHIs in the
14189 instance. Simplify getting at the vectorized reduction PHIs.
14190 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
14191 through SLP instance.
14192 (vect_slp_analyze_operations): Likewise.
14193 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
14194 (vect_transform_stmt): Likewise.
14196 2017-07-20 Tom de Vries <tom@codesourcery.com>
14198 PR tree-optimization/81489
14199 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
14200 read of phi arg location to before loop that modifies phi.
14202 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14204 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
14207 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14209 PR middle-end/81331
14210 * except.c (execute): Fix ordering issue.
14212 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14214 PR rtl-optimization/81423
14215 * combine.c (make_compound_operation_int): Don't try to optimize
14216 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
14218 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14220 PR rtl-optimization/81423
14221 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
14222 with a constant that is -1 in the truncated to mode.
14224 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14226 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
14227 (determine_unlikely_bbs): ... here.
14228 * predict.h (propagate_unlikely_bbs_forward): Declare.
14229 * cfgexpand.c (pass_expand::execute): Use it.
14230 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
14232 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
14233 propagate_unlikely_bbs_forward.
14235 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14237 PR middle-end/81331
14238 * except.c (maybe_add_nop_after_section_switch): New function.
14241 2017-07-19 Tom de Vries <tom@codesourcery.com>
14243 * gimple.h (gimple_phi_set_arg): Make assert more strict.
14245 2017-07-19 Tom de Vries <tom@codesourcery.com>
14247 * gimple.h (gimple_phi_arg): Make assert more strict.
14249 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
14251 * config.gcc (powerpc*-*-*): Add mmintrin.h.
14252 * config/rs6000/mmintrin.h: New file.
14253 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
14255 2017-07-19 Jakub Jelinek <jakub@redhat.com>
14257 PR tree-optimization/81346
14258 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
14260 2017-07-19 Tom de Vries <tom@codesourcery.com>
14262 * config/nvptx/nvptx.md (VECIM): Add V2DI.
14264 2017-07-19 Tom de Vries <tom@codesourcery.com>
14266 * config/nvptx/nvptx-modes.def: Add V2DImode.
14267 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
14268 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
14269 (nvptx_output_mov_insn): Handle lack of mov.b128.
14270 (nvptx_print_operand): Handle 'H' and 'L' codes.
14271 (nvptx_vector_mode_supported): Allow V2DImode.
14272 (nvptx_preferred_simd_mode): New function.
14273 (nvptx_data_alignment): New function.
14274 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
14275 nvptx_preferred_simd_mode.
14276 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
14278 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
14280 2017-07-19 Tom de Vries <tom@codesourcery.com>
14282 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
14283 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
14284 (nvptx_vector_mode_supported): New function. Allow V2SImode.
14285 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
14286 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
14287 (mov<VECIM>_insn): New define_insn.
14288 (define_expand "mov<VECIM>): New define_expand.
14290 2017-07-19 Tom de Vries <tom@codesourcery.com>
14292 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
14294 2017-07-19 Jakub Jelinek <jakub@redhat.com>
14296 PR tree-optimization/81346
14297 * fold-const.h (fold_div_compare, range_check_type): Declare.
14298 * fold-const.c (range_check_type): New function.
14299 (build_range_check): Use range_check_type.
14300 (fold_div_compare): No longer static, rewritten into
14301 a match.pd helper function.
14302 (fold_comparison): Don't call fold_div_compare here.
14303 * match.pd (X / C1 op C2): New optimization using fold_div_compare
14304 as helper function.
14306 2017-07-19 Nathan Sidwell <nathan@acm.org>
14308 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
14309 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
14310 * tree.c (find_decls_types_r, verify_type): Use
14311 TYPE_{MIN,MAX}_VALUE_RAW.
14312 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
14313 (hash_tree): Likewise.
14314 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14316 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14319 2017-07-18 Tom de Vries <tom@codesourcery.com>
14321 PR middle-end/81464
14322 * omp-expand.c (expand_omp_for_static_chunk): Handle
14323 equal-argument loop exit phi.
14325 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
14328 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
14329 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
14330 operand 2 predicate.
14331 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
14332 operand 2 predicate.
14333 (ror,rol -> rorx splitters): Use const_int_operand as
14334 operand 2 predicate.
14336 2017-06-18 Richard Biener <rguenther@suse.de>
14338 PR tree-optimization/81410
14339 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
14340 the gap in the ! slp_perm SLP case after each group.
14342 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14344 PR middle-end/81463
14345 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
14348 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14350 PR middle-end/81462
14351 * predict.c (set_even_probabilities): Cleanup; do not affect
14352 probabilities that are already known.
14353 (combine_predictions_for_bb): Call even when count is set.
14355 2017-07-18 Nathan Sidwell <nathan@acm.org>
14357 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
14360 2017-07-18 Bin Cheng <bin.cheng@arm.com>
14363 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
14364 optimization for loop niter analysis.
14366 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
14369 * config/avr/avr.c (avr_optimize_casesi): Don't use
14370 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
14372 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
14374 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
14375 body_cost_vec from _vect_peel_extended_info.
14376 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
14377 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
14380 2017-07-18 Bin Cheng <bin.cheng@arm.com>
14382 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
14384 2017-07-18 Richard Biener <rguenther@suse.de>
14386 PR tree-optimization/80620
14387 PR tree-optimization/81403
14388 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
14389 info when re-using a VN table entry.
14391 2017-07-18 Richard Biener <rguenther@suse.de>
14393 PR tree-optimization/81418
14394 * tree-vect-loop.c (vectorizable_reduction): Properly compute
14395 vectype_in. Verify that with lane-reducing reduction operations
14396 we have a single def-use cycle.
14398 2017-07-17 Carl Love <cel@us.ibm.com>
14400 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
14402 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14403 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14404 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14405 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14406 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14407 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14408 VMULOSW): New enum "unspec" values.
14409 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14410 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14411 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14412 altivec_vmulosw): New patterns.
14413 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14414 VMULOSW): Add definitions.
14415 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
14417 * config/alpha/alpha.c: Include predict.h.
14419 2017-07-17 Yury Gribov <tetra2005@gmail.com>
14421 * tree-vrp.c (compare_assert_loc): Fix comparison function
14422 to return predictable results.
14424 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14426 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
14428 (subdi3): Likewise.
14429 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
14430 * doc/invoke.texi (mexpand-adddi): Update text.
14432 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14434 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
14435 that also clobbers the CC register. The old expand code is moved
14437 (*arc_clzsi2): ... here.
14438 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
14439 the CC register. The old expand code is moved to ...
14440 (arc_ctzsi2): ... here.
14442 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14444 * config/arc/arc.opt (mindexed-loads): Use initial value
14445 TARGET_INDEXED_LOADS_DEFAULT.
14446 (mauto-modify-reg): Use initial value
14447 TARGET_AUTO_MODIFY_REG_DEFAULT.
14448 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
14449 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
14450 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
14451 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
14453 2017-07-17 Martin Liska <mliska@suse.cz>
14456 * opts.c (finish_options): Do not allow -fgnu-tm
14457 w/ -fsanitize={kernel-,}address. Say sorry.
14459 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14462 * tree-loop-distribution.c (classify_partition): Only assert on
14463 numer of iterations.
14464 (merge_dep_scc_partitions): Delete prameter. Update function call.
14465 (distribute_loop): Remove code handling loop with unknown niters.
14466 (pass_loop_distribution::execute): Skip loop with unknown niters.
14468 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14471 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
14472 function sort_partitions_by_post_order.
14474 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14476 PR tree-optimization/81374
14477 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
14478 the max index of basic blocks, rather than number of basic blocks.
14480 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14482 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
14484 (arc_legitimate_pic_operand_p): Likewise.
14485 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
14487 (arc_needs_pcl_p): Likewise.
14488 (arc_legitimate_pc_offset_p): Likewise.
14489 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
14490 function is also used in constrains.md.
14491 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
14492 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
14493 PLUS. Only return true/false in known cases, otherwise assert.
14494 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
14495 is already called in arc_legitimate_constant_p.
14496 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
14498 (LEGITIMATE_PIC_OPERAND_P): Use
14499 arc_raw_symbolic_reference_mentioned_p function.
14500 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
14505 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14506 Andrew Burgess <andrew.burgess@embecosm.com>
14508 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
14509 (arc_return_address_register): New function.
14510 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
14511 (arc_handle_fndecl_attribute): Add naked attribute.
14512 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14513 (TARGET_WARN_FUNC_RETURN): Likewise.
14514 (arc_allocate_stack_slots_for_args): New function.
14515 (arc_warn_func_return): Likewise.
14516 (machine_function): Change type fn_type.
14517 (arc_compute_function_type): Consider new naked function type,
14518 change function return type.
14519 (arc_must_save_register): Adapt to handle new
14520 arc_compute_function_type's return type.
14521 (arc_expand_prologue): Likewise.
14522 (arc_expand_epilogue): Likewise.
14523 (arc_return_address_regs): Delete.
14524 (arc_return_address_register): New function.
14525 (arc_epilogue_uses): Use above function.
14526 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
14527 (arc_function_type): Change encoding, add naked type.
14528 (ARC_INTERRUPT_P): Change to handle the new encoding.
14529 (ARC_FAST_INTERRUPT_P): Likewise.
14530 (ARC_NORMAL_P): Define.
14531 (ARC_NAKED_P): Likewise.
14532 (arc_compute_function_type): Delete prototype.
14533 * config/arc/arc.md (in_ret_delay_slot): Use
14534 arc_return_address_register function.
14535 (simple_return): Likewise.
14536 (p_return_i): Likewise.
14538 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14540 PR tree-optimization/81428
14541 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
14542 can't be built for those types.
14544 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14546 Remove stuff dead since r239246.
14548 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
14549 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
14550 (avr_inform_devices): Remove dead stuff.
14552 2017-07-17 Tamar Christina <tamar.christina@arm.com>
14554 * config/arm/arm_neon.h: Fix softp typo.
14556 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14558 PR tree-optimization/81365
14559 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
14560 aggregate moves onto bb predecessor edges, make sure there are no
14561 loads that could alias the lhs in between the start of bb and the
14564 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14567 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
14568 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
14569 [LSHIFTRT, outer_code = TRUNCATE]: Same.
14571 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14573 PR tree-optimization/81396
14574 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
14575 (init_symbolic_number): Initialize it to 1.
14576 (perform_symbolic_merge): Add n_ops from both operands into the new
14578 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
14579 without base_addr as useless if they need more than one operation.
14580 (bswap_replace): Handle !bswap case for NULL base_addr.
14582 2017-07-17 Tom de Vries <tom@codesourcery.com>
14585 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
14588 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
14590 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
14591 conditional builtin define __FIX_LEON3FT_B2BST.
14593 2017-07-17 Daniel Cederman <cederman@gaisler.com>
14595 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
14596 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
14599 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14601 PR rtl-optimization/81424
14602 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
14603 to remove potential trapping from operands if -fnon-call-exceptions.
14605 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14607 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
14608 profile_proability for scalling.
14609 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
14611 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14613 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
14615 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14617 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
14618 fixpoint arithmetics.
14620 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14622 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
14623 fixpoint arithmetics.
14625 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14627 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
14628 fixpoint arithmetics.
14630 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14632 * profile-count.h (profile_probability::from_reg_br_prob_note,
14633 profile_probability::to_reg_br_prob_note): New functions.
14634 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
14635 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
14636 * predict.c (probability_reliable_p): Update.
14637 (edge_probability_reliable_p): Update.
14638 (br_prob_note_reliable_p): Update.
14639 (invert_br_probabilities): Update.
14640 (add_reg_br_prob_note): New function.
14641 (combine_predictions_for_insn): Update.
14642 * asan.c (asan_clear_shadow): Update.
14643 * cfgbuild.c (compute_outgoing_frequencies): Update.
14644 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14645 (update_br_prob_note): Update.
14646 (rtl_verify_edges): Update.
14647 (purge_dead_edges): Update.
14648 (fixup_reorder_chain): Update.
14649 * emit-rtl.c (try_split): Update.
14650 * ifcvt.c (cond_exec_process_insns): Update.
14651 (cond_exec_process_if_block): Update.
14652 (dead_or_predicable): Update.
14653 * internal-fn.c (expand_addsub_overflow): Update.
14654 (expand_neg_overflow): Update.
14655 (expand_mul_overflow): Update.
14656 * loop-doloop.c (doloop_modify): Update.
14657 * loop-unroll.c (compare_and_jump_seq): Update.
14658 * optabs.c (emit_cmp_and_jump_insn_1): Update.
14659 * predict.h: Update.
14660 * reorg.c (mostly_true_jump): Update.
14662 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
14663 * config/alpha/alpha.c (emit_unlikely_jump): Update.
14664 * config/arc/arc.c: (emit_unlikely_jump): Update.
14665 * config/arm/arm.c: (emit_unlikely_jump): Update.
14666 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
14667 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
14668 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
14669 (ix86_print_operand): Update.
14670 (ix86_split_fp_branch): Update.
14671 (predict_jump): Update.
14672 * config/ia64/ia64.c (ia64_print_operand): Update.
14673 * config/mmix/mmix.c (mmix_print_operand): Update.
14674 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
14675 (rs6000_expand_split_stack_prologue): Update.
14676 * config/rs6000/rs6000.c: Update.
14677 * config/s390/s390.c (s390_expand_vec_strlen): Update.
14678 (s390_expand_vec_movstr): Update.
14679 (s390_expand_cs_tdsi): Update.
14680 (s390_expand_split_stack_prologue): Update.
14681 * config/sh/sh.c (sh_print_operand): Update.
14682 (expand_cbranchsi4): Update.
14683 (expand_cbranchdi4): Update.
14684 * config/sparc/sparc.c (output_v9branch): Update.
14685 * config/spu/spu.c (get_branch_target): Update.
14686 (ea_load_store_inline): Update.
14687 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
14688 * config/tilepro/tilepro.c: Update.
14690 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14692 * gimplify.c (mostly_copy_tree_r): Revert latest change.
14693 (gimplify_save_expr): Likewise.
14695 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14697 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
14699 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14701 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
14703 * timevar.def (TV_IPA_FNSUMMARY): Define.
14705 2017-07-16 Daniel Cederman <cederman@gaisler.com>
14707 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
14708 to back store errata sensitive sequence from being generated.
14709 (sqrtdf2_fix): Likewise.
14711 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14713 * tree-ssa-threadupdate.c (compute_path_counts,
14714 update_joiner_offpath_counts): Use profile_probability.
14716 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
14719 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14721 * config/arm/arm-c.c (arm_cpu_builtins): Define
14722 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
14724 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
14726 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14727 array entries to represent __ieee128 versions of the
14728 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
14729 scalar_extract_sig, and scalar_insert_exp built-in functions.
14730 (altivec_resolve_overloaded_builtin): Add special case handling
14731 for the __builtin_scalar_insert_exp function, as represented by
14732 the P9V_BUILTIN_VEC_VSIEDP constant.
14733 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
14734 exponent support for __ieee128 argument.
14735 (VSESQP): Add scalar extract signature support for __ieee128
14737 (VSTDCNQP): Add scalar test negative support for __ieee128
14739 (VSIEQP): Add scalar insert exponent support for __int128 argument
14740 with __ieee128 result.
14741 (VSIEQPF): Add scalar insert exponent support for __ieee128
14742 argument with __ieee128 result.
14743 (VSTDCQP): Add scalar test data class support for __ieee128
14745 (VSTDCNQP): Add overload support for scalar test negative with
14746 __ieee128 argument.
14747 (VSTDCQP): Add overload support for scalar test data class
14748 __ieee128 argument.
14749 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
14750 UNSPEC_VSX_SXSIGDP.
14751 (UNSPEC_VSX_SIEXPQP): New constant.
14752 (xsxexpqp): New insn for VSX scalar extract exponent quad
14754 (xsxsigqp): New insn for VSX scalar extract significand quad
14756 (xsiexpqpf): New insn for VSX scalar insert exponent quad
14757 precision with floating point argument.
14758 (xststdcqp): New expand for VSX scalar test data class quad
14760 (xststdcnegqp): New expand for VSX scalar test negative quad
14762 (xststdcqp): New insn to match expansions for VSX scalar test data
14763 class quad precision and VSX scalar test negative quad precision.
14764 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
14765 special case operand checking to enforce that second operand of
14766 VSX scalar test data class with quad precision argument is a 7-bit
14768 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
14769 prototypes and descriptions of __ieee128 versions of
14770 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
14771 scalar_test_data_class, and scalar_test_neg built-in functions.
14773 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14775 PR tree-optimization/81162
14776 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
14777 replace a negate with an add.
14779 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
14781 * doc/invoke.texi (arm/-mcpu): Document +crypto.
14783 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14785 * config/arm/arm-c.c (arm_cpu_builtins): Define
14786 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
14788 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14790 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
14791 (armv8-r): Set ARM Cortex-R52 as default CPU.
14792 * config/arm/arm-tables.opt: Regenerate.
14793 * config/arm/arm-tune.md: Regenerate.
14794 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14796 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
14797 extension for -mcpu=cortex-r52.
14799 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14801 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
14802 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
14803 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
14804 (fp-armv8): Define it as FP_ARMv8 only.
14805 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
14806 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
14808 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
14809 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
14810 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
14811 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
14812 than TARGET_FPU_ARMV8.
14813 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
14814 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
14815 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
14817 * config/arm/neon.md (neon_vrint): Likewise.
14818 (neon_vcvt): Likewise.
14819 (neon_<fmaxmin_op><mode>): Likewise.
14820 (<fmaxmin><mode>3): Likewise.
14821 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
14822 * config/arm/predicates.md (arm_cond_move_operator): Check against
14823 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
14825 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
14827 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
14828 to top of function.
14830 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14832 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
14833 loop in comment with memset.
14835 2017-07-14 Martin Liska <mliska@suse.cz>
14837 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
14838 * dwarf2out.c (is_java): Remove the function.
14839 (output_pubname): Remove usage of the function.
14840 (lower_bound_default): Remove usage of DW_LANG_Java.
14841 (gen_compile_unit_die): Likewise.
14842 * gcc.c: Remove compiler defaults for .java and .zip files.
14843 * gimple-expr.c (remove_suffix): Change as there's no longer
14844 extension than 4-letter one.
14845 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
14846 (gimplify_save_expr): Likewise.
14847 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
14848 as it's possible even for other languages than Java.
14849 * langhooks.h (struct lang_hooks): Remove Java from a comment.
14850 * lto-opts.c (lto_write_options): Remove reference to Java.
14851 * opts.c (strip_off_ending): Update file extension handling.
14852 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
14853 * tree-eh.c (lower_resx): Likewise.
14854 * tree.c (free_lang_data_in_type): Remove dead code.
14855 (find_decls_types_r): Likewise.
14856 (build_common_builtin_nodes): Remove Java from a comment.
14857 (verify_type): Remove dead code.
14858 * varasm.c (assemble_external): Remove Java from a comment.
14860 2017-07-14 Martin Liska <mliska@suse.cz>
14862 * opts.c (finish_options): Add quotes.
14863 (common_handle_option): Likewise.
14865 2017-07-14 Martin Liska <mliska@suse.cz>
14867 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
14868 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
14869 Remove N_SO_PASCAL.
14870 * dwarf2out.c (lower_bound_default): Do not handle
14872 (gen_compile_unit_die): Likewise.
14873 * gcc.c: Remove default extension binding for GNU Pascal.
14874 * stmt.c: Remove Pascal language from a comment.
14875 * xcoffout.c: Likewise.
14877 2017-07-13 David Malcolm <dmalcolm@redhat.com>
14880 * diagnostic-show-locus.c (fixit_cmp): New function.
14881 (layout::layout): Sort m_fixit_hints.
14882 (column_range::column_range): Assert that the values are valid.
14883 (struct char_span): New struct.
14884 (correction::overwrite): New method.
14885 (struct source_line): New struct.
14886 (line_corrections::add_hint): Add assertions. Reimplement memcpy
14887 calls in terms of classes source_line and char_span, and
14888 correction::overwrite.
14889 (selftest::test_overlapped_fixit_printing_2): New function.
14890 (selftest::diagnostic_show_locus_c_tests): Call it.
14892 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
14894 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
14895 early if there is no lhs.
14897 2017-07-13 Martin Liska <mliska@suse.cz>
14899 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
14900 (gen_reference_type_die): Likewise.
14901 * stor-layout.c: Remove Pascal-related comment.
14903 2017-07-13 Martin Liska <mliska@suse.cz>
14905 * opts.c (finish_options): Add quotes to error messages.
14906 (parse_sanitizer_options): Likewise.
14908 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
14910 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
14912 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
14914 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
14916 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
14918 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
14920 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
14922 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
14925 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
14926 provides the hardware capability bits, define the macro
14927 __BUILTIN_CPU_SUPPORTS__.
14928 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
14929 if GLIBC does not provide the hardware capability bits. Add a
14930 gcc_unreachable call if the built-in cpu function is neither
14931 __builtin_cpu_is nor __builtin_cpu_supports.
14932 (rs6000_get_function_versions_dispatcher): Change the warning
14933 that an old GLIBC is used which does not export the capability
14934 bits to be an error.
14935 * doc/extend.texi (target_clones attribute): Document the
14936 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
14937 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
14938 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
14939 the macros defined by GCC if the newer GLIBC is available.
14941 2017-07-12 Jeff Law <law@redhat.com>
14943 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
14944 remaining includes slightly.
14945 * config/riscv/riscv-builtins.c: Include profile-count.h.
14947 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14950 * config/avr/avr.c (avr_set_current_function): In diagnostic
14951 messages: Quote keywords and (parts of) identifiers.
14952 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
14955 2017-07-12 Carl Love <cel@us.ibm.com>
14957 * config/rs6000/rs6000-c.c: Add support for built-in functions
14958 vector bool char vec_revb (vector bool char);
14959 vector bool short vec_revb (vector short char);
14960 vector bool int vec_revb (vector bool int);
14961 vector bool long long vec_revb (vector bool long long);
14962 * doc/extend.texi: Update the built-in documentation file for the
14963 new built-in functions.
14965 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14967 * config/s390/s390.md: Remove movcc splitter.
14969 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14971 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
14972 load/store on condition.
14974 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14977 * config/avr/avr.c (avr_encode_section_info)
14978 [progmem && !TREE_READONLY]: Error if progmem object needs
14981 2017-07-11 Michael Collison <michael.collison@arm.com>
14983 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
14986 2017-07-11 Carl Love <cel@us.ibm.com>
14988 * config/rs6000/rs6000-c.c: Add support for builtins
14989 vector unsigned int vec_parity_lsbb (vector signed int);
14990 vector unsigned int vec_parity_lsbb (vector unsigned int);
14991 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
14992 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
14993 vector unsigned long long vec_parity_lsbb (vector signed long long);
14994 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
14995 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
14996 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
14997 * doc/extend.texi: Update the built-in documentation file for the
14998 new built-in functions.
15000 2017-07-11 David Malcolm <dmalcolm@redhat.com>
15002 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
15003 (layout::m_primary_loc): New field.
15004 (layout::layout): Initialize new field. Move location filtering
15005 logic from here to...
15006 (layout::maybe_add_location_range): ...this new method. Add
15007 support for filtering to just the lines already specified by other
15009 (layout::will_show_line_p): New method.
15010 (gcc_rich_location::add_location_if_nearby): New method.
15011 (selftest::test_add_location_if_nearby): New test function.
15012 (selftest::diagnostic_show_locus_c_tests): Call it.
15013 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
15016 2017-07-11 Tom de Vries <tom@codesourcery.com>
15018 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
15019 (bb_first_real_insn): New function.
15020 (nvptx_single): Add extra initialization of broadcasted condition
15023 2017-07-11 Nathan Sidwell <nathan@acm.org>
15025 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
15027 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
15029 * doc/extend.texi (AVR Function Attributes): Remove weblink to
15030 Binutils doc as TEXI will mess them up.
15031 * doc/invoke.texi (AVR Options): Same here.
15033 2017-07-11 Daniel Cederman <cederman@gaisler.com>
15035 * config/sparc/sparc.opt (mfix-ut700): New option.
15036 (mfix-gr712rc): Likewise.
15037 (sparc_fix_b2bst): New variable.
15038 * doc/invoke.texi (SPARC options): Document them.
15039 (ARM options): Fix warnings.
15040 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
15041 instructions to prevent sequences that can trigger the store-store
15042 errata for certain LEON3FT processors.
15043 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
15044 (sparc_option_override): Set sparc_fix_b2bst appropriately.
15045 * config/sparc/sparc.md (fix_b2bst): New attribute.
15046 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
15048 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
15051 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
15053 (*rsqrtsf2_sse): Ditto.
15055 (div<mode>3): Macroize insn from divdf3 and divsf3
15056 using MODEF mode iterator.
15058 2017-07-10 Martin Sebor <msebor@redhat.com>
15060 PR tree-optimization/80397
15061 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
15062 instead of testing for equality to INTEGER_TYPE.
15064 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
15066 * config.gcc: Remove uclibc from arc target spec.
15068 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
15070 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
15072 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15075 * lto-wrapper.c (remove_option): New function.
15076 (merge_and_complain): Merge PIC/PIE options more realistically.
15078 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15080 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
15084 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
15085 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
15086 * config.in: Regenerate.
15087 * configure: Regenerate.
15088 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
15089 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
15090 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
15091 (TARGET_GASISR_PROLOGUES): ...target mask.
15092 * common/config/avr/avr-common.c
15093 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
15094 Set -mgas-isr-prologues.
15095 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
15096 INSERT_PASS_BEFORE for it.
15097 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
15098 * config/avr/avr.c (avr_option_override)
15099 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
15100 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
15101 (avr_attribute_table) <no_gccisr>: Add new function attribute.
15102 (avr_set_current_function) <is_no_gccisr>: Init machine field.
15103 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
15105 (make_avr_pass_pre_proep): New function.
15106 (emit_push_sfr) <treg>: Add argument to function and use it
15107 instead of TMP_REG.
15108 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
15109 and set machine->gasisr.yes.
15110 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
15111 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
15112 __gcc_isr.n_pushed to .L__stack_usage.
15113 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
15114 (avr_asm_final_postscan_insn): ...this new static function.
15115 * config/avr/avr.h (machine_function)
15116 <is_no_gccisr, use_L__stack_usage>: New fields.
15117 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
15118 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
15119 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
15120 (gasisr, *gasisr): New expander and insn.
15121 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15122 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
15123 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
15125 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
15127 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
15130 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15132 Move jump-tables out of .text again.
15135 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
15136 (ASM_OUTPUT_ADDR_VEC): New function.
15137 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
15138 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
15139 INSN_ADDRESSes as asm comment.
15140 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
15141 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
15142 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
15143 * config/avr/avr.md (*tablejump): Adjust comment.
15144 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
15145 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
15147 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
15148 (avr_output_addr_vec): New proto.
15149 (avr_log_t) <insn_addresses>: New field.
15151 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
15154 * config/i386/i386.c (ix86_function_arg_advance): Set
15155 outgoing_args_on_stack to true if there are outgoing arguments
15157 (ix86_function_arg): Likewise.
15158 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
15159 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
15160 * config/i386/i386.h (machine_function): Add
15161 outgoing_args_on_stack.
15163 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
15165 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15166 supporting pthreds.
15167 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
15169 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
15171 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
15172 (REAL_H): Remove $(MACHMODE_H).
15173 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
15175 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
15176 $(MACHMODE_H) and double-int.h.
15177 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
15179 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
15182 2017-07-07 Andrew Pinski <apinski@cavium.com>
15184 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
15185 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
15187 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15189 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
15190 Add warning if GCC was not configured to link against a GLIBC that
15191 exports the hardware capability bits.
15192 (make_resolver_func): Make resolver function private and not a
15193 COMDAT function. Create the name with clone_function_name instead
15194 of make_unique_name.
15197 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
15198 correct operand in doing the split.
15200 2017-07-07 Carl Love <cel@us.ibm.com>
15202 * config/rs6000/rs6000-c: Add support for built-in function
15203 vector unsigned short vec_pack_to_short_fp32 (vector float,
15205 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
15206 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
15207 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
15208 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
15209 (convert_4f32_8i16): Add define_expand.
15210 * doc/extend.texi: Update the built-in documentation file for the
15211 new built-in function.
15213 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15215 * config/sparc/m8.md: New file.
15216 * config/sparc/sparc.md: Include m8.md.
15218 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15220 * config/sparc/sparc.opt: New option -mvis4b.
15221 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
15222 (sparc_option_override): Handle VIS4B.
15223 (enum sparc_builtins): Define
15224 SPARC_BUILTIN_DICTUNPACK{8,16,32},
15225 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
15226 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
15227 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
15228 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
15229 (check_constant_argument): New function.
15230 (sparc_vis_init_builtins): Define builtins
15231 __builtin_vis_dictunpack{8,16,32},
15232 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
15233 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
15234 __builtin_vis_fpcmpde{8,16,32}shl and
15235 __builtin_vis_fpcmpur{8,16,32}shl.
15236 (sparc_expand_builtin): Check that the constant operands to
15237 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
15238 constant and in range.
15239 * config/sparc/sparc-c.c (sparc_target_macros): Handle
15241 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
15242 (SPARC_IMM5_P): Likewise.
15243 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
15244 (enabled): Handle vis4b.
15245 (UNSPEC_DICTUNPACK): New unspec.
15246 (UNSPEC_FPCMPSHL): Likewise.
15247 (UNSPEC_FPUCMPSHL): Likewise.
15248 (UNSPEC_FPCMPDESHL): Likewise.
15249 (UNSPEC_FPCMPURSHL): Likewise.
15250 (cpu_feature): New CPU feature `vis4b'.
15251 (dictunpack{8,16,32}): New insns.
15252 (FPCSMODE): New mode iterator.
15253 (fpcscond): New code iterator.
15254 (fpcsucond): Likewise.
15255 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
15256 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
15257 (fpcmpde{8,16,32}{si,di}shl): Likewise.
15258 (fpcmpur{8,16,32}{si,di}shl): Likewise.
15259 * config/sparc/constraints.md: Define constraints `q' for unsigned
15260 2-bit integer constants and `t' for unsigned 5-bit integer
15262 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
15264 (imm5_operand_dictunpack16): Likewise.
15265 (imm5_operand_dictunpack32): Likewise.
15266 (imm2_operand): Likewise.
15267 * doc/invoke.texi (SPARC Options): Document -mvis4b.
15268 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
15269 ditunpack* and fpcmp*shl builtins.
15271 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15273 * config.gcc: Handle m8 in --with-{cpu,tune} options.
15274 * config.in: Add HAVE_AS_SPARC6 define.
15275 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
15277 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
15279 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
15280 (CPP_CPU_SPEC): Handle m8.
15281 (ASM_CPU_SPEC): Likewise.
15282 * config/sparc/sparc-opts.h (enum processor_type): Add
15284 * config/sparc/sparc.c (m8_costs): New struct.
15285 (sparc_option_override): Handle TARGET_CPU_m8.
15286 (sparc32_initialize_trampoline): Likewise.
15287 (sparc64_initialize_trampoline): Likewise.
15288 (sparc_issue_rate): Likewise.
15289 (sparc_register_move_cost): Likewise.
15290 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
15291 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
15292 (ASM_CPU64_DEFAULT_SPEC): Likewise.
15293 (CPP_CPU_SPEC): Handle M8.
15294 (ASM_CPU_SPEC): Likewise.
15295 (AS_M8_FLAG): Define.
15296 * config/sparc/sparc.md: Add m8 to the cpu attribute.
15297 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
15298 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
15300 * configure: Regenerate.
15301 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
15304 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15306 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
15308 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
15309 ("*movdi_insn_sp32"): Do not set v3pipe.
15310 ("*movsi_insn"): Likewise.
15311 ("*movdi_insn_sp64"): Likewise.
15312 ("*movsf_insn"): Likewise.
15313 ("*movdf_insn_sp32"): Likewise.
15314 ("*movdf_insn_sp64"): Likewise.
15315 ("*zero_extendsidi2_insn_sp64"): Likewise.
15316 ("*sign_extendsidi2_insn"): Likewise.
15317 ("*mov<VM32:mode>_insn"): Likewise.
15318 ("*mov<VM64:mode>_insn_sp64"): Likewise.
15319 ("*mov<VM64:mode>_insn_sp32"): Likewise.
15320 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
15321 ("<vlop:code><VL:mode>3"): Likewise.
15322 ("*not_<vlop:code><VL:mode>3"): Likewise.
15323 ("*nand<VL:mode>_vis"): Likewise.
15324 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
15325 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
15326 ("one_cmpl<VL:mode>2"): Likewise.
15327 ("faligndata<VM64:mode>_vis"): Likewise.
15328 ("alignaddrsi_vis"): Likewise.
15329 ("alignaddrdi_vis"): Likweise.
15330 ("alignaddrlsi_vis"): Likewise.
15331 ("alignaddrldi_vis"): Likewise.
15332 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
15333 ("bmaskdi_vis"): Likewise.
15334 ("bmasksi_vis"): Likewise.
15335 ("bshuffle<VM64:mode>_vis"): Likewise.
15336 ("cmask8<P:mode>_vis"): Likewise.
15337 ("cmask16<P:mode>_vis"): Likewise.
15338 ("cmask32<P:mode>_vis"): Likewise.
15339 ("pdistn<P:mode>_vis"): Likewise.
15340 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
15342 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15344 * config/sparc/sparc.md ("subtype"): New insn attribute.
15345 ("*wrgsr_sp64"): Set insn subtype.
15346 ("*rdgsr_sp64"): Likewise.
15347 ("alignaddrsi_vis"): Likewise.
15348 ("alignaddrdi_vis"): Likewise.
15349 ("alignaddrlsi_vis"): Likewise.
15350 ("alignaddrldi_vis"): Likewise.
15351 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
15352 ("fexpand_vis"): Likewise.
15353 ("fpmerge_vis"): Likewise.
15354 ("faligndata<VM64:mode>_vis"): Likewise.
15355 ("bshuffle<VM64:mode>_vis"): Likewise.
15356 ("cmask8<P:mode>_vis"): Likewise.
15357 ("cmask16<P:mode>_vis"): Likewise.
15358 ("cmask32<P:mode>_vis"): Likewise.
15359 ("fchksm16_vis"): Likewise.
15360 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
15361 ("fmean16_vis"): Likewise.
15362 ("fp<plusminus_insn>64_vis"): Likewise.
15363 ("<plusminus_insn>v8qi3"): Likewise.
15364 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
15365 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
15366 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
15367 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
15368 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
15369 ("*movqi_insn"): Likewise.
15370 ("*movhi_insn"): Likewise.
15371 ("*movsi_insn"): Likewise.
15372 ("movsi_pic_gotdata_op"): Likewise.
15373 ("*movdi_insn_sp32"): Likewise.
15374 ("*movdi_insn_sp64"): Likewise.
15375 ("movdi_pic_gotdata_op"): Likewise.
15376 ("*movsf_insn"): Likewise.
15377 ("*movdf_insn_sp32"): Likewise.
15378 ("*movdf_insn_sp64"): Likewise.
15379 ("*zero_extendhisi2_insn"): Likewise.
15380 ("*zero_extendqihi2_insn"): Likewise.
15381 ("*zero_extendqisi2_insn"): Likewise.
15382 ("*zero_extendqidi2_insn"): Likewise.
15383 ("*zero_extendhidi2_insn"): Likewise.
15384 ("*zero_extendsidi2_insn_sp64"): Likewise.
15385 ("ldfsr"): Likewise.
15386 ("prefetch_64"): Likewise.
15387 ("prefetch_32"): Likewise.
15388 ("tie_ld32"): Likewise.
15389 ("tie_ld64"): Likewise.
15390 ("*tldo_ldub_sp32"): Likewise.
15391 ("*tldo_ldub1_sp32"): Likewise.
15392 ("*tldo_ldub2_sp32"): Likewise.
15393 ("*tldo_ldub_sp64"): Likewise.
15394 ("*tldo_ldub1_sp64"): Likewise.
15395 ("*tldo_ldub2_sp64"): Likewise.
15396 ("*tldo_ldub3_sp64"): Likewise.
15397 ("*tldo_lduh_sp32"): Likewise.
15398 ("*tldo_lduh1_sp32"): Likewise.
15399 ("*tldo_lduh_sp64"): Likewise.
15400 ("*tldo_lduh1_sp64"): Likewise.
15401 ("*tldo_lduh2_sp64"): Likewise.
15402 ("*tldo_lduw_sp32"): Likewise.
15403 ("*tldo_lduw_sp64"): Likewise.
15404 ("*tldo_lduw1_sp64"): Likewise.
15405 ("*tldo_ldx_sp64"): Likewise.
15406 ("*mov<VM32:mode>_insn"): Likewise.
15407 ("*mov<VM64:mode>_insn_sp64"): Likewise.
15408 ("*mov<VM64:mode>_insn_sp32"): Likewise.
15410 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15412 * config/sparc/sparc.md ("type"): New insn type viscmp.
15413 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
15415 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
15416 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
15417 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
15418 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
15420 ("n7_vis_logical_11cycle"): Likewise.
15421 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
15422 * config/sparc/niagara2.md ("niag3_vis": Likewise.
15423 * config/sparc/niagara.md ("niag_vis"): Likewise.
15424 * config/sparc/ultra3.md ("us3_fga"): Likewise.
15425 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
15427 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15429 * config/sparc/sparc.md: New instruction type `bmask'.
15430 (bmaskdi_vis): Use the `bmask' type.
15431 (bmasksi_vis): Likewise.
15432 * config/sparc/ultra3.md (us3_array): Likewise.
15433 * config/sparc/niagara7.md (n7_array): Likewise.
15434 * config/sparc/niagara4.md (n4_array): Likewise.
15435 * config/sparc/niagara2.md (niag2_vis): Likewise.
15436 (niag3_vis): Likewise.
15437 * config/sparc/niagara.md (niag_vis): Likewise.
15439 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15441 * ipa-comdats.c: Remove optimize check from gate.
15442 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
15443 for functions not optimized.
15444 (ipa_fn_summary_read): Skip optimize check.
15445 (ipa_fn_summary_write): Likewise.
15446 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
15448 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
15450 (can_inline_edge_p): Check flag_pcc_struct_return for match.
15451 (check_callers): Give up on caller which is not optimized.
15452 (inline_small_functions): Likewise.
15453 (ipa_inline): Do not give up when not optimizing.
15454 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
15455 away unoptimizes cdtors.
15456 (whole_program_function_and_variable_visibility): Do
15457 ipa_discover_readonly_nonaddressable_vars in LTO mode.
15458 * ipa.c (process_references): Do not check optimize.
15459 (symbol_table::remove_unreachable_nodes): Update optimize check.
15460 (set_writeonly_bit): Update optimize check.
15461 (pass_ipa_cdtor_merge::gate): Do not check optimize.
15462 (pass_ipa_single_use::gate): Remove.
15464 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15466 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
15467 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
15468 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
15469 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
15470 permute_load, permute_store, adjust_extract, adjust_splat,
15471 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
15472 replace_swap_with_copy, dump_swap_insn_table,
15473 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
15474 recombine_lvx_pattern, recombine_stvx_pattern,
15475 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
15476 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
15477 to file rs6000-p8swap.c.
15478 * config/rs6000/rs6000-p8swap.c: New file.
15479 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
15480 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
15481 and rs6000*-*-* targets.
15483 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15485 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
15487 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15489 * lto-wrapper.c (merge_and_complain): Do not merge
15490 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
15491 fsigned_zeros, ftrapping_math, fwrapv.
15492 (append_compiler_options): Do not track these options.
15493 (append_linker_options): Likewie
15495 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15497 * cgraphunit.c (cgraph_node::finalize_function): When
15498 !flag_toplevel_reorde set no_reorder flag.
15499 (varpool_node::finalize_decl): Likewise.
15500 (symbol_table::compile): Drop no toplevel reorder path.
15502 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15504 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
15505 edges; zero probability is not better than uninitialized.
15507 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
15509 * asan.h (asan_sanitize_allocas_p): Declare.
15510 * asan.c (asan_sanitize_allocas_p): New function.
15511 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
15512 (handle_builtin_alloca): Likewise.
15513 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
15514 if !asan_sanitize_allocas_p.
15515 * params.def (asan-instrument-allocas): Add new option.
15516 * params.h (ASAN_PROTECT_ALLOCAS): Define.
15517 * opts.c (common_handle_option): Disable allocas sanitization for
15520 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
15522 * asan.c: Include gimple-fold.h.
15523 (get_last_alloca_addr): New function.
15524 (handle_builtin_stackrestore): Likewise.
15525 (handle_builtin_alloca): Likewise.
15526 (asan_emit_allocas_unpoison): Likewise.
15527 (get_mem_refs_of_builtin_call): Add new parameter, remove const
15528 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
15529 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
15530 (instrument_builtin_call): Pass gimple iterator to
15531 get_mem_refs_of_builtin_call.
15532 (last_alloca_addr): New global.
15533 * asan.h (asan_emit_allocas_unpoison): Declare.
15534 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
15535 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
15536 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
15537 if function calls alloca.
15538 * gimple-fold.c (replace_call_with_value): Remove static keyword.
15539 * gimple-fold.h (replace_call_with_value): Declare.
15540 * internal-fn.c: Include asan.h.
15541 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
15542 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
15544 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15546 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
15547 (C_SELFTEST_FLAGS): New.
15548 (CPP_SELFTEST_FLAGS): New.
15549 (SELFTEST_DEPS): New, from deps of s-selftest.
15550 (C_SELFTEST_DEPS): New, from deps of s-selftest.
15551 (CPP_SELFTEST_DEPS): New.
15552 (selftest): Add dependency on s-selftest-c++.
15553 (s-selftest): Rename to...
15554 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
15555 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
15556 than SELFTEST_FLAGS.
15557 (selftest-gdb): Rename to...
15558 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
15560 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
15561 (selftest-valgrind): Rename to...
15562 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
15564 (selftest-valgrind): Reintroduce as an alias for
15565 selftest-c-valgrind.
15566 (s-selftest-c++): New.
15567 (selftest-c++-gdb): New.
15568 (selftest-c++-valgrind): New.
15570 2017-07-06 Olivier Hainque <hainque@adacore.com>
15572 * gcc.c (process_command): When deciding if undefined variables
15573 should be ignored when processing specs, accept "gcc -v" as well.
15575 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15577 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
15578 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
15580 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15582 * config/arm/arm-cpus.in (armv8-r): Add new entry.
15583 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
15584 * config/arm/arm-tables.opt: Regenerate.
15585 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
15587 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
15589 2017-07-06 Carl Love <cel@us.ibm.com>
15591 * ChangeLog: Clean up from mid air collision
15593 2017-07-06 Carl Love <cel@us.ibm.com>
15595 * config/rs6000/rs6000-c.c: Add support for built-in functions
15596 vector signed int vec_subc (vector signed int, vector signed int);
15597 vector signed __int128 vec_subc (vector signed __int128,
15598 vector signed __int128);
15599 vector unsigned __int128 vec_subc (vector unsigned __int128,
15600 vector unsigned __int128);
15601 vector signed int vec_sube (vector signed int, vector signed int,
15602 vector signed int);
15603 vector unsigned int vec_sube (vector unsigned int,
15604 vector unsigned int,
15605 vector unsigned int);
15606 vector signed __int128 vec_sube (vector signed __int128,
15607 vector signed __int128,
15608 vector signed__int128);
15609 vector unsigned __int128 vec_sube (vector unsigned __int128,
15610 vector unsigned __int128,
15611 vector unsigned __int128);
15612 vector signed int vec_subec (vector signed int, vector signed int,
15613 vector signed int);
15614 vector unsigned int vec_subec (vector unsigned int,
15615 vector unsigned int,
15616 vector unsigned int);
15617 vector signed __int128 vec_subec (vector signed __int128,
15618 vector signed __int128,
15619 vector signed__int128);
15620 vector unsigned __int128 vec_subec (vector unsigned __int128,
15621 vector unsigned __int128,
15622 vector unsigned __int128);
15623 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
15624 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
15625 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
15626 BU_ALTIVEC_OVERLOAD_X definitions.
15627 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
15628 * doc/extend.texi: Update the built-in documentation file for the new
15629 built-in functions.
15631 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15634 * diagnostic-show-locus.c (layout::layout): Use start and finish
15635 spelling location for the start and finish of each range.
15636 * genmatch.c (linemap_client_expand_location_to_spelling_point):
15637 Add unused aspect param.
15638 * input.c (expand_location_1): Add "aspect" param, and use it
15639 to access the correct part of the location.
15640 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
15642 (expand_location_to_spelling_point): Likewise.
15643 (linemap_client_expand_location_to_spelling_point): Add "aspect"
15644 param, and pass it to expand_location_1.
15646 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
15648 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
15649 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
15650 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
15651 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
15652 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
15653 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
15654 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
15655 _mm_maskz_getmant_ss): New intrinsics.
15656 (__builtin_ia32_getexpss128_mask): Changed to ...
15657 __builtin_ia32_getexpss128_round ... this.
15658 (__builtin_ia32_getexpsd128_mask): Changed to ...
15659 __builtin_ia32_getexpsd128_round ... this.
15660 * config/i386/i386-builtin-types.def
15661 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
15662 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
15663 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
15664 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
15665 __builtin_ia32_getmantss_mask_round): New builtins.
15666 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
15667 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
15668 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
15669 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
15670 * config/i386/sse.md
15671 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
15672 avx512f_sgetexp<mode><mask_scalar_name>
15673 <round_saeonly_scalar_name> ... this.
15674 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
15675 %0, %1, %2<round_saeonly_op3>}): Changed to ...
15676 vgetexp<ssescalarmodesuffix>
15677 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15678 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
15679 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
15680 avx512f_vgetmant<mode><mask_scalar_name>
15681 <round_saeonly_scalar_name> ... this.
15682 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
15683 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
15684 vgetmant<ssescalarmodesuffix>
15685 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
15686 %0<mask_scalar_operand4>, %1, %2
15687 <round_saeonly_scalar_mask_op4>, %3} ... this.
15688 * config/i386/subst.md (mask_scalar_operand4,
15689 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
15690 round_saeonly_scalar_nimm_predicate): New subst attributes.
15692 2017-07-06 Julia Koval <julia.koval@intel.com>
15694 * config/i386/i386.c (ix86_erase_embedded_rounding):
15695 Remove code for old rounding pattern.
15697 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
15699 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
15701 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
15703 * doc/sourcebuild.texi (Test Directives, Variants of
15704 dg-require-support): Add documentation for dg-require-stack-check.
15706 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
15708 * config/i386/subst.md (mask_scalar, round_scalar,
15709 round_saeonly_scalar): New meta-templates.
15710 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
15711 round_scalar_mask_operand3, round_scalar_mask_op3,
15712 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
15713 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
15714 round_saeonly_scalar_constraint,
15715 round_saeonly_scalar_prefix): New subst attribute.
15716 * config/i386/sse.md
15717 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
15718 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
15719 <round_scalar_name> ... this.
15720 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
15721 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
15722 <round_scalar_name> ... this.
15723 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
15724 <sse>_vm<code><mode>3<mask_scalar_name>
15725 <round_saeonly_scalar_name> ... this.
15726 (v<plusminus_mnemonic><ssescalarmodesuffix>
15727 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
15728 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
15729 v<plusminus_mnemonic><ssescalarmodesuffix>
15730 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15731 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
15732 (v<multdiv_mnemonic><ssescalarmodesuffix>
15733 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
15734 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
15735 v<multdiv_mnemonic><ssescalarmodesuffix>
15736 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15737 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
15738 (v<maxmin_float><ssescalarmodesuffix>
15739 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
15740 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
15741 v<maxmin_float><ssescalarmodesuffix>
15742 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15743 %0<mask_scalar_operand3>, %1, %<iptr>2
15744 <round_saeonly_scalar_mask_op3>} ... this.
15746 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
15748 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
15749 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
15751 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
15752 Alan Hayward <alan.hayward@arm.com>
15753 David Sherwood <david.sherwood@arm.com>
15755 * combine.c (simplify_if_then_else): Remove "enum" before
15757 * compare-elim.c (can_eliminate_compare): Likewise.
15758 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15760 (aarch64_lookup_simd_builtin_type): Likewise.
15761 (aarch64_simd_builtin_type): Likewise.
15762 (aarch64_init_simd_builtin_types): Likewise.
15763 (aarch64_simd_expand_args): Likewise.
15764 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
15766 (aarch64_reverse_mask): Likewise.
15767 (aarch64_simd_emit_reg_reg_move): Likewise.
15768 (aarch64_gen_adjusted_ldpstp): Likewise.
15769 (aarch64_ccmp_mode_to_code): Likewise.
15770 (aarch64_operands_ok_for_ldpstp): Likewise.
15771 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15772 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15774 (aarch64_min_divisions_for_recip_mul): Likewise.
15775 (aarch64_reassociation_width): Likewise.
15776 (aarch64_get_condition_code_1): Likewise.
15777 (aarch64_simd_emit_reg_reg_move): Likewise.
15778 (aarch64_simd_attr_length_rglist): Likewise.
15779 (aarch64_reverse_mask): Likewise.
15780 (aarch64_operands_ok_for_ldpstp): Likewise.
15781 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15782 (aarch64_gen_adjusted_ldpstp): Likewise.
15783 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
15785 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
15786 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
15787 (arm_lookup_simd_builtin_type): Likewise.
15788 (arm_simd_builtin_type): Likewise.
15789 (arm_init_simd_builtin_types): Likewise.
15790 (arm_expand_builtin_args): Likewise.
15791 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
15792 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
15793 (ft32_setup_incoming_varargs): Likewise.
15794 (ft32_function_arg): Likewise.
15795 (ft32_function_arg_advance): Likewise.
15796 (ft32_pass_by_reference): Likewise.
15797 (ft32_arg_partial_bytes): Likewise.
15798 (ft32_valid_pointer_mode): Likewise.
15799 (ft32_addr_space_pointer_mode): Likewise.
15800 (ft32_addr_space_legitimate_address_p): Likewise.
15801 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
15803 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
15804 (ix86_emit_outlined_ms2sysv_restore): Likewise.
15805 (iamcu_alignment): Likewise.
15806 (canonicalize_vector_int_perm): Likewise.
15807 (ix86_noce_conversion_profitable_p): Likewise.
15808 (ix86_mpx_bound_mode): Likewise.
15809 (ix86_operands_ok_for_move_multiple): Likewise.
15810 * config/microblaze/microblaze-protos.h
15811 (microblaze_expand_conditional_branch_reg): Likewise.
15812 * config/microblaze/microblaze.c
15813 (microblaze_expand_conditional_branch_reg): Likewise.
15814 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15816 (rs6000_reassociation_width): Likewise.
15817 (rs6000_invalid_binary_op): Likewise.
15818 (fusion_p9_p): Likewise.
15819 (emit_fusion_p9_load): Likewise.
15820 (emit_fusion_p9_store): Likewise.
15821 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
15823 (riscv_hard_regno_mode_ok_p): Likewise.
15824 (riscv_address_insns): Likewise.
15825 (riscv_split_symbol): Likewise.
15826 (riscv_legitimize_move): Likewise.
15827 (riscv_function_value): Likewise.
15828 (riscv_hard_regno_nregs): Likewise.
15829 (riscv_expand_builtin): Likewise.
15830 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
15831 (riscv_build_integer): Likewise.
15832 (riscv_split_integer): Likewise.
15833 (riscv_legitimate_constant_p): Likewise.
15834 (riscv_cannot_force_const_mem): Likewise.
15835 (riscv_regno_mode_ok_for_base_p): Likewise.
15836 (riscv_valid_base_register_p): Likewise.
15837 (riscv_valid_offset_p): Likewise.
15838 (riscv_valid_lo_sum_p): Likewise.
15839 (riscv_classify_address): Likewise.
15840 (riscv_legitimate_address_p): Likewise.
15841 (riscv_address_insns): Likewise.
15842 (riscv_load_store_insns): Likewise.
15843 (riscv_force_binary): Likewise.
15844 (riscv_split_symbol): Likewise.
15845 (riscv_force_address): Likewise.
15846 (riscv_legitimize_address): Likewise.
15847 (riscv_move_integer): Likewise.
15848 (riscv_legitimize_const_move): Likewise.
15849 (riscv_legitimize_move): Likewise.
15850 (riscv_address_cost): Likewise.
15851 (riscv_subword): Likewise.
15852 (riscv_output_move): Likewise.
15853 (riscv_canonicalize_int_order_test): Likewise.
15854 (riscv_emit_int_order_test): Likewise.
15855 (riscv_function_arg_boundary): Likewise.
15856 (riscv_pass_mode_in_fpr_p): Likewise.
15857 (riscv_pass_fpr_single): Likewise.
15858 (riscv_pass_fpr_pair): Likewise.
15859 (riscv_get_arg_info): Likewise.
15860 (riscv_function_arg): Likewise.
15861 (riscv_function_arg_advance): Likewise.
15862 (riscv_arg_partial_bytes): Likewise.
15863 (riscv_function_value): Likewise.
15864 (riscv_pass_by_reference): Likewise.
15865 (riscv_setup_incoming_varargs): Likewise.
15866 (riscv_print_operand): Likewise.
15867 (riscv_elf_select_rtx_section): Likewise.
15868 (riscv_save_restore_reg): Likewise.
15869 (riscv_for_each_saved_reg): Likewise.
15870 (riscv_register_move_cost): Likewise.
15871 (riscv_hard_regno_mode_ok_p): Likewise.
15872 (riscv_hard_regno_nregs): Likewise.
15873 (riscv_class_max_nregs): Likewise.
15874 (riscv_memory_move_cost): Likewise.
15875 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
15876 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
15877 (rl78_addr_space_address_mode): Likewise.
15878 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15880 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15881 (rs6000_reassociation_width): Likewise.
15882 (rs6000_invalid_binary_op): Likewise.
15883 (fusion_p9_p): Likewise.
15884 (emit_fusion_p9_load): Likewise.
15885 (emit_fusion_p9_store): Likewise.
15886 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
15887 (ok_for_simple_move_operands): Likewise.
15888 (ok_for_simple_move_strict_operands): Likewise.
15889 (ok_for_simple_arith_logic_operands): Likewise.
15890 (visium_legitimize_reload_address): Likewise.
15891 (visium_select_cc_mode): Likewise.
15892 (output_cbranch): Likewise.
15893 (visium_split_double_move): Likewise.
15894 (visium_expand_copysign): Likewise.
15895 (visium_expand_int_cstore): Likewise.
15896 (visium_expand_fp_cstore): Likewise.
15897 * config/visium/visium.c (visium_pass_by_reference): Likewise.
15898 (visium_function_arg): Likewise.
15899 (visium_function_arg_advance): Likewise.
15900 (visium_libcall_value): Likewise.
15901 (visium_setup_incoming_varargs): Likewise.
15902 (visium_legitimate_constant_p): Likewise.
15903 (visium_legitimate_address_p): Likewise.
15904 (visium_legitimize_address): Likewise.
15905 (visium_secondary_reload): Likewise.
15906 (visium_register_move_cost): Likewise.
15907 (visium_memory_move_cost): Likewise.
15908 (prepare_move_operands): Likewise.
15909 (ok_for_simple_move_operands): Likewise.
15910 (ok_for_simple_move_strict_operands): Likewise.
15911 (ok_for_simple_arith_logic_operands): Likewise.
15912 (visium_function_value_1): Likewise.
15913 (rtx_ok_for_offset_p): Likewise.
15914 (visium_legitimize_reload_address): Likewise.
15915 (visium_split_double_move): Likewise.
15916 (visium_expand_copysign): Likewise.
15917 (visium_expand_int_cstore): Likewise.
15918 (visium_expand_fp_cstore): Likewise.
15919 (visium_split_cstore): Likewise.
15920 (visium_select_cc_mode): Likewise.
15921 (visium_split_cbranch): Likewise.
15922 (output_cbranch): Likewise.
15923 (visium_print_operand_address): Likewise.
15924 * expmed.c (flip_storage_order): Likewise.
15925 * expmed.h (emit_cstore): Likewise.
15926 (flip_storage_order): Likewise.
15927 * genrecog.c (validate_pattern): Likewise.
15928 * hsa-gen.c (gen_hsa_addr): Likewise.
15929 * internal-fn.c (expand_arith_overflow): Likewise.
15930 * ira-color.c (allocno_copy_cost_saving): Likewise.
15931 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15932 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
15933 (process_invariant_for_inheritance): Likewise.
15934 * lra-eliminations.c (move_plus_up): Likewise.
15935 * omp-low.c (lower_oacc_reductions): Likewise.
15936 * simplify-rtx.c (simplify_subreg): Likewise.
15937 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
15938 (TARGET_CHKP_BOUND_MODE): Likewise..
15939 * targhooks.c (default_chkp_bound_mode): Likewise.
15940 (default_setup_incoming_vararg_bounds): Likewise.
15941 * targhooks.h (default_chkp_bound_mode): Likewise.
15942 (default_setup_incoming_vararg_bounds): Likewise.
15943 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
15944 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
15945 (have_whole_vector_shift): Likewise.
15946 * tree-vect-stmts.c (vectorizable_load): Likewise.
15947 * doc/tm.texi: Regenerate.
15949 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15951 Graceful degrade if Binutils PR21472 is not available.
15954 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
15955 .rodata in flash test fails.
15956 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
15957 * confgure: Regenerate.
15958 * config.in: Regenerate.
15959 * config/avr/avr.c (avr_asm_named_section)
15960 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
15961 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
15962 (avr_asm_init_sections): Same.
15964 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15966 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
15967 (fma<VH:mode>4_intrinsic): Likewise.
15968 (*fmsub<VCVTF:mode>4): Likewise.
15969 (*fmsub<VH:mode>4_intrinsic): Likewise.
15971 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15974 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
15975 Don't depend on "optimize > 0".
15976 (out_movhi_r_mr, out_movqi_mr_r): Same.
15977 (out_movhi_mr_r, out_movqi_r_mr): Same.
15978 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
15979 io_address_operand on "optimize > 0".
15981 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15983 * tree-loop-distribution.c: Add general explanantion on the pass.
15984 (generate_loops_for_partition): Mark distributed loop.
15985 (pg_add_dependence_edges): New parameter. Handle alias data
15986 dependence specially and record it in the parameter if asked.
15987 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
15988 (init_partition_graph_vertices, add_partition_graph_edge): New.
15989 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
15990 (free_partition_graph_vdata, build_partition_graph): New.
15991 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
15992 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
15993 (data_ref_segment_size, latch_dominated_by_data_ref): New.
15994 (compute_alias_check_pairs, version_loop_by_alias_check): New.
15995 (version_for_distribution_p, finalize_partitions): New.
15996 (distribute_loop): Handle alias data dependence specially. Factor
15997 out loop fusion code as functions and call these functions.
15999 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16001 * tree-loop-distribution.c (classify_partition): New parameter and
16002 better handle reduction statement.
16003 (rdg_build_partitions): Revise comment.
16004 (distribute_loop): Compute statements in all partitions and pass it
16005 to classify_partition.
16007 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16009 * tree-loop-distribution.c (enum partition_type): New.
16010 (struct partition): New field type.
16011 (partition_merge_into): Add parameter. Update partition type.
16012 (data_dep_in_cycle_p, update_type_for_merge): New functions.
16013 (build_rdg_partition_for_vertex): Compute partition type.
16014 (rdg_build_partitions): Dump partition type.
16015 (distribute_loop): Update calls to partition_merge_into.
16017 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16019 * tree-loop-distribution.c (struct ddr_hasher): New.
16020 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
16022 (classify_partition): Call get_data_dependence.
16023 (pg_add_dependence_edges): Ditto.
16024 (distribute_loop): Release data dependence hash table.
16026 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16028 * tree-loop-distribution.c (ref_base_address): Delete.
16029 (similar_memory_accesses): Rename ...
16030 (share_memory_accesses): ... to this. Check if partitions access
16031 the same memory reference.
16032 (distribute_loop): Call share_memory_accesses.
16034 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16036 * tree-loop-distribution.c (struct partition): New field recording
16037 its data reference.
16038 (partition_alloc, partition_free): Init and release data refs.
16039 (partition_merge_into): Merge data refs.
16040 (build_rdg_partition_for_vertex): Collect data refs for partition.
16041 (pg_add_dependence_edges): Change parameters from vector to bitmap.
16043 (distribute_loop): Remve data refs from vertice data of partition
16046 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16048 * tree-loop-distribution.c (params.h): Include header file.
16049 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
16050 (datarefs_vec): New global var.
16051 (create_rdg_vertices): Use datarefs_vec directly.
16052 (free_rdg): Don't free data references.
16053 (build_rdg): Update use. Don't free data references.
16054 (distribute_loop): Compute global variable for data references.
16055 Bail out if there are too many data references.
16057 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16059 * tree-loop-distribution.c (loop_nest): New global var.
16060 (build_rdg): Use loop directly, rather than loop nest.
16061 (pg_add_dependence_edges): Remove loop nest parameter. Use global
16063 (distribute_loop): Compute global variable loop nest. Update use.
16065 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16067 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
16068 (partition_merge_into): New parameter. Dump reason for fusion.
16069 (distribute_loop): Update use of partition_merge_into.
16071 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16073 * tree-loop-distribution.c (bb_top_order_index): New.
16074 (bb_top_order_index_size, bb_top_order_cmp): New.
16075 (stmts_from_loop): Use topological order.
16076 (pass_loop_distribution::execute): Compute and release topological
16077 order for basic blocks.
16079 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16081 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
16084 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16086 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
16087 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
16088 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
16089 * internal-fn.def (LOOP_DIST_ALIAS): New.
16090 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
16091 (fold_loop_internal_call): ... this.
16092 (vect_loop_dist_alias_call): New function.
16093 (set_uid_loop_bbs): Call fold_loop_internal_call.
16094 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
16097 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16100 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
16101 Require dead FLAGS_REG at the beginning of a peephole.
16103 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16106 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
16107 arguments in the call to __builtin_ia32_sbb_u32.
16108 (_subborrow_u64): Swap _X and _Y arguments in the call to
16109 __builtin_ia32_sbb_u64.
16111 2017-07-04 Jakub Jelinek <jakub@redhat.com>
16114 * tree-vrp.c (compare_assert_loc): Turn into a function template
16115 with stable template parameter. Only test if a->e is NULL,
16116 !a->e == !b->e has been verified already. Use e == NULL or
16117 e != NULL instead of e or ! e tests. If stable is true, don't use
16118 iterative_hash_expr, on the other side allow a or b or both NULL
16119 and sort the NULLs last.
16120 (process_assert_insertions): Sort using compare_assert_loc<false>
16121 instead of compare_assert_loc, later sort using
16122 compare_assert_loc<true> before calling process_assert_insertions_for
16123 in a loop. Use break instead of continue once seen NULL pointer.
16125 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16127 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16128 Cortex-R7 and Cortex-R8 processors.
16130 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16132 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
16133 uninitialized while src is not.
16135 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
16137 * common/config/arm/arm-common.c: Adjust include path for
16139 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
16140 (arm-cpu.h): Create in build directory. Adjust dependency rules.
16141 (arm-cpu-data.h): Likewise.
16142 (arm-cpu-cdata.h): Likewise.
16143 * config/arm/arm-cpu.h: Delete.
16144 * config/arm/arm-cpu-cdata.h: Delete.
16145 * config/arm/arm-cpu-data.h: Delete.
16147 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
16149 * config/arm/arm-cpus.in (cortex-a55): New.
16150 (cortex-a75): Likewise.
16151 (cortex-a75.cortex-a55): Likewise.
16152 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
16154 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
16155 * config/arm/arm-cpu-cdata.h: Regenerate.
16156 * config/arm/arm-cpu-data.h: Regenerate.
16157 * config/arm/arm-cpu.h: Regenerate.
16158 * config/arm/arm-tables.opt: Regenerate.
16159 * config/arm/arm-tune.md: Regenerate.
16161 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16163 * haifa-sched.c (sched_create_recovery_edges): Update profile.
16165 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16167 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
16170 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
16172 PR tree-optimization/81292
16173 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
16174 full_string_p, also call adjust_related_strinfos if the adjustment
16175 is simple, otherwise invalidate related strinfos.
16177 2017-07-04 Martin Liska <mliska@suse.cz>
16180 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
16181 newly created variable as DECL_IGNORED_P.
16183 2017-07-04 Martin Liska <mliska@suse.cz>
16186 * ipa-inline.c (inline_small_functions):
16187 Use xstrdup_for_dump.
16189 2017-07-04 Tom de Vries <tom@codesourcery.com>
16191 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
16193 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16196 * config/darwin.c (darwin_function_switched_text_sections):
16199 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
16201 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
16203 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
16205 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
16207 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16209 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
16210 min_profitable_iters, and th as inclusive lower bounds.
16211 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
16212 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
16213 for min_profitable_iters and min_profitable_estimate.
16214 (vect_transform_loop): Treat th as an inclusive lower bound.
16215 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16217 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16220 * config/darwin.c (darwin_function_switched_text_sections):
16221 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
16222 in two pieces, and suppress the use of buf.
16224 2017-07-03 Nathan Sidwell <nathan@acm.org>
16226 * hash-table.h (hash_table_mod1): Fix indentation.
16228 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16230 PR middle-end/81290
16231 * predict.c (force_edge_cold): Be more careful about propagation
16233 * profile-count.h (profile_probability::guessed,
16234 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
16236 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
16238 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
16240 * doc/invoke.texi (rcpc architecture extension): Document it.
16242 2017-07-03 Richard Biener <rguenther@suse.de>
16244 PR tree-optimization/60510
16245 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
16246 the scalar reduction PHI and use it.
16247 (vectorizable_reduction): Properly guard the single_defuse_cycle
16248 path for non-SLP reduction chains where we cannot use it.
16249 Rework reduc_def/index and vector type deduction. Rework
16250 vector operand gathering during reduction op code-gen.
16251 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
16252 chains dissolve the chain and leave it to non-SLP reduction
16255 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16257 * tree-data-ref.h (dr_alignment): Declare.
16258 * tree-data-ref.c (dr_alignment): New function.
16259 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
16260 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
16262 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
16264 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16266 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
16267 and base_misalignment fields.
16268 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
16269 * tree-data-ref.c: Include builtins.h.
16270 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
16271 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
16272 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
16273 * tree-vect-data-refs.c: Include tree-cfg.h.
16274 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
16275 fields instead of calculating an alignment here.
16276 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
16277 innermost_loop_behavior fields.
16279 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16281 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
16283 (DR_STEP_ALIGNMENT): New macro.
16284 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
16285 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
16286 (create_data_ref): Print it.
16287 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
16288 to tell whether the step preserves vector (mis)alignment.
16289 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
16290 Move the check for an integer step and generalise to all INTEGER_CST.
16291 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
16292 Print the outer step alignment.
16294 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16296 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
16297 with offset_alignment.
16298 (DR_ALIGNED_TO): Delete.
16299 (DR_OFFSET_ALIGNMENT): New macro.
16300 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
16301 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
16302 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
16303 (create_data_ref): Likewise.
16304 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
16305 (vect_analyze_data_refs): Likewise.
16306 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
16307 creating dummy innermost behavior.
16309 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16311 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
16312 with a "innermost_loop_behavior *" and refeence tree.
16313 * tree-data-ref.c (dr_analyze_innermost): Likewise.
16314 (create_data_ref): Update call accordingly.
16315 * tree-predcom.c (find_looparound_phi): Likewise.
16317 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16319 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
16320 fields with dr_wrt_vec_loop.
16321 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
16322 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
16323 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
16324 (vect_dr_behavior): New function.
16325 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
16326 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
16327 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
16328 track whether the step preserves the misalignment.
16329 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
16330 Use vect_dr_behavior.
16331 (vect_setup_realignment): Update call accordingly.
16332 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
16333 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
16334 call to vect_create_addr_base_for_vector_ref.
16335 (vect_create_cond_for_align_checks): Likewise.
16336 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
16337 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
16338 (vect_recog_mask_conversion_pattern): Likewise.
16339 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
16340 (new_stmt_vec_info): Remove redundant zeroing.
16342 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
16344 * common/config/arm/arm-common.c (arm_be8_option): New function.
16345 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
16346 (ISA_ARMv6): Add isa_bit_be8.
16347 * config/arm/arm.h (arm_be8_option): Add prototype.
16348 (BE8_SPEC_FUNCTION): New define.
16349 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
16350 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
16351 (mlittle-endian): Similarly.
16352 (mbe8, mbe32): New options.
16353 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
16354 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
16356 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16358 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
16360 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16362 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
16363 (cleanup_tree_cfg_bb): Use it.
16364 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
16366 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
16368 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16371 * loop-doloop.c (add_test): Update profile.
16373 2017-07-03 Martin Liska <mliska@suse.cz>
16376 * sanopt.c (rewrite_usage_of_param): New function.
16377 (sanitize_rewrite_addressable_params): Likewise.
16378 (pass_sanopt::execute): Call rewrite_usage_of_param.
16380 2017-07-03 Richard Biener <rguenther@suse.de>
16382 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
16383 back to using VIEW_CONVERT_EXPR.
16385 2017-07-03 Martin Liska <mliska@suse.cz>
16388 * doc/extend.texi: Document when a resolver function is
16389 generated for target_clones.
16391 2017-07-03 Martin Liska <mliska@suse.cz>
16393 * asan.c (asan_emit_stack_protection): Unpoison just red zones
16394 and shadow memory of auto variables which are subject of
16395 use-after-scope sanitization.
16396 (asan_expand_mark_ifn): Add do set only when is_poison.
16398 2016-07-03 Richard Biener <rguenther@suse.de>
16400 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
16402 (vect_force_simple_reduction): Record reduction def -> phi mapping.
16403 (vectorizable_reduction): Perform reduction PHI creation when
16404 visiting a reduction PHI and adjust and simplify code generation
16405 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
16406 (vect_transform_loop): Visit reduction PHIs.
16407 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
16408 defs into the SLP tree.
16409 (vect_build_slp_tree): Reduction defs terminate the recursion.
16410 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
16412 (vect_get_vec_defs_for_stmt_copy): Export.
16413 (vect_get_vec_defs): Likewise.
16414 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
16416 (vect_get_vec_defs_for_stmt_copy): Declare.
16417 (vect_get_vec_defs): Likewise.
16419 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16421 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
16422 parameter with a "loop" parameter and use it instead of the
16423 loop containing DR_STMT. Don't check simple_iv when doing
16424 BB analysis. Describe the two analysis modes in the comment.
16426 2017-07-03 Tom de Vries <tom@codesourcery.com>
16428 PR tree-optimization/69468
16429 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
16430 (find_same_succ_bb): Handle ignore_edge_flags.
16432 2017-07-03 Tom de Vries <tom@codesourcery.com>
16434 PR tree-optimization/81192
16435 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
16437 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
16439 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
16441 2017-07-03 Tom de Vries <tom@codesourcery.com>
16443 PR tree-optimization/81192
16444 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
16445 BB_SAME_SUCC (bb) == NULL.
16447 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16449 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
16452 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16454 * dumpfile.c: Include profile-count.h
16455 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
16457 (insert_cond_bb): Update profile.
16458 * tree-cfg.h (insert_cond_bb): Update prototype.
16459 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
16460 * tree-dump.c: Do not include tree-cfg.
16462 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16464 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
16466 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16468 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
16471 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16473 * tree-complex.c (expand_complex_div_wide): update profile.
16475 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16476 Alan Hayward <alan.hayward@arm.com>
16477 David Sherwood <david.sherwood@arm.com>
16479 * Makefile.in (MACHMODE_H): Remove insn-modes.h
16480 (CORETYPES_H): New define.
16481 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
16482 (insn-modes-inline.h, s-modes-inline-h): New rules.
16483 (generated_files): Add insn-modes-inline.h.
16484 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
16485 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
16486 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
16487 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
16488 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
16489 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
16490 (build/gencodes.o, build/genconditions.o): Likewise.
16491 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
16492 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
16493 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
16494 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
16495 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
16496 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
16497 * coretypes.h: Include everything up to real.h for generators.
16498 Include insn-modes.h first. Include wide-int-print.h after
16499 wide-int.h. Include insn-modes-inline.h and then machmode.h.
16500 * machmode.h: Don't include insn-modes.h here.
16501 * function-tests.c: Remove includes of signop.h, machmode.h,
16502 double-int.h and wide-int.h.
16504 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
16506 * optc-save-gen.awk: Likewise.
16507 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
16508 * godump.c: Remove include of wide-int-print.h.
16509 * pretty-print.h: Likewise.
16510 * wide-int-print.cc: Likewise.
16511 * wide-int.cc: Likewise.
16512 * hash-map-tests.c: Remove include of signop.h.
16513 * hash-set-tests.c: Likewise.
16514 * rtl-tests.c: Likewise.
16515 * mkconfig.sh: Remove include of machmode.h.
16516 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
16518 (emit_insn_modes_inline_h): ...this new function. Emit the code
16519 into an insn-modes-inline.h header file, adding appropriate
16520 include guards and end comments.
16521 (emit_insn_modes_c_header): Remove include of machmode.h.
16522 (emit_min_insn_modes_c_header): Include coretypes.h rather than
16524 (main): Handle -i flag and call emit_insn_modes_inline_h when
16527 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16529 * tree-ssa-strlen.c (strinfo): Rename the length field to
16530 nonzero_chars. Add a full_string_p field.
16531 (compare_nonzero_chars, zero_length_string_p): New functions.
16532 (get_addr_stridx): Add an offset_out parameter.
16533 Use compare_nonzero_chars.
16534 (get_stridx): Update accordingly. Use compare_nonzero_chars.
16535 (new_strinfo): Update after above changes to strinfo.
16536 (set_endptr_and_length): Set full_string_p.
16537 (get_string_length): Update after above changes to strinfo.
16538 (unshare_strinfo): Update call to new_strinfo.
16539 (maybe_invalidate): Likewise.
16540 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
16541 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
16542 as a uhwi instead of an shwi. Update after above changes to
16543 strinfo and new_strinfo.
16544 (zero_length_string): Assert that chainsi contains full strings.
16545 Use zero_length_string_p. Update call to new_strinfo.
16546 (adjust_related_strinfos): Update after above changes to strinfo.
16547 Copy full_string_p from origsi.
16548 (adjust_last_stmt): Use zero_length_string_p.
16549 (handle_builtin_strlen): Update after above changes to strinfo and
16550 new_strinfo. Install the lhs as the string length if the previous
16551 entry didn't describe a full string.
16552 (handle_builtin_strchr): Update after above changes to strinfo
16554 (handle_builtin_strcpy): Likewise.
16555 (handle_builtin_strcat): Likewise.
16556 (handle_builtin_malloc): Likewise.
16557 (handle_pointer_plus): Likewise.
16558 (handle_builtin_memcpy): Likewise. Track nonzero characters
16559 that aren't necessarily followed by a nul terminator.
16560 (handle_char_store): Likewise.
16562 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16564 PR tree-optimization/80769
16565 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
16566 for malloc and calloc. Document the new invariant that all related
16567 strinfos have delayed lengths or none do.
16568 (verify_related_strinfos): Move earlier in file.
16569 (set_endptr_and_length): New function, split out from...
16570 (get_string_length): ...here. Also set the lengths of related
16572 (zero_length_string): Assert that chainsi has known (rather than
16574 (adjust_related_strinfos): Likewise.
16576 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16578 PR tree-optimization/81136
16579 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
16580 assert that two references with the same misalignment have the same
16581 compile-time misalignment if those compile-time misalignments
16584 2017-07-01 Andi Kleen <ak@linux.intel.com>
16586 * print-tree.c (print_node): Print all attributes.
16588 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16590 * cfg.c (scale_bbs_frequencies): New function.
16591 * cfg.h (scale_bbs_frequencies): Declare it.
16592 * cfgloopanal.c (single_likely_exit): Cleanup.
16593 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
16595 (scale_loop_profile): Likewise.
16596 (loop_version): Likewise.
16597 (create_empty_loop_on_edge): Update.
16598 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
16599 scale_loop_frequencies, scale_loop_profile, loopify,
16600 loop_version): Update prototypes.
16601 * modulo-sched.c (sms_schedule): Update.
16602 * predict.c (unlikely_executed_edge_p): Also check probability.
16603 (probably_never_executed_edge_p): Fix typo.
16604 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16605 * tree-parloops.c (gen_parallel_loop): Update.
16606 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
16607 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16608 * tree-ssa-loop-split.c (split_loop): Update.
16609 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16610 * tree-vect-loop-manip.c (vect_do_peeling): Update.
16611 (vect_loop_versioning): Update.
16612 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16614 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16616 * trans-mem.c (split_bb_make_tm_edge): Update profile.
16618 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16620 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
16621 to keep profile consistent.
16623 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16625 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
16626 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
16627 * profile-count.h (max_safe_multiplier): Make unsigned.
16628 (profile_count::guessed_zero): New.
16630 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16632 * bb-reorder.c (fix_up_crossing_landing_pad,
16633 fix_crossing_conditional_branches): Use make_single_succ_edge
16634 to keep profile consistent.
16636 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16638 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
16641 2017-07-01 Jakub Jelinek <jakub@redhat.com>
16644 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
16645 the right scopes, make sure cond_jump isn't preserved between multiple
16646 iterations. Search for fallthru edge whenever there are 3+ edges and
16647 use find_fallthru_edge for it.
16649 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16651 Patch by Alexander Monakov <amonakov@ispras.ru>
16652 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
16653 probabilities consistently.
16655 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16657 * pa.c (pa_expand_compare_and_swap_loop): Update call of
16658 emit_cmp_and_jump_insns.
16660 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16663 * tree-inline.c (expand_call_inline): Combine profile statuses.
16665 2017-06-30 Andrew Pinski <apinski@cavium.com>
16667 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
16668 fold_stmt returned true.
16670 2017-06-30 Nathan Sidwell <nathan@acm.org>
16672 * ggc.h (empty_string): Delete.
16673 * cfgexpand.c (expand_asm_stmt): Use plain "".
16674 * optabs.c (expand_asm_memory_barrier): Likewise.
16675 * stringpool.c (empty_string): Delete.
16676 (digit_vector, digit_string): Delete.
16677 (ggc_alloc_string): Use plain "", don't optimize single digit
16678 strings. Use ggc_alloc_atomic.
16680 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
16682 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
16683 comparison set and one other set, use the cost of the non-comparison
16686 2017-06-30 Nathan Sidwell <nathan@acm.org>
16688 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
16691 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
16693 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
16694 loops. Remove now unneeded calls to gimple_switch_set_label() that
16695 just set removed labels to NULL_TREE.
16697 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
16699 * tree-ssanames.c (set_range_info_raw): Abstract from ...
16700 (set_range_info): ...here. Only call set_range_info_raw if domain
16702 (set_nonzero_bits): Call set_range_info_raw.
16703 * tree-ssanames.h (set_range_info_raw): New.
16705 2017-06-30 Jakub Jelinek <jakub@redhat.com>
16708 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
16709 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
16710 of nonimmediate_operand and <store_mask_constraint> instead of m
16711 for the input operand. For V8FI iterator, always split if input
16712 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
16713 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
16714 <store_mask_predicate> instead of register_operand and
16715 <store_mask_constraint> instead of v for the input operand. Make
16716 sure both operands aren't MEMs for if not <mask_applied>.
16718 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
16720 * lto-wrapper.c (copy_file) Close both file descriptors before
16723 2017-06-30 Martin Liska <mliska@suse.cz>
16726 * multiple_target.c (create_dispatcher_calls): Make ifunc
16727 also for function that don't have calls or are not referenced.
16729 2017-06-30 Richard Biener <rguenther@suse.de>
16731 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
16732 analyze the first scalar stmt. Move vector type computation
16733 for the BB case here from ...
16734 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
16735 live operation processing in the SLP case properly.
16737 2017-06-30 Richard Biener <rguenther@suse.de>
16739 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
16741 2017-06-30 Martin Liska <mliska@suse.cz>
16744 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
16745 before BUILT_IN_UNWIND_RESUME when ASAN is used.
16747 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
16749 * doc/invoke.texi (AArch64): Add missing options and remove redundant
16752 2017-06-30 Richard Biener <rguenther@suse.de>
16754 PR tree-optimization/81249
16755 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
16756 condition reduction result to original scalar type.
16758 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16760 * profile-count.h (enum profile_quality): Fix typos and whitespace
16763 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16765 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
16766 type for branch probabilities.
16768 2017-06-29 Julian Brown <julian@codesourcery.com>
16769 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16771 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
16772 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
16773 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
16774 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
16776 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16778 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
16779 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
16780 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
16781 CC usage from generic code to here.
16782 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
16783 CC usage into the target macros.
16785 2017-06-29 Maya Rashish <coypu@sdf.org>
16787 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
16790 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16792 * arm/arm-builtins.c: Include profile-count.h
16793 * except.c (sjlj_emit_function_enter): Use
16794 profile_probability::unlikely.
16796 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16798 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
16799 and tocrel_offset be pointer args rather than implicitly using
16801 (legitimate_constant_pool_address_p, rs6000_emit_move,
16802 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
16803 tocrel_offset and use in toc_relative_expr_p call.
16804 (print_operand, print_operand_address): Use static tocrel_base_oac
16805 and tocrel_offset_oac.
16806 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
16809 2017-06-29 Maya Rashish <coypu@sdf.org>
16811 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
16813 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
16815 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
16816 objects, take into account only the alignment of 'op0' and 'mode1' if
16819 2017-06-29 Steve Ellcey <sellcey@cavium.com>
16821 * ccmp.c (ccmp_tree_comparison_p): New function.
16822 (ccmp_candidate_p): Update to use above function.
16823 (get_compare_parts): New function.
16824 (expand_ccmp_next): Update to use new functions.
16825 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
16827 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
16828 take mode as argument.
16829 * ccmp.h (expand_ccmp_expr): Add mode as argument.
16830 * expr.c (expand_expr_real_1): Pass mode as argument.
16832 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
16834 * combine.c (combine_instructions): Print insns to dump_file, together
16837 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16839 * asan.c (asan_emit_stack_protection): Update.
16840 (create_cond_insert_point): Update.
16841 * auto-profile.c (afdo_propagate_circuit): Update.
16842 * basic-block.h (struct edge_def): Turn probability to
16843 profile_probability.
16844 (EDGE_FREQUENCY): Update.
16845 * bb-reorder.c (find_traces_1_round): Update.
16846 (better_edge_p): Update.
16847 (sanitize_hot_paths): Update.
16848 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
16849 (make_single_succ_edge): Update.
16850 (check_bb_profile): Update.
16851 (dump_edge_info): Update.
16852 (update_bb_profile_for_threading): Update.
16853 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
16854 probabilitycount to 0.
16855 * cfgbuild.c (compute_outgoing_frequencies): Update.
16856 * cfgcleanup.c (try_forward_edges): Update.
16857 (outgoing_edges_match): Update.
16858 (try_crossjump_to_edge): Update.
16859 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
16860 (expand_gimple_tailcall): Update.
16861 (construct_init_block): Use make_single_succ_edge.
16862 (construct_exit_block): Use make_single_succ_edge.
16863 * cfghooks.c (verify_flow_info): Update.
16864 (redirect_edge_succ_nodup): Update.
16865 (split_edge): Update.
16866 (account_profile_record): Update.
16867 * cfgloopanal.c (single_likely_exit): Update.
16868 * cfgloopmanip.c (scale_loop_profile): Update.
16869 (set_zero_probability): Remove.
16870 (duplicate_loop_to_header_edge): Update.
16871 * cfgloopmanip.h (loop_version): Update prototype.
16872 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
16873 (force_nonfallthru_and_redirect): Update.
16874 (update_br_prob_note): Update.
16875 (rtl_verify_edges): Update.
16876 (purge_dead_edges): Update.
16877 (rtl_lv_add_condition_to_bb): Update.
16878 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
16879 * cgraphunit.c (init_lowered_empty_function): Update.
16880 (cgraph_node::expand_thunk): Update.
16881 * cilk-common.c: Include profile-count.h
16882 * dojump.c (inv): Remove.
16883 (jumpifnot): Update.
16884 (jumpifnot_1): Update.
16885 (do_jump_1): Update.
16887 (do_jump_by_parts_greater_rtx): Update.
16888 (do_compare_rtx_and_jump): Update.
16889 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
16890 do_jump_1. do_compare_rtx_and_jump): Update prototype.
16891 * dwarf2cfi.c: Include profile-count.h
16892 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
16893 (sjlj_emit_dispatch_table): Likewise.
16894 * explow.c: Include profile-count.h
16895 * expmed.c (emit_store_flag_force): Update.
16896 (do_cmp_and_jump): Update.
16897 * expr.c (compare_by_pieces_d::generate): Update.
16898 (compare_by_pieces_d::finish_mode): Update.
16899 (emit_block_move_via_loop): Update.
16900 (store_expr_with_bounds): Update.
16901 (store_constructor): Update.
16902 (expand_expr_real_2): Update.
16903 (expand_expr_real_1): Update.
16904 * expr.h (try_casesi, try_tablejump): Update prototypes.
16905 * gimple-pretty-print.c (dump_probability): Update.
16906 (dump_profile): New.
16907 (dump_gimple_label): Update.
16908 (dump_gimple_bb_header): Update.
16909 * graph.c (draw_cfg_node_succ_edges): Update.
16910 * hsa-gen.c (convert_switch_statements): Update.
16911 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
16912 (find_if_case_1): Update.
16913 (find_if_case_2): Update.
16914 * internal-fn.c (expand_arith_overflow_result_store): Update.
16915 (expand_addsub_overflow): Update.
16916 (expand_neg_overflow): Update.
16917 (expand_mul_overflow): Update.
16918 (expand_vector_ubsan_overflow): Update.
16919 * ipa-cp.c (good_cloning_opportunity_p): Update.
16920 * ipa-split.c (split_function): Use make_single_succ_edge.
16921 * ipa-utils.c (ipa_merge_profiles): Update.
16922 * loop-doloop.c (add_test): Update.
16923 (doloop_modify): Update.
16924 * loop-unroll.c (compare_and_jump_seq): Update.
16925 (unroll_loop_runtime_iterations): Update.
16926 * lra-constraints.c (lra_inheritance): Update.
16927 * lto-streamer-in.c (input_cfg): Update.
16928 * lto-streamer-out.c (output_cfg): Update.
16929 * mcf.c (adjust_cfg_counts): Update.
16930 * modulo-sched.c (sms_schedule): Update.
16931 * omp-expand.c (expand_omp_for_init_counts): Update.
16932 (extract_omp_for_update_vars): Update.
16933 (expand_omp_ordered_sink): Update.
16934 (expand_omp_for_ordered_loops): Update.
16935 (expand_omp_for_generic): Update.
16936 (expand_omp_for_static_nochunk): Update.
16937 (expand_omp_for_static_chunk): Update.
16938 (expand_cilk_for): Update.
16939 (expand_omp_simd): Update.
16940 (expand_omp_taskloop_for_outer): Update.
16941 (expand_omp_taskloop_for_inner): Update.
16942 * omp-simd-clone.c (simd_clone_adjust): Update.
16943 * optabs.c (expand_doubleword_shift): Update.
16944 (expand_abs): Update.
16945 (emit_cmp_and_jump_insn_1): Update.
16946 (expand_compare_and_swap_loop): Update.
16947 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
16948 * predict.c (predictable_edge_p): Update.
16949 (edge_probability_reliable_p): Update.
16950 (set_even_probabilities): Update.
16951 (combine_predictions_for_insn): Update.
16952 (combine_predictions_for_bb): Update.
16953 (propagate_freq): Update.
16954 (estimate_bb_frequencies): Update.
16955 (force_edge_cold): Update.
16956 * profile-count.c (profile_count::dump): Add missing space into dump.
16957 (profile_count::debug): Add newline.
16958 (profile_count::differs_from_p): Explicitly convert to unsigned.
16959 (profile_count::stream_in): Update.
16960 (profile_probability::dump): New member function.
16961 (profile_probability::debug): New member function.
16962 (profile_probability::differs_from_p): New member function.
16963 (profile_probability::differs_lot_from_p): New member function.
16964 (profile_probability::stream_in): New member function.
16965 (profile_probability::stream_out): New member function.
16966 * profile-count.h (profile_count_quality): Rename to ...
16967 (profile_quality): ... this one.
16968 (profile_probability): New.
16969 (profile_count): Update.
16970 * profile.c (compute_branch_probabilities): Update.
16971 * recog.c (peep2_attempt): Update.
16972 * sched-ebb.c (schedule_ebbs): Update.
16973 * sched-rgn.c (find_single_block_region): Update.
16974 (compute_dom_prob_ps): Update.
16975 (schedule_region): Update.
16976 * sel-sched-ir.c (compute_succs_info): Update.
16977 * stmt.c (struct case_node): Update.
16978 (do_jump_if_equal): Update.
16979 (get_outgoing_edge_probs): Update.
16980 (conditional_probability): Update.
16981 (emit_case_dispatch_table): Update.
16982 (expand_case): Update.
16983 (expand_sjlj_dispatch_table): Update.
16984 (emit_case_nodes): Update.
16985 * targhooks.c: Update.
16986 * tracer.c (better_p): Update.
16987 (find_best_successor): Update.
16988 * trans-mem.c (expand_transaction): Update.
16989 * tree-call-cdce.c: Update.
16990 * tree-cfg.c (gimple_split_edge): Upate.
16991 (move_sese_region_to_fn): Upate.
16992 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
16993 * tree-eh.c (lower_resx): Upate.
16994 (cleanup_empty_eh_move_lp): Upate.
16995 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16996 * tree-inline.c (copy_edges_for_bb): Update.
16997 (copy_cfg_body): Update.
16998 * tree-parloops.c (gen_parallel_loop): Update.
16999 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
17000 (gimple_gen_time_profiler): Update.
17001 * tree-ssa-dce.c (remove_dead_stmt): Update.
17002 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17003 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
17004 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
17005 (unloop_loops): Update.
17006 (try_peel_loop): Update.
17007 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17008 * tree-ssa-loop-split.c (connect_loops): Update.
17009 (split_loop): Update.
17010 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17011 (hoist_guard): Update.
17012 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
17013 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
17014 (value_replacement): Update.
17015 * tree-ssa-reassoc.c (branch_fixup): Update.
17016 * tree-ssa-tail-merge.c (replace_block_by): Update.
17017 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
17018 (create_edge_and_update_destination_phis): Update.
17019 (compute_path_counts): Update.
17020 (recompute_probabilities): Update.
17021 (update_joiner_offpath_counts): Update.
17022 (freqs_to_counts_path): Update.
17023 (duplicate_thread_path): Update.
17024 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
17025 (struct switch_conv_info): Update.
17026 (gen_inbound_check): Update.
17027 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17028 (vect_do_peeling): Update.
17029 (vect_loop_versioning): Update.
17030 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17031 (optimize_mask_stores): Update.
17032 * ubsan.c (ubsan_expand_null_ifn): Update.
17033 * value-prof.c (gimple_divmod_fixed_value): Update.
17034 (gimple_divmod_fixed_value_transform): Update.
17035 (gimple_mod_pow2): Update.
17036 (gimple_mod_pow2_value_transform): Update.
17037 (gimple_mod_subtract): Update.
17038 (gimple_mod_subtract_transform): Update.
17039 (gimple_ic): Update.
17040 (gimple_stringop_fixed_value): Update.
17041 (gimple_stringops_transform): Update.
17042 * value-prof.h: Update.
17044 2017-06-29 Carl Love <cel@us.ibm.com>
17046 * config/rs6000/rs6000-c.c: Add support for built-in functions
17047 vector signed int vec_signed (vector float);
17048 vector signed long long vec_signed (vector double);
17049 vector signed int vec_signed2 (vector double, vector double);
17050 vector signed int vec_signede (vector double);
17051 vector signed int vec_signedo (vector double);
17052 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
17053 instruction generator.
17054 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17055 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
17056 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
17058 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
17059 vunsignede_v2df): Add define_expands.
17060 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
17061 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
17062 VEC_UNSIGNEDO): Add definitions.
17063 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17064 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
17065 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
17066 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
17067 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
17068 * config/rs6000/altivec.h (vec_signed, vec_signed2,
17069 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
17070 vec_unsignede, vec_unsignedo): Add builtin defines.
17071 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
17073 * doc/extend.texi: Update the built-in documentation file for the
17074 new built-in functions.
17076 2017-06-29 Richard Biener <rguenther@suse.de>
17078 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
17079 reduction chains to LOOP_VINFO_REDUCTIONS.
17080 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
17081 SLP reductions after processing reduction chains.
17083 2017-06-29 Nathan Sidwell <nathan@acm.org>
17085 * builtins.c (fold_builtin_FUNCTION): Use
17086 lang_hooks.decl_printable_name.
17088 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
17090 PR middle-end/81194
17091 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
17092 with only one label.
17093 * stmt.c (expand_case): Assert NCASES is greater than one.
17095 2017-06-29 Richard Biener <rguenther@suse.de>
17097 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
17099 (group_case_labels): Likewise.
17100 (find_taken_edge): Push sanity checking on val to workers...
17101 (find_taken_edge_cond_expr): ... here
17102 (find_taken_edge_switch_expr): ... and here, handle cases
17103 with just a default label.
17104 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
17105 (group_case_labels): Likewise.
17106 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
17107 group_case_labels does anything cleanup the CFG again.
17109 2017-06-29 Bin Cheng <bin.cheng@arm.com>
17111 PR tree-optimization/81196
17112 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
17113 exit condition comparing two IVs.
17115 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
17117 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
17118 profile to the dummy entry at the end of the list of architectures.
17119 * config/arm/arm-cpu-cdata.h: Regenerated.
17121 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17122 Michael Collison <michael.collison@arm.com>
17125 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
17127 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
17128 (*aarch64_reg_<mode>3_minus_mask): New pattern.
17129 (*aarch64_<optab>_reg_di3_mask2): New pattern.
17130 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
17131 of shift when the shift amount is masked with constant equal to
17132 the size of the mode.
17133 * config/aarch64/predicates.md (subreg_lowpart_operator): New
17136 2017-06-29 Martin Liska <mliska@suse.cz>
17138 * config/i386/i386.opt: Change range from [1,5] to [0,5].
17140 2017-06-29 Yury Gribov <tetra2005@gmail.com>
17143 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
17146 (edge_growth_cache_entry::edge_growth_cache_entry): New
17148 (reset_edge_growth_cache): Update to use constructor.
17150 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17152 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
17153 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17154 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17156 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
17158 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
17159 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
17161 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
17163 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
17164 (*-linux-uclibc*): Add t-uclibc tmake_file.
17165 * config/t-musl: New.
17166 * config/t-uclibc: New.
17168 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
17170 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
17172 (gen_comm_data): Emit architectural setting of arch_prof.
17173 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
17175 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
17176 (armv8-m.base, armv8-m.main): Likewise.
17177 * arm-protos.h (arm_build_target): Add profile field.
17178 (arch_option): Likewise.
17179 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
17181 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
17182 arm_active_target.profile.
17184 2017-06-28 Richard Biener <rguenther@suse.de>
17186 PR middle-end/81227
17187 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
17188 TYPE_OVERFLOW_WRAPS.
17189 * match.pd (negate_expr_p): Likewise.
17190 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
17191 fold_build2, not fold_binary.
17193 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17195 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
17196 Convert memory address to Pmode.
17197 (aarch64_print_operand): Assert MEM operands are always Pmode.
17199 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17202 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
17203 Remove redundant if.
17204 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
17205 * config/arm/aarch-common-protos.h
17206 (aarch_forward_to_shift_is_not_shifted_re): Remove.
17207 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
17209 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
17212 * multiple_target.c (create_dispatcher_calls): Set the default
17213 clone to be static, not public.
17215 2017-06-28 Richard Biener <rguenther@suse.de>
17217 * tree-vect-loop.c (vectorizable_reduction): Move special
17218 cond reduction IV var creation ...
17219 (vect_create_epilog_for_reduction): ... here. Remove induction_index
17220 parameter. Use STMT_VINFO_VECTYPE.
17221 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
17224 2017-06-28 Martin Liska <mliska@suse.cz>
17227 * ipa-visibility.c (non_local_p): Handle visibility.
17229 2017-06-28 Martin Liska <mliska@suse.cz>
17232 * common.opt: Add IntegerRange to various options.
17233 * opt-functions.awk (integer_range_info): New function.
17234 * optc-gen.awk: Add integer_range_info to cl_options struct.
17235 * opts-common.c (decode_cmdline_option): Handle
17236 CL_ERR_INT_RANGE_ARG.
17237 (cmdline_handle_error): Likewise.
17238 * opts.c (print_filtered_help): Show valid interval in
17239 when --help is provided.
17240 * opts.h (struct cl_option): Add range_min and range_max fields.
17241 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
17243 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
17245 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
17246 (x * C EQ/NE y * C): New transformation.
17248 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
17250 * genmultilib (combination_space): Accept '+' in option names.
17252 2017-06-28 Martin Liska <mliska@suse.cz>
17255 * asan.c (instrument_derefs): Bail out inner references
17256 that are hard register variables.
17258 2017-06-28 Jakub Jelinek <jakub@redhat.com>
17261 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
17262 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
17264 2017-06-28 Richard Biener <rguenther@suse.de>
17266 * tree-vectorizer.h (vect_get_vec_defs): Remove.
17267 (vect_get_slp_defs): Adjust.
17268 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
17270 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
17272 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
17273 get_initial_defs_for_reduction instead of vect_get_vec_defs.
17274 (vectorizable_reduction): Adjust.
17275 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
17277 (vect_get_slp_defs): Likewise.
17278 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
17279 (vectorizable_bswap): Adjust.
17280 (vectorizable_call): Likewise.
17281 (vectorizable_conversion): Likewise.
17282 (vectorizable_assignment): Likewise.
17283 (vectorizable_shift): Likewise.
17284 (vectorizable_operation): Likewise.
17285 (vectorizable_store): Likewise.
17286 (vectorizable_condition): Likewise.
17287 (vectorizable_comparison): Likewise.
17289 2017-06-28 Michael Collison <michael.collison@arm.com>
17292 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
17294 (arm_gen_movmemqi): Removed unused variable 'i'.
17295 Convert 'for' loop into 'while' loop.
17296 (arm_expand_prologue): Remove last unnecessary set of insn.
17297 (thumb_pop): Remove unused variable 'pushed_words'.
17298 (thumb_exit): Remove last unnecessary set of regs_to_pop.
17300 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17302 * config/s390/predicates.md: Use s390_rel_address_ok_p.
17303 * config/s390/s390-protos.h: Add prototype of
17304 s390_rel_address_ok_p.
17305 * config/s390/s390.c (s390_got_symbol): New function.
17306 (s390_rel_address_ok_p): New function.
17307 (legitimize_pic_address): Use s390_rel_address_ok_p.
17308 (s390_load_got): Use s390_got_symbol.
17309 (s390_option_override): Issue error if
17310 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
17311 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
17313 * config/s390/s390.opt: New option mpic-data-is-text-relative.
17315 2017-06-27 Andrew Pinski <apinski@cavium.com>
17317 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
17318 (X * copysign (1.0, X)): New pattern.
17319 (X * copysign (1.0, -X)): New pattern.
17320 (copysign (-1.0, CST)): New pattern.
17322 2017-06-27 Joseph Myers <joseph@codesourcery.com>
17324 * genmultilib (combination_space): Remove variable.
17325 Validate reuse rules against regular expression for any sequence
17326 of multilib options in any order.
17328 2017-06-27 Michael Collison <michael.collison@arm.com>
17330 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
17331 call aarch64_split_simd_combine.
17332 * (aarch64_combine_internal<mode>): Delete pattern.
17333 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
17334 Allow register and subreg operands.
17336 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17338 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
17339 specific need, just fallback on defaults.
17340 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
17342 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17343 Olivier Hainque <hainque@adacore.com>
17345 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
17347 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
17348 targets. Pick a default if no particular attempt applied.
17349 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
17352 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17354 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
17355 (x86_64-wrs-vxworks7): Likewise.
17357 2017-06-27 Marek Polacek <polacek@redhat.com>
17360 * ubsan.c (instrument_null): Check get_base_address's result for null.
17362 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
17364 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
17366 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
17368 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
17369 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
17370 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
17371 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
17372 New function types.
17373 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
17374 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
17375 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
17376 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
17377 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
17378 BUILT_IN_FEUPDATEENV): New builtins.
17379 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
17380 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
17381 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
17382 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
17384 (builtin_structptr_types): Adjust size.
17385 * tree.c (builtin_structptr_types): Add four entries.
17387 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17388 Olivier Hainque <hainque@adacore.com>
17390 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
17391 (TLS_SYM): New local macro, forcing reference to __tls__ on
17392 link command lines for VxWorks 7 RTPs, triggering initialization
17394 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
17395 OS features TLS support, true for RTPs on VxWorks 7.
17396 * config/vxworks.c (vxworks_override_options): Setup emutls
17399 2017-06-27 Jakub Jelinek <jakub@redhat.com>
17401 * predict.c (test_prediction_value_range): Use -1U instead of -1
17402 to avoid narrowing conversion warning.
17403 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
17404 to avoid narrowing conversion warning.
17405 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
17407 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
17409 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17411 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
17412 64bit configurations.
17413 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
17414 (SIZE_TYPE): Likewise.
17415 * config/vxworks.c (vxworks_emutls_var_fields): Use
17416 long_unsigned_type_node instead of unsigned_type_node as the offset
17417 field type, which is "pointer" mode in emutls.c.
17419 2017-06-27 Jakub Jelinek <jakub@redhat.com>
17422 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
17424 PR middle-end/81207
17425 * gimple-fold.c (replace_call_with_call_and_fold): Handle
17426 gimple_vuse copying separately from gimple_vdef copying.
17428 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17430 * value-prof.c (free_hist): Remove call to memset and the enclosing if
17433 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
17434 Olivier Hainque <hainque@adacore.com>
17436 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
17437 for all vxworks7 targets.
17438 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
17439 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
17440 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
17441 variations for VX6/VX7 and 32/64bits later on in ...
17442 (VXWORKS_LIB_SPEC): Leverage new macros.
17443 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
17444 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
17446 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
17448 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
17449 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
17451 2017-06-26 Carl Love <cel@us.ibm.com>
17453 * config/rs6000/rs6000-c.c: Add support for built-in functions
17454 vector bool char vec_reve (vector bool char);
17455 vector signed char vec_reve (vector signed char);
17456 vector unsigned char vec_reve (vector unsigned char);
17457 vector bool int vec_reve (vector bool int);
17458 vector signed int vec_reve (vector signed int);
17459 vector unsigned int vec_reve (vector unsigned int);
17460 vector bool long long vec_reve (vector bool long long);
17461 vector signed long long vec_reve (vector signed long long);
17462 vector unsigned long long vec_reve (vector unsigned long long);
17463 vector bool short vec_reve (vector bool short);
17464 vector signed short vec_reve (vector signed short);
17465 vector double vec_reve (vector double);
17466 vector float vec_reve (vector float);
17467 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
17468 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
17469 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
17470 (altivec_vreve): New pattern.
17471 * config/rs6000/altivec.h (vec_reve): New define.
17472 * doc/extend.texi (vec_rev): Update the built-in documentation file
17473 for the new built-in functions.
17475 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17477 PR tree-optimization/71815
17478 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
17480 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
17482 (slsr_process_phi): Likewise.
17483 (replace_uncond_cands_and_profitable_phis): Don't replace a
17484 multiply candidate with a stride of 1 (copy or cast).
17485 (phi_incr_cost): Call uses_consumed_by_stmt rather than
17487 (lowest_cost_path): Likewise.
17488 (total_savings): Likewise.
17490 2017-06-26 Richard Biener <rguenther@suse.de>
17493 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
17494 Use def_builtin_pure for all gather builtins.
17496 2017-06-26 Richard Biener <rguenther@suse.de>
17498 PR tree-optimization/81203
17499 * tree-tailcall.c (find_tail_calls): Do not move stmts into
17500 non-dominating BBs.
17502 2017-06-26 Marek Polacek <polacek@redhat.com>
17505 * doc/invoke.texi: Document -Wmultistatement-macros.
17507 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
17509 * doc/sourcebuild.texi (ARM-specific attributes): Document new
17510 arm_neon_ok_no_float_abi effective target.
17512 2017-06-26 Richard Biener <rguenther@suse.de>
17514 PR tree-optimization/80928
17515 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
17516 (copy_bbs): Set BB_DUPLICATED flag early.
17517 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
17519 (execute_on_shrinking_pred): Likewise.
17520 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
17521 BB_DUPLICATED blocks.
17522 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
17523 iterate over all PHIs considering removal of *gsi.
17525 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
17527 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
17530 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17532 * config/rs6000/rs6000-string.c: (expand_block_clear,
17533 do_load_for_compare, select_block_compare_mode,
17534 compute_current_alignment, expand_block_compare,
17535 expand_strncmp_align_check, expand_strn_compare,
17536 expand_block_move, rs6000_output_load_multiple)
17537 Move functions related to string/block move/compare
17538 to a separate file.
17539 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
17540 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
17541 for this function which is now used in two files.
17542 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
17543 * config.gcc: Add rs6000-string.o to extra_objs for
17544 targets powerpc*-*-* and rs6000*-*-*.
17546 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
17549 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
17550 32-bit, since indexed is not valid for DImode.
17551 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
17552 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
17553 (define_peephole2 for Altivec d-form load): Add 32-bit support.
17554 (define_peephole2 for Altivec d-form store): Likewise.
17557 * multiple_target.c (create_dispatcher_calls): Only create the
17558 dispatcher call if the function is the default clone of a
17559 versioned function.
17561 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
17563 PR middle-end/80902
17564 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
17565 a call, force the call to not be a tail call.
17567 2017-06-23 Jeff Law <law@redhat.com>
17569 * doc/contrib.texi: Add entry for Steven Pemberton's work on
17572 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
17574 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
17575 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
17576 handling for early expansion of vector shifts (sl,sr,sra,rl).
17577 (builtin_function_type): Add vector shift right instructions
17578 to the unsigned argument list.
17580 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
17582 rtl-optimizatoin/79286
17583 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
17584 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
17585 trap. PIC register plus a const unspec without offset can never trap.
17587 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
17589 * tree.h (builtin_structptr_type): New type.
17590 (builtin_structptr_types): Declare new array.
17591 * tree.c (builtin_structptr_types): New array.
17592 (free_lang_data, build_common_tree_nodes): Use it.
17594 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
17597 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
17600 2017-06-22 Matt Turner <mattst88@gmail.com>
17602 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
17603 Lake models to skylake case. Assume skylake for unknown
17604 models with clflushopt.
17606 2017-06-22 Jeff Law <law@redhat.com>
17608 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
17609 frame sizes that do not satisfy aarch64_uimm12_shift.
17611 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
17613 * profile-count.h (apply_probability,
17614 apply_scale, probability_in): Fix checks for zero.
17616 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17618 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
17619 * doc/cppdiropts.texi (-I @var{dir}): Document it.
17621 2016-06-22 Richard Biener <rguenther@suse.de>
17623 * tree-vect-loop.c (vect_model_reduction_cost): Handle
17624 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
17625 REDUC_MAX_EXPR support.
17626 (vectorizable_reduction): Likewise.
17627 (vect_create_epilog_for_reduction): Likewise.
17629 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
17631 * match.pd (A / (1 << B) -> A >> B): New.
17632 * generic-match-head.c: Include optabs-tree.h.
17633 * gimple-match-head.c: Likewise.
17634 * optabs-tree.h (target_supports_op_p): New.
17635 * optabs-tree.c (target_supports_op_p): New.
17637 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17639 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
17640 $gcc_cv_ld --help output.
17641 (gcc_cv_ld_demangle): Likewise.
17642 (gcc_cv_ld_eh_frame_hdr): Likewise.
17643 (gcc_cv_ld_pie): Likewise.
17644 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
17645 (gcc_cv_ld_buildid): Likewise.
17646 (gcc_cv_ld_sysroot): Likewise.
17647 (ld_bndplt_support): Likewise.
17648 (ld_pushpopstate_support): Likewise.
17649 * configure: Regenerate.
17650 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
17652 2017-06-21 Jakub Jelinek <jakub@redhat.com>
17655 * config/i386/sse.md (round<mode>2): Renumber match_dup and
17656 operands indexes to avoid gap between operands and match_dups.
17658 2017-06-21 Andrew Pinski <apinski@cavium.com>
17660 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
17661 Increment Arith_shift and Arith_shift_reg by 1.
17662 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
17664 * config/aarch64/aarch64.c (thunderx_tunings): Enable
17665 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
17666 (aarch64_strip_extend): Add new argument and test for it.
17667 (aarch64_cheap_mult_shift_p): New function.
17668 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
17669 add a cost if it is true.
17670 Update calls to aarch64_strip_extend.
17671 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
17673 2017-06-21 Andrew Pinski <apinski@cavium.com>
17675 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
17677 (thunderxt88): Likewise.
17678 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
17679 (thunderx_prefetch_tune): New variable.
17680 (thunderx2t99_prefetch_tune): Update for the correct values.
17681 (thunderxt88_tunings): New variable.
17682 (thunderx_tunings): Use thunderx_prefetch_tune instead of
17683 generic_prefetch_tune.
17684 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
17686 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17688 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
17689 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
17690 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
17691 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
17692 (aarch64_atomic_cas<mode>, GPI): Likewise.
17694 2017-06-21 Martin Liska <mliska@suse.cz>
17696 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
17697 statements on cold and hot labels.
17698 * predict.c (tree_estimate_probability_bb): Remove the
17699 prediction from this place.
17701 2017-06-21 Martin Liska <mliska@suse.cz>
17703 PR tree-optimization/79489
17704 * gimplify.c (maybe_add_early_return_predict_stmt): New
17706 (gimplify_return_expr): Call the function.
17707 * predict.c (tree_estimate_probability_bb): Remove handling
17709 * predict.def: Update comment about early return predictor.
17710 * gimple-predict.h (is_gimple_predict): New function.
17711 * predict.def: Change default value of early return to 66.
17712 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
17714 * passes.def: Put pass_strip_predict_hints to the beginning of
17717 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
17719 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
17720 FUNCTION_DECL declarations.
17721 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
17723 (dwaf2out_decl): Likewise.
17724 * godump.c (go_early_global_decl): Skip call to the real debug hook
17725 for FUNCTION_DECL declarations.
17726 * passes.c (rest_of_decl_compilation): Skip call to the
17727 early_global_decl debug hook for FUNCTION_DECL declarations, unless
17728 -fdump-go-spec is passed.
17730 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
17732 * config/i386/i386.c (struct builtin_isa): New field pure_p.
17733 Reorder for compactness.
17734 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
17735 (def_builtin_pure, def_builtin_pure2): New functions.
17736 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
17738 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
17740 * match.pd (nop_convert): New predicate.
17741 ((A +- CST1) +- CST2): Allow some NOP conversions.
17743 2017-06-21 Jakub Jelinek <jakub@redhat.com>
17746 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
17747 with ctors/dtors if GOVD_SHARED is set.
17749 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17751 * config/aarch64/aarch64.md (movti_aarch64):
17752 Emit mov rather than orr.
17753 (movtf_aarch64): Likewise.
17754 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
17755 Emit mov rather than orr.
17757 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17759 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
17760 Swap alternatives, make integer dup more expensive.
17762 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17764 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
17765 Return true for non-tls symbols.
17767 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
17769 * config/aarch64/aarch64-cores.def (cortex-a55): New.
17770 (cortex-a75): Likewise.
17771 (cortex-a75.cortex-a55): Likewise.
17772 * config/aarch64/aarch64-tune.md: Regenerate.
17773 * doc/invoke.texi (-mtune): Document new values for -mtune.
17775 2017-06-21 Tom de Vries <tom@codesourcery.com>
17777 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
17778 stack_size feature.
17779 (Effective-Target Keywords, Other attributes): Suggest using
17780 dg-add-options stack_size feature to get stack limit in stack_size
17781 effective target documentation.
17783 2017-06-21 Julian Brown <julian@codesourcery.com>
17784 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17786 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
17787 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
17788 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
17790 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
17791 attribute type list for neon_multiply.
17792 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
17793 attribute type list for neon_multiply.
17794 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
17795 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
17796 attribute type list for neon_multiply.
17797 * config/arm/types.md (crypto_pmull): Add.
17798 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
17799 attribute type list.
17801 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
17803 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
17806 2017-06-20 Jakub Jelinek <jakub@redhat.com>
17808 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
17809 to make sure not to dereference a NULL cost_classes_ptr pointer.
17811 2017-06-20 Carl Love <cel@us.ibm.com>
17813 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17814 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17815 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17816 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17817 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17818 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17819 VMULOSW): New enum "unspec" values.
17820 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17821 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17822 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17823 altivec_vmulosw): New patterns.
17824 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17825 VMULOSW): Add definitions.
17827 2017-06-20 Julia Koval <julia.koval@intel.com>
17829 * config/i386/i386.c: Fix rounding expand for new pattern.
17830 * config/i386/subst.md: Fix pattern (parallel -> unspec).
17832 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17834 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
17835 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
17837 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17839 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
17842 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17844 * config/aarch64/aarch64-cores.def: Rearrange to sort by
17845 architecture, then by implementer ID.
17846 * config/aarch64/aarch64-tune.md: Regenerate.
17848 2017-06-20 Richard Biener <rguenther@suse.de>
17850 PR middle-end/81097
17851 * fold-const.c (split_tree): Fold to type before negating.
17853 2017-06-20 David Malcolm <dmalcolm@redhat.com>
17855 * diagnostic-show-locus.c
17856 (selftest::test_fixit_deletion_affecting_newline): New function.
17857 (selftest::diagnostic_show_locus_c_tests): Call it.
17859 2017-06-20 Andreas Schwab <schwab@suse.de>
17862 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
17865 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17867 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
17868 __ARM_FEATURE_COPROC according to support.
17870 2017-06-20 Jakub Jelinek <jakub@redhat.com>
17872 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17873 Rewritten to avoid overflow for > 32-bit pointers.
17876 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
17877 by removing enum keyword.
17878 (ubsan_type_descriptor): Likewise. Formatting fix.
17881 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
17882 splitter): Require TARGET_SSE2 in the condition.
17884 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
17887 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
17888 for doing vector set of SFmode on ISA 3.0.
17889 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
17890 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
17892 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
17893 SFmode value into a V4SF variable that was extracted from another
17894 V4SF variable without converting the element to double precision
17895 and back to single precision vector format.
17896 (vsx_insert_extract_v4sf_p9_2): Likewise.
17898 2017-06-19 Jakub Jelinek <jakub@redhat.com>
17900 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
17901 in UWHI to avoid undefined overflow.
17904 * ubsan.h (enum ubsan_encode_value_phase): New.
17905 (ubsan_encode_value): Change second argument to
17906 enum ubsan_encode_value_phase with default value of
17907 UBSAN_ENCODE_VALUE_GENERIC.
17908 * ubsan.c (ubsan_encode_value): Change second argument to
17909 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
17910 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
17911 create_tmp_var_raw instead of create_tmp_var and use a
17913 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
17914 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
17915 ubsan_encode_value callers.
17918 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
17919 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
17920 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
17922 2017-06-19 Richard Biener <rguenther@suse.de>
17924 PR middle-end/81118
17925 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
17926 estimates if we changed anything.
17928 2017-06-19 Richard Biener <rguenther@suse.de>
17930 PR tree-optimization/80887
17931 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
17932 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
17933 simplified lookups, then reset mprts_hook.
17934 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
17936 (try_to_simplify): Likewise.
17938 2017-06-19 Martin Liska <mliska@suse.cz>
17941 * gimplify.c (gimplify_switch_expr):
17942 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
17944 2017-06-19 Martin Liska <mliska@suse.cz>
17946 * doc/install.texi: Document that PGO runs in 4 stages.
17948 2017-06-19 Martin Liska <mliska@suse.cz>
17951 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
17952 to dispatcher function name.
17953 * multiple_target.c (replace_function_decl): New function.
17954 (create_dispatcher_calls): Redirect both edges and references.
17956 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
17958 * profile-count.c (profile_count::dump): Dump quality.
17959 (profile_count::differs_from_p): Update for unsigned val.
17960 * profile-count.h (profile_count_quality): New enum.
17961 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
17963 2017-06-19 Richard Biener <rguenther@suse.de>
17965 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
17966 struct function as arg.
17967 (estimate_numbers_of_iterations): Export overload with loop arg.
17968 (free_numbers_of_iterations_estimates_loop): Use an overload of
17969 free_numbers_of_iterations_estimates instead.
17970 * tree-cfg.c (remove_bb): Adjust.
17971 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
17972 * tree-parloops.c (gen_parallel_loop): Likewise.
17973 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
17975 (tree_unroll_loops_completely): Likewise.
17976 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
17977 Use an overload instead and export.
17978 (estimated_loop_iterations): Adjust.
17979 (max_loop_iterations): Likewise.
17980 (likely_max_loop_iterations): Likewise.
17981 (estimate_numbers_of_iterations): Take struct function as arg
17983 (loop_exits_before_overflow): Adjust.
17984 (free_numbers_of_iterations_estimates_loop): Use an overload.
17985 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
17986 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
17988 2017-06-19 Richard Biener <rguenther@suse.de>
17991 * ipa-prop.c (find_constructor_constant_at_offset): Handle
17992 RANGE_EXPR conservatively.
17994 2017-06-16 Carl Love <cel@us.ibm.com>
17996 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17997 definitions for vec_float, vec_float2, vec_floato,
17998 vec_floate built-ins.
17999 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
18000 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
18002 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
18003 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
18004 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
18005 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
18006 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
18007 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
18008 vec_floato): Add builtin defines.
18009 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
18010 Update the built-in documentation file for the new built-in
18013 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18015 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
18016 (mthumb): Mark as the negative of -marm.
18018 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18020 * doc/invoke.texi (ARM Options, -mcpu): Document supported
18022 (ARM Options, -mtune): Document that this accepts the same
18023 extension options as -mcpu.
18024 (ARM Options, -mfpu): Document addition of -mfpu=auto.
18026 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18028 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
18029 permitted extensions.
18031 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18033 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
18034 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
18035 (armv8-m.main): Add option +nodsp.
18036 * config/arm/arm-cpu-cdata.h: Regenerated.
18038 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18040 * config/arm/t-fuchsia: New file.
18041 * config.gcc (arm*-*-fuchsia*): Use it.
18043 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18045 * config/arm/t-symbian: Rewrite for new option infrastructure.
18047 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18049 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
18050 (MULTILIB_REQUIRED): Likewise.
18052 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18054 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
18055 (MULTILIB_RESUE): Likewise.
18056 (MULTILIB_MATCHES): Likewise.
18057 (MULTLIB_REQUIRED): Likewise.
18059 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18061 * config/arm/t-rtems: Rewrite for new option framework.
18063 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18065 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
18066 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
18067 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
18068 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
18069 * config/arm/t-multilib: ... here.
18070 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
18071 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
18072 armv7-a and armv8*-a when A-profile libraries have not been built.
18073 * config/arm/t-rmprofile: Rewrite.
18075 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18077 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
18078 with a backslash. Remove the backslash after substituting unescaped
18080 * doc/fragments.texi (MULTILIB_REUSE): Document it.
18082 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18084 * config.gcc: (arm*-*-*): When building a-profile libraries, force
18085 the driver to pass through the default setting of -mfloat-abi.
18086 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
18088 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
18089 (all_feat_combs): New rule.
18090 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
18092 * config/arm/t-aprofile: Rewrite.
18094 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18096 * config/arm/arm.h (FPUTYPE_AUTO): Define.
18097 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
18098 fpu is not specified by the user/command-line.
18099 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
18100 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
18101 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
18102 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
18103 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
18104 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
18106 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18108 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
18109 * config/arm/t-arm-elf: Rewritten.
18111 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18113 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
18114 have some floating-point instructions.
18115 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
18116 (TARGET_MAYBE_HARD_FLOAT): New macro.
18117 * config/arm/arm-builtins.c (arm_init_builtins): Use
18118 TARGET_MAYBE_HARD_FLOAT.
18119 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
18121 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18123 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
18124 (configargs.h): Include it.
18125 (arm_print_hint_for_fpu_option): New function.
18126 (arm_parse_fpu_option): New function.
18127 (candidate_extension): New class.
18128 (arm_canon_for_multilib): New function.
18129 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
18130 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
18131 (ARCH_CANONICAL_SPECS): New macro.
18132 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
18134 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18136 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
18137 are set after handling multilib fragments. Set target_cpu_default2
18140 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18142 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
18144 (arm*-*-*): Set target_cpu_default2 to a quoted string.
18145 * config/arm/parsecpu.awk (check_cpu): Validate any extension
18147 (check_arch): Likewise.
18148 * config/arm/arm.c (arm_configure_build_target): Handle
18149 TARGET_CPU_DEFAULT being a string constant. Scan any feature
18150 options in the default.
18152 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18154 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
18155 when an option is an alias of another.
18156 * config/arm/parsecpu.awk (optalias): New parser token.
18157 (gen_comm_data): Mark non-alias options as such. Emit entries
18158 for extension aliases.
18159 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
18160 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
18161 (armv6kz, armv6zk, armv6t2): Likewise.
18162 (armv7): Make vfpv3-d16 an alias.
18163 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
18165 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
18166 Sort in canonical order.
18167 (armv8-a): Sort in canonical order.
18168 (armv8.1-a, armv8.2-a): Likewise.
18169 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
18171 (cortex-a9): Sort in canonical order.
18172 * config/arm/arm.c (selftests.h): Include it.
18173 (arm_test_cpu_arch_data): New function.
18174 (arm_run_self_tests): New function.
18175 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
18176 (targetm): Move declaration to the end of the file.
18177 * arm-cpu-cdata.h: Regenerated.
18179 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18181 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
18182 call to target_mode_check describing the type of option passed.
18183 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
18184 (arm_target_thumb_only): Use arm_parse_arch_option_name or
18185 arm_parse_cpu_option_name to match parameters against list of
18187 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
18188 arm_arch_core_flags data structure.
18189 * config/arm/arm-cpu_cdata.h: Regenerated.
18191 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18193 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
18195 (arm_print_hint_for_cpu_option): Likewise.
18196 (arm_print_hint_for_arch_option): Likewise.
18197 (arm_parse_cpu_option_name): Likewise.
18198 (arm_parse_arch_option_name): Likewise.
18199 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
18200 of entries in the all_fpus list.
18201 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
18202 (arm_parse_cpu_option_name): Declare.
18203 (arm_parse_arch_option_name): Declare.
18204 (arm_parse_option_features): Declare.
18205 (arm_intialize_isa): Declare.
18206 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
18208 (gen_comm_data): ... here. Make definitions non-static.
18209 * config/arm/arm-cpu-data.h: Regenerated.
18210 * config/arm/arm-cpu-cdata.h: Regenerated.
18212 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18214 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
18215 (cpu_arch_extension): New structure.
18216 (cpu_arch_option, arch_option, cpu_option): New structures.
18217 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
18218 architecture types.
18219 (gen_data): Generate new format data tables.
18220 * config/arm/arm.c (cpu_tune): New structure.
18221 (cpu_option, processors): Delete.
18222 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
18223 (arm_print_hint_for_cpu_option): ... this and ...
18224 (arm_print_hint_for_arch_option): ... this.
18225 (arm_parse_arch_cpu_name): Delete. Replace with ...
18226 (arm_parse_cpu_option_name): ... this and ...
18227 (arm_parse_arch_option_name): ... this.
18228 (arm_unrecognized_feature): Change type of target parameter to
18230 (arm_parse_arch_cpu_features): Delete. Replace with ...
18231 (arm_parse_option_features): ... this.
18232 (arm_configure_build_target): Rework to use new configuration data
18234 (arm_print_tune_info): Rework for new configuration data tables.
18235 * config/arm/arm-cpu-data.h: Regenerated.
18236 * config/arm/arm-cpu.h: Regenerated.
18238 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18240 * Makefile.in (OBJS): Move sbitmap.o from here ...
18241 (OBJS-libcommon): ... to here.
18243 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18245 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
18246 (ISA_ALL_CRYPTO): New macro.
18247 (ISA_ALL_SIMD): New macro
18248 (ISA_ALL_FP): New macro.
18249 * config/arm/arm.c (fpu_bitlist): Update initializer.
18250 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
18252 (arm9e): Add fpu. Add option for nofp
18253 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
18254 (arm926ej-s, arm1026ej-s): Likewise.
18255 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
18256 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
18257 neon-fp16, neon-vfpv4, nofp and nosimd.
18258 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
18259 (cortex-a8): Add fpu. Add option for nofp.
18260 (cortex-a9): Add fpu. Add options for nosimd and nofp.
18261 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
18262 (cortex-r4f): Add fpu.
18263 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
18264 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
18266 (cortex-r8): Likewise.
18267 (cortex-m4): Add fpu. Add option for nofp.
18268 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
18269 (cortex-a17.cortex-a7): Likewise.
18270 (cortex-a32): Add fpu. Add options for crypto and nofp.
18271 (cortex-a35, cortex-a53): Likewise.
18272 (cortex-a57): Add fpu. Add option for crypto.
18273 (cortex-a72, cortex-a73): Likewise.
18274 (exynos-m1): Likewise.
18275 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
18276 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
18277 (cortex-m33): Add fpu. Add option for nofp.
18278 * config/arm/arm-cpu-cdata.h: Regenerated
18279 * config/arm/arm-cpu-data.h: Regenerated.
18281 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18283 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
18284 (armv5te, armv5tej): Likewise.
18285 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
18286 (armv7): Add options fp and vfpv3-d16.
18287 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
18288 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
18290 (armv7ve): Likewise.
18291 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
18292 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
18293 (armv8-a): Add nocrypto option.
18294 (armv8.1-a, armv8.2-a): Likewise.
18295 (armv8-m.main): add options fp, fp.dp and nofp.
18297 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18299 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
18301 (armv8-a+crc): Delete.
18302 (armv8.1-a): Add options simd, crypto and nofp.
18303 (armv8.2-a): Add options fp16, simd, crypto and nofp.
18304 (armv8.2-a+fp16): Delete.
18305 (armv8-m.main): Add option dsp.
18306 (armv8-m.main+dsp): Delete.
18307 (cortex-a8): Add fpu. Add nofp option.
18308 (cortex-a9): Add fpu. Add nofp and nosimd options.
18309 * config/arm/parsecpu.awk (gen_data): Generate option tables and
18310 link to main cpu and architecture data structures.
18311 (gen_comm_data): Only put isa attributes from the main architecture
18313 (option): New statement for architecture and CPU entries.
18314 * arm.c (struct cpu_option): New structure.
18315 (struct processors): Add entry for options.
18316 (arm_unrecognized_feature): New function.
18317 (arm_parse_arch_cpu_name): Ignore any characters after the first
18319 (arm_parse_arch_cpu_feature): New function.
18320 (arm_configure_build_target): Separate out any CPU and architecture
18321 features and parse separately. Don't error out if -mfpu=auto is
18322 used with only an architecture string.
18323 (arm_print_asm_arch_directives): New function.
18324 (arm_file_start): Call it.
18325 * config/arm/arm-cpu-cdata.h: Regenerated.
18326 * config/arm/arm-cpu-data.h: Likewise.
18327 * config/arm/arm-tables.opt: Likewise.
18329 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18331 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
18332 assembler when it is not -mfpu=auto.
18334 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18336 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
18337 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
18338 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
18339 (ASM_CPU_SPEC): Rewrite.
18340 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
18341 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
18342 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
18343 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
18344 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
18345 copied string is NUL-terminated. Also strip any characters prefixed
18347 (arm_rewrite_selected_arch): New function.
18348 (arm_rewrite_march): New function.
18350 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18352 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
18353 (x_arm_cpu_string, x_arm_tune_string): Likewise.
18354 (march, mcpu, mtune): Convert to string-based options.
18355 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
18356 (arm_parse_arch_cpu_name): New function.
18357 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
18358 identify selected architecture or CPU.
18359 (arm_option_save): New function.
18360 (TARGET_OPTION_SAVE): Redefine.
18361 (arm_option_restore): Restore string options.
18362 (arm_option_print): Print string options.
18364 2017-06-16 Martin Sebor <msebor@redhat.com>
18366 PR tree-optimization/80933
18367 PR tree-optimization/80934
18368 * builtins.c (fold_builtin_3): Do not handle bcmp here.
18369 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
18370 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
18371 (gimple_fold_builtin): Call them.
18373 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18375 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
18376 as unlikely; update profile.
18378 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18380 * predict.c (force_edge_cold): Handle declaring edges impossible
18383 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18385 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
18387 (try_unroll_loop_completely): Fix reporting.
18389 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18391 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
18393 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
18396 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
18397 if given a non-constant argument for an intrinsic which requires a
18400 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18402 * profile.c (compare_freqs): New function.
18403 (branch_prob): Sort edge list.
18404 (find_spanning_tree): Assume that the list is priority sorted.
18406 2017-06-16 Richard Biener <rguenther@suse.de>
18408 PR tree-optimization/81090
18409 * passes.def (pass_record_bounds): Remove.
18410 * tree-pass.h (make_pass_record_bounds): Likewise.
18411 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
18412 make_pass_record_bounds): Likewise.
18413 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
18414 not free niter estimates at the beginning but at the end.
18415 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
18417 2017-06-16 Richard Biener <rguenther@suse.de>
18419 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
18420 initializer to workaround ICE in host GCC 4.8.
18422 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18424 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
18426 (clone_inlined_nodes): Update.
18428 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18430 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
18431 prefetch settings, and enable prefetching by default at -O3.
18433 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18435 * config/aarch64/aarch64.c (aarch64_override_options_internal):
18436 Set flag_prefetch_loop_arrays according to tuning data.
18438 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18440 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
18441 New tune structure.
18442 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
18443 [Unrelated to main purpose of the patch] Place the pointer field last
18444 to enable type checking errors when tune structure are wrongly merged.
18445 * config/aarch64/aarch64.c (generic_prefetch_tune,)
18446 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
18447 (thunderx2t99_prefetch_tune): New tune constants.
18448 (tune_params *_tunings): Update all tunings (no functional change).
18449 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
18450 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
18451 from tunings structures.
18453 2017-06-16 Jakub Jelinek <jakub@redhat.com>
18456 * ubsan.c (instrument_null): Add T argument, use it instead
18457 of computing it based on IS_LHS.
18458 (instrument_object_size): Likewise.
18459 (pass_ubsan::execute): Adjust instrument_null and
18460 instrument_object_size callers to pass gimple_get_lhs or
18461 gimple_assign_rhs1 result to it. Use instrument_null instead of
18462 calling get_base_address and instrument_mem_ref. Handle
18463 aggregate call arguments for object-size sanitization.
18465 2017-06-16 Yury Gribov <tetra2005@gmail.com>
18467 PR tree-optimization/81089
18468 * tree-vrp.c (is_masked_range_test): Validate operands of
18471 2017-06-15 Martin Sebor <msebor@redhat.com>
18474 * dumpfile.c (dump_register): Avoid calling memset to initialize
18475 a class with a default ctor.
18476 * gcc.c (struct compiler): Remove const qualification.
18477 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
18478 * hash-table.h: Ditto.
18479 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
18481 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
18482 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
18484 * params.h (struct param_info): Make struct members non-const.
18485 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
18486 with default initialization.
18487 * vec.h (vec_copy_construct, vec_default_construct): New helper
18489 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
18490 with vec_copy_construct.
18491 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
18492 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
18493 * doc/invoke.texi (-Wclass-memaccess): Document.
18495 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18497 * emit-rtl.h (is_leaf): Update comment about local
18498 register allocator.
18500 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18503 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
18504 for a variable to have a section before checking if the section has a
18506 Set section to.persistent if persistent attribute is set.
18507 Warn if .persistent attribute is used on an automatic variable.
18509 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
18511 PR rtl-optimization/80474
18512 * reorg.c (update_block): Do not ignore instructions in a delay slot.
18514 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
18516 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
18519 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
18521 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
18522 (casesi): Emit bounds checking as RTL.
18523 (casesi_internal_mips16_<mode>): Remove bounds checking.
18525 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
18527 * config/xtensa/xtensa.c (xtensa_option_override): Append
18528 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
18529 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
18530 xtensa_doloop_hooks): Define unconditionally.
18531 (xtensa_reorg_loops): Only call reorg_loops in the presence of
18533 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
18534 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
18535 for it in xtensa_option_override.
18536 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
18537 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
18539 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
18541 * doc/cppopts.texi: Document '-' special value to -MF.
18543 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
18545 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
18546 (cortex_a53_fconst): Likewise.
18547 (cortex_a53_fpmul): Likewise.
18548 (cortex_a53_f_load_64): Likewise.
18549 (cortex_a53_f_load_many): Likewise.
18550 (cortex_a53_advsimd_alu): Likewise.
18551 (cortex_a53_advsimd_alu_q): Likewise.
18552 (cortex_a53_advsimd_mul): Likewise.
18553 (cortex_a53_advsimd_mul_q): Likewise.
18554 (fpmac bypass): Add new bypass for fpmac-fpmac case.
18555 Add missing fmul, r2f_cvt and fconst cases.
18557 2017-06-14 Richard Biener <rguenther@suse.de>
18559 PR middle-end/81088
18560 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
18562 (fold_binary_loc): When associating do not treat pre-existing
18563 TREE_OVERFLOW on literal constants as a reason to allow
18564 TREE_OVERFLOW on associated literal constants.
18566 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18568 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
18569 (MASK_FEATURES): New macro.
18570 * config/sparc/sparc.c (sparc_option_override): Remove the special
18571 handling of -mfpu and generalize it to all MASK_FEATURES switches.
18573 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18575 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
18576 a division of 0 if non-call exceptions are enabled.
18578 2017-06-14 Andrew Pinski <apinski@cavium.com>
18579 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18582 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
18583 Improve vector initialization code gen for only variable case.
18585 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18587 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
18589 2017-06-14 Richard Biener <rguenther@suse.de>
18591 PR tree-optimization/81083
18592 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
18595 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18597 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
18598 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
18599 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
18600 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
18601 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
18602 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
18604 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18606 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
18607 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
18609 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18611 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
18613 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18615 * config/rs6000/t-rtems: Don't handle SPE.
18617 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18619 * config/rs6000/t-linux: Don't handle SPE.
18621 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18623 * config/rs6000/eabispe.h: Delete file.
18625 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18627 * config/rs6000/t-spe: Delete file.
18629 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18631 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
18632 (rs6000_legitimate_offset_address_p): Return false for anything in
18633 V2SImode or V2SFmode.
18635 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18637 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
18638 except V2SF and V2SI. Rearrange the vector modes, and add comments.
18639 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
18641 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
18642 (rs6000_legitimate_offset_address_p): Ditto.
18643 (rs6000_emit_move): Ditto.
18644 (rs6000_init_builtins): Remove V4HI_type_node.
18646 2017-06-13 Martin Liska <mliska@suse.cz>
18649 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
18650 (gate_asan): Likewise.
18651 * asan.h (asan_no_sanitize_address_p): Remove the function.
18652 (sanitize_flags_p): New function.
18653 * builtins.def: Fix coding style.
18654 * common.opt: Use renamed enum value.
18655 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
18656 * doc/extend.texi: Document no_sanitize attribute.
18657 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
18658 to SANITIZE_UNDEFINED_NONDEFAULT.
18659 * gcc.c (sanitize_spec_function): Use the renamed enum value.
18660 * gimple-fold.c (optimize_atomic_compare_exchange_p):
18661 Use sanitize_flags_p.
18662 * gimplify.c (gimplify_function_tree): Likewise.
18663 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
18664 * opts.c (parse_no_sanitize_attribute): New function.
18665 (common_handle_option): Use renamed enum value.
18666 * opts.h (parse_no_sanitize_attribute): Declare.
18667 * tree.c (sanitize_flags_p): New function.
18668 * tree.h: Declared here.
18669 * tsan.c: Use sanitize_flags_p.
18670 * ubsan.c (ubsan_expand_null_ifn): Likewise.
18671 (instrument_mem_ref): Likewise.
18672 (instrument_bool_enum_load): Likewise.
18673 (do_ubsan_in_current_function): Remove the function.
18674 (pass_ubsan::execute): Use sanitize_flags_p.
18675 * ubsan.h: Remove do_ubsan_in_current_function
18676 * tree-cfg.c (print_no_sanitize_attr_value): New function.
18677 (dump_function_to_file): Use it here.
18679 2017-06-13 Martin Jambor <mjambor@suse.cz>
18681 PR tree-optimization/80803
18682 PR tree-optimization/81063
18683 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
18684 (propagate_subaccesses_across_link): Enqueue subtree whenever
18685 necessary instead of relying on the caller.
18687 2017-06-13 Martin Jambor <mjambor@suse.cz>
18689 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
18690 that have a first_link.
18691 (sort_and_splice_var_accesses): Do not check first_link before
18693 (subtree_mark_written_and_enqueue): Likewise.
18694 (propagate_all_subaccesses): Likewise and do not stop at first
18695 parent with a first_link.
18697 2017-06-13 Martin Jambor <mjambor@suse.cz>
18699 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
18702 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18704 * match.pd: New pattern.
18706 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18708 * tree-vrp.c (is_masked_range_test): New function.
18709 (register_edge_assert_for): Determine ranges for
18712 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18714 PR tree-optimization/67328
18715 * fold-const.c (maskable_range_p): New function.
18716 (build_range_check): Generate bittests if possible.
18718 2017-06-13 Martin Liska <mliska@suse.cz>
18720 * gimple-pretty-print.c (dump_probability): Add new argument.
18721 (dump_edge_probability): Dump both probability and count.
18722 (dump_gimple_label): Likewise.
18723 (dump_gimple_bb_header): Likewise.
18725 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
18728 * config/avr/avr-devices.c: Fix indentation.
18729 * config/avr/gen-avr-mmcu-specs.c: Dito.
18731 2017-06-13 Richard Biener <rguenther@suse.de>
18733 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
18734 instead get vector type from stmt_info.
18735 (vectorizable_reduction): Adjust. Remove dead code.
18737 2017-06-13 Richard Biener <rguenther@suse.de>
18739 PR middle-end/81065
18740 * fold-const.c (extract_muldiv_1): Remove bogus distribution
18741 case of C * (x * C2 + C3).
18742 (fold_addr_of_array_ref_difference): Properly fold index difference.
18744 2017-06-12 David S. Miller <davem@davemloft.net>
18747 * config/sparc/sparc.md (return expander): Emit frame blockage if
18748 function uses alloca.
18750 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
18752 * combine.c (make_field_assignment): Check len rather than the mode
18753 precision when calling force_to_mode.
18755 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
18757 Support multilibs and devices that see flash in RAM address range.
18760 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
18761 (avr_mcu_t) <flash_pm_offset>: New field.
18762 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
18763 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
18764 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
18765 (AVR_TINY_PM_OFFSET): Remove macro.
18766 * config/avr/avr.opt (-mshort-calls): New option.
18767 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18768 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
18769 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
18770 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
18771 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
18772 instead of avr_arch->have_jmp_call.
18773 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
18774 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
18775 avr_arch->flash_pm_offset to define.
18776 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
18777 new field flash_pm_offset. Add entry for avrxmega3.
18778 (avr_texinfo): Add entry for avrxmega3.
18779 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
18780 attiny212, attiny214,
18781 attiny412, attiny414, attiny416, attiny417,
18782 attiny814, attiny816, attiny817,
18783 attiny1614, attiny1616, attiny1617,
18784 attiny3214, attiny3216, attiny3217.
18785 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
18786 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
18787 (avr_print_operand_address) [AVR_TINY]: Same.
18788 (avr_asm_init_sections) <readonly_data_section>: Only patch
18789 callback if avr_arch->flash_pm_offset = 0.
18790 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
18791 for rodata if avr_arch->flash_pm_offset != 0.
18792 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
18793 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
18794 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
18795 (m_options): Append opt_rcall.
18796 (m_dirnames): Append dir_rcall.
18797 * config/avr/t-multilib: Regenerate.
18799 * configure.ac [target=avr]: Check whether avrxmega3 default
18800 linker description file works as needed.
18801 * configure: Regenerate.
18802 * doc/avr-mmcu.texi: Regenerate.
18803 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
18804 <__AVR_ARCH__>: Document avrxmega3 and 103.
18805 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
18806 <__AVR_SHORT_CALLS__>: Document it.
18807 <__AVR_PM_BASE_ADDRESS__>: Document it.
18808 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
18809 (AVR Variable Attributes) <progmem>: Document this is
18810 not needed for avrxmega3.
18811 (AVR Named Address Spaces) <__flash>: Dito.
18813 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
18815 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
18817 2017-06-12 Doug Rupp <rupp@adacore.com>
18819 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
18820 Append vxworks-stdint.h to the tm_file list.
18821 * config/vxworks-stdint.h: New file.
18823 2017-06-12 Martin Liska <mliska@suse.cz>
18825 PR tree-optimization/81041
18826 * tree-profile.c (gimple_gen_ic_func_profiler):
18827 Create an extra BB in profile-generate
18828 (gimple_gen_time_profiler): Likewise.
18830 2017-06-12 Jakub Jelinek <jakub@redhat.com>
18832 PR tree-optimization/81003
18833 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
18834 (update_range_test): Use it instead of force_gimple_operand_gsi.
18836 2017-06-12 Richard Biener <rguenther@suse.de>
18838 PR tree-optimization/81053
18839 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
18840 with backedge value not defined in loop. Simplify def stmt
18843 2017-06-11 Tom de Vries <tom@codesourcery.com>
18846 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
18848 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
18849 nvptx_cannot_force_const_mem.
18851 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18853 * opts.c (finish_options): Move test for flag_split_stack after
18854 it has been initialized.
18856 2017-06-11 Jason Merrill <jason@redhat.com>
18858 * tree.h (id_equal): New.
18859 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
18860 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
18861 instead of strcmp of IDENTIFIER_POINTER.
18863 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18865 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
18866 (mark_all_inlined_calls_cdtor): Fix formating.
18867 (inline_transform): Rescale profile before inlining.
18869 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18871 * cgraph.h (cgraph_edge::clone): Update prototype.
18872 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
18873 (cgraph_node::create_clone): Update.
18874 (cgraph_node::create_version_clone): Update.
18875 * tree-inline.c (copy_bb): Update.
18876 (expand_call_inline): Update.
18878 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
18880 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
18881 factored out from ...
18882 (rs6000_emit_prologue): ... here.
18884 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
18886 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
18887 factored out from ...
18888 (rs6000_emit_prologue): ... here.
18890 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18892 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
18894 (handle_missing_profiles): Fix computation of tp_first_run.
18895 (counts_to_freqs): Do not touch freqs when count is 0.
18897 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18899 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
18902 2017-06-10 Tom de Vries <tom@codesourcery.com>
18904 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18905 attributes): Document signal effective target.
18907 2017-06-10 Tom de Vries <tom@codesourcery.com>
18909 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18910 Document effective target stack_size.
18912 2017-06-09 David Malcolm <dmalcolm@redhat.com>
18914 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
18915 to the edit_context if they can be auto-applied.
18917 2017-06-9 Ian Lance Taylor <iant@golang.org>
18919 * opts.c (finish_options): If -fsplit-stack, disable implicit
18920 -forder-blocks-and-partition.
18921 * doc/invoke.texi (Optimize Options): Document that when using
18922 -fsplit-stack -forder-blocks-and-partition is not implicitly
18925 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18927 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
18928 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
18929 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
18930 * builtins.def (abort, trap, unreachable): Declare cold.
18931 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
18932 * tree-core.h (ECF_COLD): New.
18933 * tree.c (set_call_expr_flags): Handle ECF_COLD.
18934 (build_common_builtin_nodes): Mark unreachable and abort as cold.
18936 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18938 * predict.c (unlikely_executed_stmt_p): Cleanup.
18940 2017-06-09 Richard Biener <rguenther@suse.de>
18942 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
18943 model if the ref is always written to.
18945 2017-06-09 Tamar Christina <tamar.christina@arm.com>
18947 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
18949 2017-06-09 Tamar Christina <tamar.christina@arm.com>
18951 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
18954 2017-06-09 Tom de Vries <tom@codesourcery.com>
18957 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
18958 "target cannot support label values" when encountering LABEL_REF.
18960 2017-06-09 Martin Liska <mliska@suse.cz>
18962 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
18963 (gimple_gen_ic_func_profiler): Emit direct comparison
18964 of __gcov_indirect_call_callee with NULL.
18965 (gimple_gen_time_profiler): Change probability from
18966 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
18968 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18970 * profile.c (edge_gcov_counts): Turn to pointer.
18971 (compute_branch_probabilities, compute_branch_probabilities): Update.
18972 (branch_prob): Do not clear edge_gcov_count.
18973 * profile.h (edge_gcov_counts): Turn to pointer.
18974 (edge_gcov_count): Update.
18976 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18978 * gimple.h (gimple_check_failed): Mark cold.
18980 2017-06-09 Richard Biener <rguenther@suse.de>
18982 PR tree-optimization/66623
18983 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
18984 refactor check_reduction into two parts, properly computing
18985 whether we have to check reduction validity for outer loop
18988 2017-06-09 Richard Biener <rguenther@suse.de>
18990 PR tree-optimization/79483
18991 * graphite-scop-detection.c (order): New global.
18992 (get_order): Compute bb to order mapping that satisfies code
18993 generation constraints.
18994 (cmp_pbbs): New helper.
18995 (build_scops): Start domwalk at entry block, sort generated
18998 2017-06-09 Richard Biener <rguenther@suse.de>
19000 PR middle-end/81007
19001 * ipa-polymorphic-call.c
19002 (ipa_polymorphic_call_context::restrict_to_inner_class):
19003 Skip FIELD_DECLs with error_mark_node type.
19004 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
19007 2017-06-09 Martin Liska <mliska@suse.cz>
19009 * predict.c (struct branch_predictor): New struct.
19010 (test_prediction_value_range): New test.
19011 (predict_c_tests): New function.
19012 * selftest-run-tests.c (selftest::run_tests): Run the function.
19013 * selftest.h: Declare new tests.
19015 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
19018 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
19019 gen_add3_insn did not fail.
19020 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
19021 r0, construct that number in a temporary reg and add that reg to r0.
19022 If asked to put the result in r0 as well, fail.
19024 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
19026 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19027 for early expansion of vec_eqv.
19029 2017-06-08 Jakub Jelinek <jakub@redhat.com>
19031 PR middle-end/81005
19032 * ubsan.c (instrument_null): Avoid pointless code temporary.
19033 (pass_ubsan::execute): Instrument aggregate arguments of calls.
19035 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
19039 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
19042 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
19045 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19047 * predict.c (unlikely_executed_edge_p): Move ahead.
19048 (probably_never_executed_edge_p): Use it.
19050 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19052 PR middle-end/79988
19053 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
19054 gimple_call_builtin_p call.
19056 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19058 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
19059 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
19060 rtl_check_failed_type2, rtl_check_failed_code1,
19061 rtl_check_failed_code2, rtl_check_failed_code_mode,
19062 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
19063 rtvec_check_failed_bounds, rtl_check_failed_flag,
19064 _fatal_insn_not_found, _fatal_insn): Likewise.
19065 * tree.h (tree_contains_struct_check_failed,
19066 tree_check_failed, tree_not_check_failed,
19067 tree_class_check_failed, tree_range_check_failed,
19068 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
19069 tree_vec_elt_check_failed, phi_node_elt_check_failed,
19070 tree_operand_check_failed, omp_clause_check_failed,
19071 omp_clause_operand_check_failed, omp_clause_range_check_failed):
19074 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19076 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
19077 flag_branch_probabilities.
19078 * ipa-inline.c (edge_badness): Likewise.
19079 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
19080 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
19081 * predict.c (maybe_hot_frequency_p): Likewise.
19082 (probably_never_executed): Likewise.
19083 * sched-ebb.c (schedule_ebbs): Likewise.
19084 * sched-rgn.c (find_single_block_region): Likewise.
19085 * tracer.c (tail_duplicate): Likewise.
19087 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19089 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
19090 longer requires x_flag_profile_use.
19092 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19094 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
19095 instead of flag_reorder_blocks_and_partition.
19096 * dbxout.c (dbxout_function_end): Likewise.
19097 * dwarf2out.c (gen_subprogram_die): Likewise.
19098 * haifa-sched.c (sched_create_recovery_edges): Likewise.
19099 * hw-doloop.c (reorg_loops): Likewise.
19100 * varasm.c (assemble_start_function,
19101 assemble_end_function): Likewise.
19102 (decide_function_section): Do not check for
19103 flag_reorder_blocks_and_partition.
19105 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19107 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19109 (chkp_get_hard_register_fake_addr_expr): Ditto.
19110 (chkp_build_addr_expr): Add check for hard reg case.
19111 (chkp_parse_array_and_component_ref): Ditto.
19112 (chkp_find_bounds_1): Ditto.
19113 (chkp_process_stmt): Don't generate bounds store for
19116 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19118 * predict.c (maybe_hot_bb_p): Do not check profile status.
19119 (maybe_hot_edge_p): Likewise.
19120 (probably_never_executed): Check for zero counts even if profile
19122 (unlikely_executed_edge_p): New function.
19123 (unlikely_executed_stmt_p): New function.
19124 (unlikely_executed_bb_p): New function.
19125 (set_even_probabilities): Use unlikely predicates.
19126 (combine_predictions_for_bb): Likewise.
19127 (predict_paths_for_bb): Likewise.
19128 (predict_paths_leading_to_edge): Likewise.
19129 (determine_unlikely_bbs): New function.
19130 (estimate_bb_frequencies): Use it.
19131 (compute_function_frequency): Use zero counts even if profile is
19133 * profile-count.h: Fix typo.
19135 2017-08-08 Julia Koval <julia.koval@intel.com>
19137 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
19138 _mm512_mask_cvtsepi16_storeu_epi8,
19139 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
19140 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
19141 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
19142 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
19143 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
19144 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
19145 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
19146 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
19147 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
19148 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
19149 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
19150 __builtin_ia32_pmovuswb256mem_mask,
19151 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
19152 __builtin_ia32_pmovwb512mem_mask): New builtins.
19154 2017-08-08 Julia Koval <julia.koval@intel.com>
19156 PR target/73350,80862
19157 * config/i386/subst.md (round): Fix round pattern.
19158 * config/i386/i386.c (ix86_erase_embedded_rounding):
19159 Fix erasing rounding for the fixed pattern.
19161 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19163 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
19165 2017-06-08 Martin Liska <mliska@suse.cz>
19167 PR gcov-profile/80911
19168 * gcov.c (block_info::block_info): New constructor.
19170 2017-06-07 Carl Love <cel@us.ibm.com>
19172 * config/rs6000/rs6000-c: The return type of the following
19173 built-in functions was implemented as int not long long. Fix sign
19174 of return value for the unsigned version of vec_mulo and vec_mule.
19175 vector unsigned long long vec_bperm (vector unsigned long long,
19176 vector unsigned char)
19177 vector signed long long vec_mule (vector signed int,
19179 vector unsigned long long vec_mule (vector unsigned int,
19180 vector unsigned int)
19181 vector signed long long vec_mulo (vector signed int,
19183 vector unsigned long long vec_mulo (vector unsigned int,
19184 vector unsigned int)
19185 * doc/extend.texi: Fix the documentation for the built-in
19188 2017-06-07 Carl Love <cel@us.ibm.com>
19191 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
19194 2017-06-07 Carl Love <cel@us.ibm.com>
19196 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
19197 support, Generate doublehv for signed int/float for BE case only.
19199 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
19201 * doc/invoke.texi (mcx16): Rewrite.
19203 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19205 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
19206 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
19207 *mov<mode>_softfloat, and an anonymous splitter): Use
19208 nonimmediate_operand instead of rs6000_nonimmediate_operand.
19210 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19212 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
19214 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19215 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
19216 (rs6000_debug_reg_global): Adjust.
19217 (rs6000_init_hard_regno_mode_ok): Adjust.
19218 (rs6000_dbx_register_number): Adjust.
19219 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
19220 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19221 Remove SPE_ACC and SPEFSCR.
19222 (REG_ALLOC_ORDER): Ditto.
19223 (FRAME_POINTER_REGNUM): Change to 111.
19224 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
19225 (REG_CLASS_NAMES): Ditto.
19226 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
19227 (REGISTER_NAMES): Ditto.
19228 (ADDITIONAL_REG_NAMES): Ditto.
19229 (rs6000_reg_names): Ditto.
19230 * config/rs6000/rs6000.md: Renumber some register number
19233 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19235 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
19237 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19238 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
19240 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
19241 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19242 Delete the SPE high registers.
19243 (REG_ALLOC_ORDER): Ditto.
19244 (enum reg_class): Remove SPE_HIGH_REGS.
19245 (REG_CLASS_NAMES): Ditto.
19246 (REG_CLASS_CONTENTS): Delete the SPE high registers.
19247 (REGISTER_NAMES): Ditto.
19248 (rs6000_reg_names): Ditto.
19249 * doc/tm.texi.in: Remove SPE as example.
19250 * doc/tm.texi: Regenerate.
19252 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19254 * config/rs6000/8540.md (ppc8540_brinc): Delete.
19255 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
19256 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
19257 * config/rs6000/rs6000.md (type): Remove "brinc".
19259 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19261 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
19262 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
19263 * config/rs6000/linuxspe.h: Delete file.
19264 * config/rs6000/rs6000.md: Don't include spe.md.
19265 * config/rs6000/spe.h: Delete file.
19266 * config/rs6000/spe.md: Delete file.
19267 * config/rs6000/t-rs6000: Remove spe.md.
19269 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19271 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
19272 (reg_or_none500mem_operand): Delete.
19273 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
19274 instead of reg_or_none500mem_operand.
19276 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19278 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19279 handling of SPE flags.
19280 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
19282 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19284 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
19286 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
19287 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
19288 paired_divv2sf3): Similar.
19289 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
19290 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
19291 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
19293 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
19294 Rename the paired_* instruction patterns.
19295 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
19297 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
19298 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
19299 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
19300 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
19301 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
19302 PAIRED_VECTOR_MODE.
19303 (struct machine_function): Delete field spe_insn_chain_scanned_p.
19304 (spe_func_has_64bit_regs_p): Delete.
19305 (spe_expand_predicate_builtin): Delete.
19306 (spe_expand_evsel_builtin): Delete.
19307 (TARGET_DWARF_REGISTER_SPAN): Do not define.
19308 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
19309 (invalid_e500_subreg): Delete.
19310 (rs6000_legitimize_address): Always force_reg op2 as well, for
19311 paired single memory accesses.
19312 (rs6000_member_type_forces_blk): Delete.
19313 (rs6000_spe_function_arg): Delete.
19314 (rs6000_expand_unop_builtin): Delete SPE handling.
19315 (rs6000_expand_binop_builtin): Ditto.
19316 (spe_expand_stv_builtin): Delete.
19317 (bdesc_2arg_spe): Delete.
19318 (spe_expand_builtin): Delete.
19319 (spe_expand_predicate_builtin): Delete.
19320 (spe_expand_evsel_builtin): Delete.
19321 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
19322 (spe_init_builtins): Delete.
19323 (spe_func_has_64bit_regs_p): Delete.
19324 (savres_routine_name): Delete "info" parameter. Adjust callers.
19325 (rs6000_emit_stack_reset): Ditto.
19326 (rs6000_dwarf_register_span): Delete.
19327 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
19328 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
19329 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
19331 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
19333 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
19334 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
19335 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
19336 mulv2sf3, divv2sf3): Delete expanders.
19338 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19340 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
19342 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19344 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
19345 * config/rs6000/rs6000.c: Ditto.
19347 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19349 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
19350 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
19351 comparison_operator.
19353 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19355 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
19356 * config/rs6000/rs6000.opt: Ditto.
19357 * config/rs6000/t-rtems: Ditto.
19359 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19361 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
19362 TARGET_E500_SINGLE by 0, simplify.
19363 * config/rs6000/rs6000.c: Ditto.
19364 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
19365 (spe_build_register_parallel): Delete.
19366 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
19367 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
19368 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
19369 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
19370 (E500_CONVERT): Delete.
19371 * config/rs6000/spe.md: Remove many patterns and all define_constants.
19373 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19375 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
19376 * config/rs6000/dfp.md: Ditto.
19377 (negdd2, *negdd2_fpr): Merge.
19378 (absdd2, *absdd2_fpr): Merge.
19379 (negtd2, *negtd2_fpr): Merge.
19380 (abstd2, *abstd2_fpr): Merge.
19381 * config/rs6000/e500.h: Delete file.
19382 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
19383 TARGET_FPRS by 1 and simplify.
19384 * config/rs6000/rs6000-c.c: Ditto.
19385 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
19386 TARGET_DF_SPE by 0.
19387 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
19389 * config/rs6000/rs6000.md: Ditto.
19390 (floatdidf2, *floatdidf2_fpr): Merge.
19391 (move_from_CR_gt_bit): Delete.
19392 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
19393 (E500_CR_IOR_COMPARE): Delete.
19394 (All patterns that require !TARGET_FPRS): Delete.
19395 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
19397 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19399 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
19401 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19403 * graphds.c (add_edge): Intitialize edge's attached data.
19404 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
19405 pointer parameter. Call pointed function on each edge during
19406 graph traversing. Skip traversing the edge when the function
19408 (graphds_dfs, graphds_scc): Ditto.
19409 (for_each_edge): New parameter. Pass the new parameter to callback
19411 * graphds.h (skip_edge_callback): New function pointer type.
19412 (graphds_dfs, graphds_scc): New function pointer parameter.
19413 (graphds_edge_callback, for_each_edge): New parameter.
19415 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19417 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
19418 out code checking if runtime alias check is possible to below ...
19419 Call the new function.
19420 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
19421 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
19423 2017-06-07 Marek Polacek <polacek@redhat.com>
19426 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
19427 TYPE_OVERFLOW_WRAPS check.
19429 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19431 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
19432 if versioning is required.
19433 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
19434 peeling with the check for versioning.
19436 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19438 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
19439 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
19440 Set true to new parameter if new ssa variable is defined.
19441 (vect_gen_vector_loop_niters): Refactor. Set range information
19442 for the new vector loop bound variable.
19443 (vect_do_peeling): Ditto.
19445 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19447 * tree-affine.c (ssa.h): Include header file.
19448 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
19449 has wrapping overflow behavior.
19451 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19453 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
19455 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19457 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
19458 (tree_to_aff_combination): ... here.
19460 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19462 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
19463 reg_pressure model function.
19464 (ivopts_global_cost_for_size): Delete.
19465 (determine_set_costs, iv_ca_recount_cost): Call new model function
19466 ivopts_estimate_reg_pressure.
19468 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19470 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
19471 expensive than udiv. Remove floating point cases from mod.
19473 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19475 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
19476 Increase idiv cost.
19478 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19480 * config/aarch64/aarch64.md
19481 (copysignsf3): Fix mask generation.
19483 2017-06-07 Jakub Jelinek <jakub@redhat.com>
19485 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
19487 (class dump_manager): Add register_dumps method.
19488 * dumpfile.c: Include langhooks.h.
19489 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
19490 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
19491 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
19492 (dump_manager::dump_register): Start with 512 entries instead of 32.
19493 (dump_manager::register_dumps): New method.
19494 * toplev.c (general_init): Instead of invoking register_dumps
19495 langhook, invoke register_dumps method on the dump manager.
19496 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
19499 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
19501 * doc/md.texi: Clarify the restrictions on a define_insn condition.
19502 Say that # requires an associated define_split to exist, and that
19503 the define_split must be suitable for use after register allocation.
19505 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19507 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
19508 (compute_outgoing_frequencies): Also initialize zero counts.
19509 (find_many_sub_basic_blocks): Do not produce uninitialized profile
19510 around loops; preserve more of profile when nothing changes.
19512 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
19514 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
19516 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
19517 * config/arm/arm-cpu-cdata.h: Regenerate.
19518 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
19519 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
19520 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
19521 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
19522 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
19524 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
19526 * config/arm/t-rmprofile: Likewise.
19527 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
19529 2017-06-06 David S. Miller <davem@davemloft.net>
19532 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
19533 blockage if function uses alloca.
19535 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19537 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
19538 New "uid" fields to hold pretty-print IDs of group and ref.
19539 Memory references are now identified as <group_id>:<ref_id>
19540 instead of using [random] addresses.
19541 (dump_mem_details): Simplify, no functional change.
19542 (dump_mem_ref): Simplify and make output more concise.
19543 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
19544 (find_or_create_group): Initialize group uid.
19545 (record_ref): Initialize ref uid. Improve debug output.
19546 (prune_group_by_reuse, should_issue_prefetch_p,)
19547 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
19548 (mark_nontemporal_store, determine_loop_nest_reuse):
19549 Improve debug output.
19551 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19553 * dbgcnt.def (prefetch): New debug counter.
19554 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
19555 (schedule_prefetches): Stop issueing prefetches if debug counter
19558 2017-06-06 Tom de Vries <tom@codesourcery.com>
19560 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
19561 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
19563 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19565 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
19566 Use aarch64_reg_or_zero predicate for operand 4.
19567 (aarch64_compare_and_swap<mode> define_insn_and_split):
19568 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
19569 (aarch64_store_exclusive<mode>): Likewise for operand 2.
19571 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19573 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
19574 (arm_compute_save_core_reg_mask): This.
19575 (thumb1_compute_save_reg_mask): Rename into ...
19576 (thumb1_compute_save_core_reg_mask): This.
19577 (arm_compute_save_reg0_reg12_mask): Adapt comment.
19578 (arm_compute_frame_layout): Likewise.
19580 2017-06-06 Richard Biener <rguenther@suse.de>
19582 PR tree-optimization/80974
19583 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
19584 keep or clear leaders SSA info.
19586 2017-06-06 Tom de Vries <tom@codesourcery.com>
19588 * config/nvptx/nvptx.c (split_mode_p): New function.
19589 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
19591 2017-06-06 Tom de Vries <tom@codesourcery.com>
19593 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
19595 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19598 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
19601 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19603 * shrink-wrap.c (handle_simple_exit): Update profile.
19604 (try_shrink_wrapping): Upate profile.
19606 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19608 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
19609 (tree_guess_outgoing_edge_probabilities): New.
19610 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
19611 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
19613 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19615 * ipa-split.c (split_function): Initialize return bb profile.
19617 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19619 * profile.c (compute_branch_probabilities): Also initialize
19620 EXIT_BLOCK profile.
19622 2017-06-06 Richard Biener <rguenther@suse.de>
19624 PR tree-optimization/80928
19625 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
19626 (vect_analyze_loop_operations): Properly guard analysis for
19628 (vect_transform_loop): Likewise.
19629 (vect_analyze_loop_2): Also reset SLP type on PHIs.
19630 (vect_model_induction_cost): Do not cost for pure SLP.
19631 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
19632 of induction in inner loop vectorization.
19633 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
19634 (vect_get_and_check_slp_defs): Handle vect_induction_def.
19635 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
19637 (vect_analyze_slp_cost_1): Cost induction.
19638 (vect_detect_hybrid_slp_stmts): Handle PHIs.
19639 (vect_get_slp_vect_defs): Likewise.
19640 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
19641 (vect_transform_stmt): Handle SLP reductions.
19642 * tree-vectorizer.h (vectorizable_induction): Adjust.
19644 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19646 * config/rs6000/rs6000.c (make_resolver_func): Update
19647 init_lowered_empty_function call.
19649 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
19651 * doc/invoke.texi: Document the -fprofile-abs-path option.
19652 * common.opt (fprofile-abs-path): New option.
19653 * gcov-io.h (gcov_write_filename): Declare.
19654 * gcov-io.c (gcov_write_filename): New function.
19655 * coverage.c (coverage_begin_function): Use gcov_write_filename.
19656 * profile.c (output_location): Likewise.
19658 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19660 * shring-wrap.c: Revert accidental commit.
19662 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
19664 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
19666 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19668 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
19670 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
19671 profile in callgraph edge.
19672 * profile-count.h (apply_probability): If THIS is 0, then result is 0
19673 (apply_scale): Likewise.
19674 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
19675 Also scale profile when inlining function with zero profile.
19676 (initialize_cfun): Update exit block profile even when it is zero.
19677 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
19678 when profile is read.
19680 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19682 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
19683 (CLONE_*): New constants to define the processors we can generate
19684 code for with the target_clone attribute.
19685 (rs6000_clone_map): New array to identify which clone processors
19686 the current program is running on.
19687 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
19688 target_clone attribute.
19689 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
19690 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
19691 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
19692 (cpu_expand_builtin): Add support for target_clone attribute.
19693 (rs6000_valid_attribute_p): Allow "default" attribute.
19694 (get_decl_name): New debug function to simplify printing the
19695 current function name in debugging statements.
19696 (rs6000_clone_priority): New functions to support the target_clone
19697 attribute, and be able to generate code to switch between ISA 2.05
19698 through ISA 3.0 (power6 through power9).
19699 (rs6000_compare_version_priority): Likewise.
19700 (rs6000_get_function_versions_dispatcher): Likewise.
19701 (make_resolver_func): Likewise.
19702 (add_condition_to_bb): Likewise.
19703 (dispatch_function_versions): Likewise.
19704 (rs6000_generate_version_dispatcher_body): Likewise.
19705 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
19706 (fusion_gpr_load_p): Fix a spacing issue.
19707 * doc/extend.texi (Common Function Attributes): Document that the
19708 PowerPC supports the target_clone attribute.
19710 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19712 * config/arm/arm.h: explain F symbol found in description of ARM
19713 register allocation in its legend.
19715 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19717 * config/mips/frame-header-opt.c: Include profile-count.h.
19718 * config/riscv/riscv.c: Include profile-count.h
19720 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19722 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
19724 (sm_set_flag_if_changed): Add bbs field.
19725 (execute_sm_if_changed_flag_set): Pass BBS.
19726 (execute_sm): Update.
19728 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19730 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
19733 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19735 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
19736 (peephole2): New peephole2 to emit the above.
19737 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
19739 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19741 * config/aarch64/aarch64.c (define_peephole2 above
19742 *sub_<shift>_<mode>): New peephole.
19744 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19746 * config/i386/i386.c (make_resolver_func): Update.
19747 * Makefile.in: Add profile-count.h and profile-count.o
19748 * auto-profile.c (afdo_indirect_call): Update to new API.
19749 (afdo_set_bb_count): Update.
19750 (afdo_propagate_edge): Update.
19751 (afdo_propagate_circuit): Update.
19752 (afdo_calculate_branch_prob): Update.
19753 (afdo_annotate_cfg): Update.
19754 * basic-block.h: Include profile-count.h
19755 (struct edge_def): Turn count to profile_count.
19756 (struct basic_block_def): Likewie.
19757 (REG_BR_PROB_BASE): Move to profile-count.h
19758 (RDIV): Move to profile-count.h
19759 * bb-reorder.c (max_entry_count): Turn to profile_count.
19760 (find_traces): Update.
19761 (rotate_loop):Update.
19762 (connect_traces):Update.
19763 (sanitize_hot_paths):Update.
19764 * bt-load.c (migrate_btr_defs): Update.
19765 * cfg.c (RDIV): Remove.
19766 (init_flow): Use alloc_block.
19767 (alloc_block): Uninitialize count.
19768 (unchecked_make_edge): Uninitialize count.
19769 (check_bb_profile): Update.
19770 (dump_edge_info): Update.
19771 (dump_bb_info): Update.
19772 (update_bb_profile_for_threading): Update.
19773 (scale_bbs_frequencies_int): Update.
19774 (scale_bbs_frequencies_gcov_type): Update.
19775 (scale_bbs_frequencies_profile_count): New.
19776 * cfg.h (update_bb_profile_for_threading): Update.
19777 (scale_bbs_frequencies_profile_count): Declare.
19778 * cfgbuild.c (compute_outgoing_frequencies): Update.
19779 (find_many_sub_basic_blocks): Update.
19780 * cfgcleanup.c (try_forward_edges): Update.
19781 (try_crossjump_to_edge): Update.
19782 * cfgexpand.c (expand_gimple_tailcall): Update.
19783 (construct_exit_block): Update.
19784 * cfghooks.c (verify_flow_info): Update.
19785 (dump_bb_for_graph): Update.
19786 (split_edge): Update.
19787 (make_forwarder_block): Update.
19788 (duplicate_block): Update.
19789 (account_profile_record): Update.
19790 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
19791 (get_estimated_loop_iterations): Update.
19792 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
19793 (single_likely_exit): Update.
19794 * cfgloopmanip.c (scale_loop_profile): Update.
19796 (set_zero_probability): Update.
19797 (lv_adjust_loop_entry_edge): Update.
19798 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
19799 (purge_dead_edges): Update.
19800 (rtl_account_profile_record): Update.
19801 * cgraph.c (cgraph_node::create): Uninitialize count.
19802 (symbol_table::create_edge): Uninitialize count.
19803 (cgraph_update_edges_for_call_stmt_node): Update.
19804 (cgraph_edge::dump_edge_flags): Update.
19805 (cgraph_node::dump): Update.
19806 (cgraph_edge::maybe_hot_p): Update.
19807 * cgraph.h: Include profile-count.h
19808 (create_clone), create_edge, create_indirect_edge): Update.
19809 (cgraph_node): Turn count to profile_count.
19810 (cgraph_edge0: Likewise.
19811 (make_speculative, clone): Update.
19812 (create_edge): Update.
19813 (init_lowered_empty_function): Update.
19814 * cgraphclones.c (cgraph_edge::clone): Update.
19815 (duplicate_thunk_for_node): Update.
19816 (cgraph_node::create_clone): Update.
19817 * cgraphunit.c (cgraph_node::analyze): Update.
19818 (cgraph_node::expand_thunk): Update.
19819 * final.c (dump_basic_block_info): Update.
19820 * gimple-streamer-in.c (input_bb): Update.
19821 * gimple-streamer-out.c (output_bb): Update.
19822 * graphite.c (print_global_statistics): Update.
19823 (print_graphite_scop_statistics): Update.
19824 * hsa-brig.c: Include basic-block.h.
19825 * hsa-dump.c: Include basic-block.h.
19826 * hsa-gen.c (T sum_slice): Update.
19827 (convert_switch_statements):Update.
19828 * hsa-regalloc.c: Include basic-block.h.
19829 * ipa-chkp.c (chkp_produce_thunks): Update.
19830 * ipa-cp.c (struct caller_statistics): Update.
19831 (init_caller_stats): Update.
19832 (gather_caller_stats): Update.
19833 (ipcp_cloning_candidate_p): Update.
19834 (good_cloning_opportunity_p): Update.
19835 (get_info_about_necessary_edges): Update.
19836 (dump_profile_updates): Update.
19837 (update_profiling_info): Update.
19838 (update_specialized_profile): Update.
19839 (perhaps_add_new_callers): Update.
19840 (decide_about_value): Update.
19841 (ipa_cp_c_finalize): Update.
19842 * ipa-devirt.c (struct odr_type_warn_count): Update.
19843 (struct decl_warn_count): Update.
19844 (struct final_warning_record): Update.
19845 (possible_polymorphic_call_targets): Update.
19846 (ipa_devirt): Update.
19847 * ipa-fnsummary.c (redirect_to_unreachable): Update.
19848 * ipa-icf.c (sem_function::merge): Update.
19849 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
19850 * ipa-inline.c (compute_uninlined_call_time): Update.
19851 (compute_inlined_call_time): Update.
19852 (want_inline_small_function_p): Update.
19853 (want_inline_self_recursive_call_p): Update.
19854 (edge_badness): Update.
19855 (lookup_recursive_calls): Update.
19856 (recursive_inlining): Update.
19857 (inline_small_functions): Update.
19858 (dump_overall_stats): Update.
19859 (dump_inline_stats): Update.
19860 * ipa-profile.c (ipa_profile_generate_summary): Update.
19861 (ipa_propagate_frequency): Update.
19862 (ipa_profile): Update.
19863 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19864 * ipa-utils.c (ipa_merge_profiles): Update.
19865 * loop-doloop.c (doloop_modify): Update.
19866 * loop-unroll.c (report_unroll): Update.
19867 (unroll_loop_runtime_iterations): Update.
19868 * lto-cgraph.c (lto_output_edge): Update.
19869 (lto_output_node): Update.
19870 (input_node): Update.
19871 (input_edge): Update.
19872 (merge_profile_summaries): Update.
19873 * lto-streamer-in.c (input_cfg): Update.
19874 * lto-streamer-out.c (output_cfg): Update.
19875 * mcf.c (create_fixup_graph): Update.
19876 (adjust_cfg_counts): Update.
19877 (sum_edge_counts): Update.
19878 * modulo-sched.c (sms_schedule): Update.
19879 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
19880 * predict.c (maybe_hot_count_p): Update.
19881 (probably_never_executed): Update.
19882 (dump_prediction): Update.
19883 (combine_predictions_for_bb): Update.
19884 (propagate_freq): Update.
19885 (handle_missing_profiles): Update.
19886 (counts_to_freqs): Update.
19887 (rebuild_frequencies): Update.
19888 (force_edge_cold): Update.
19889 * predict.h: Include profile-count.h
19890 (maybe_hot_count_p, counts_to_freqs): UPdate.
19891 * print-rtl-function.c: Do not include cfg.h
19892 * print-rtl.c: Include basic-block.h
19893 * profile-count.c: New file.
19894 * profile-count.h: New file.
19895 * profile.c (is_edge_inconsistent): Update.
19896 (correct_negative_edge_counts): Update.
19897 (is_inconsistent): Update.
19898 (set_bb_counts): Update.
19899 (read_profile_edge_counts): Update.
19900 (compute_frequency_overlap): Update.
19901 (compute_branch_probabilities): Update; Initialize and deinitialize
19903 (branch_prob): Update.
19904 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
19905 (edge_gcov_count): New.
19906 (bb_gcov_count): New.
19907 * shrink-wrap.c (try_shrink_wrapping): Update.
19908 * tracer.c (better_p): Update.
19909 * trans-mem.c (expand_transaction): Update.
19910 (ipa_tm_insert_irr_call): Update.
19911 (ipa_tm_insert_gettmclone_call): Update.
19912 * tree-call-cdce.c: Update.
19913 * tree-cfg.c (gimple_duplicate_sese_region): Update.
19914 (gimple_duplicate_sese_tail): Update.
19915 (gimple_account_profile_record): Update.
19916 (execute_fixup_cfg): Update.
19917 * tree-inline.c (copy_bb): Update.
19918 (copy_edges_for_bb): Update.
19919 (initialize_cfun): Update.
19920 (freqs_to_counts): Update.
19921 (copy_cfg_body): Update.
19922 (expand_call_inline): Update.
19923 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19924 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
19925 (try_unroll_loop_completely): Update.
19926 (try_peel_loop): Update.
19927 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19928 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
19929 * tree-ssa-loop-split.c (connect_loops): Update.
19930 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
19931 * tree-ssa-reassoc.c (branch_fixup): Update.
19932 * tree-ssa-tail-merge.c (replace_block_by): Update.
19933 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
19934 (compute_path_counts): Update.
19935 (update_profile): Update.
19936 (recompute_probabilities): Update.
19937 (update_joiner_offpath_counts): Update.
19938 (estimated_freqs_path): Update.
19939 (freqs_to_counts_path): Update.
19940 (clear_counts_path): Update.
19941 (ssa_fix_duplicate_block_edges): Update.
19942 (duplicate_thread_path): Update.
19943 * tree-switch-conversion.c (case_bit_test_cmp): Update.
19944 (struct switch_conv_info): Update.
19945 * tree-tailcall.c (decrease_profile): Update.
19946 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19947 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19948 * value-prof.c (check_counter): Update.
19949 (gimple_divmod_fixed_value): Update.
19950 (gimple_mod_pow2): Update.
19951 (gimple_mod_subtract): Update.
19952 (gimple_ic_transform): Update.
19953 (gimple_stringop_fixed_value): Update.
19954 * value-prof.h (gimple_ic): Update.
19956 2017-06-02 Carl Love <cel@us.ibm.com>
19958 * config/rs6000/rs6000-c: Add support for built-in functions
19959 vector double vec_doublee (vector signed int);
19960 vector double vec_doublee (vector unsigned int);
19961 vector double vec_doublee (vector float);
19962 vector double vec_doubleh (vector signed int);
19963 vector double vec_doubleh (vector unsigned int);
19964 vector double vec_doubleh (vector float);
19965 vector double vec_doublel (vector signed int);
19966 vector double vec_doublel (vector unsigned int);
19967 vector double vec_doublel (vector float);
19968 vector double vec_doubleo (vector signed int);
19969 vector double vec_doubleo (vector unsigned int);
19970 vector double vec_doubleo (vector float);.
19971 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
19972 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
19974 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
19975 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
19976 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
19978 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
19979 vec_doublel, vec_doubleh.
19980 * doc/extend.texi: Update the built-in documentation file for the
19981 new built-in functions.
19983 2017-06-02 David Malcolm <dmalcolm@redhat.com>
19986 * ipa-inline-analysis.c (free_growth_caches): Set
19987 edge_removal_hook_holder to NULL after removing it.
19989 2017-06-02 Sudakshina Das <sudi.das@arm.com>
19991 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
19992 comparision with zero.
19994 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19995 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19996 for early expansion of vec_min and vec_max builtins.
19997 (builtin_function_type): Add min/max unsigned variants to those
19998 identified as having unsigned arguments.
20000 2017-06-02 Olivier Hainque <hainque@adacore.com>
20002 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
20004 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20006 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
20007 Use VALL_F16 iterator rather than VALL.
20009 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20011 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
20012 Emit CBNZ inside loop when doing a strong exchange and comparing
20013 against zero. Generate the CC flags after the loop.
20015 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
20017 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
20018 (dl_section_ref): New.
20019 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
20020 On AIX, append an expression to subtract the size of the
20021 section length to dl_section_ref.
20023 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20025 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20026 for early expansion of vector absolute builtins.
20028 2017-06-02 Richard Biener <rguenther@suse.de>
20030 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
20031 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
20033 2017-06-02 Richard Biener <rguenther@suse.de>
20035 PR tree-optimization/80948
20036 * tree-tailcall.c (find_tail_calls): Track stmts to move in
20037 stmt order as well.
20039 2017-06-02 Richard Biener <rguenther@suse.de>
20041 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
20043 * tree-vect-stmts.c (process_use): Do not mark backedge defs
20044 for inductions as relevant.
20046 2017-06-02 Richard Biener <rguenther@suse.de>
20048 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
20049 (vectorizable_induction): ... this. Remove dead code.
20051 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
20053 * builtins. (expand_builtin_alloca): Remove second parameter and
20054 infer its value from the first parameter instead.
20055 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
20057 2017-06-02 Jakub Jelinek <jakub@redhat.com>
20059 PR rtl-optimization/80903
20060 * loop-doloop.c (add_test): Unshare sequence.
20062 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
20064 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
20066 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
20068 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
20070 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
20071 xlogue_layout::get_instance, logue_layout::xlogue_layout,
20072 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
20073 (xlogue_layout::get_stub_rtx): Make static.
20074 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
20075 (xlogue_layout::compute_stub_managed_regs): Rename to...
20076 (xlogue_layout::count_stub_managed_regs): ...this.
20077 (xlogue_layout::is_stub_managed_reg): New function.
20078 (xlogue_layout::m_stub_names): Rename to...
20079 (xlogue_layout::s_stub_names): ...this, make static.
20080 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
20081 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
20082 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
20083 xlogue_layout::s_stub_names): Instantiate statics.
20084 (stub_managed_regs): Remove.
20085 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
20086 (disable_call_ms2sysv_xlogues): Rename to...
20087 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
20088 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
20090 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
20091 change after reload_completed.
20092 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
20094 (ix86_expand_prologue): Likewise.
20095 (ix86_expand_epilogue): Likewise.
20096 (ix86_expand_split_stack_prologue): Likewise.
20097 (ix86_compute_frame_layout): Remove frame parameter ...
20098 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
20099 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
20101 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
20102 (ix86_frame): Move from here ...
20103 * config/i386/i386.h (ix86_frame): ... to here.
20104 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
20105 complete ix86_frame data structure instead. Remove some_ld_name.
20107 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20109 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
20110 symbols that hold a DECL_VALUE_EXPR.
20112 2017-06-01 Martin Jambor <mjambor@suse.cz>
20114 PR tree-optimization/80898
20115 * tree-sra.c (process_subtree_disqualification): Removed.
20116 (disqualify_candidate): Do not acll
20117 process_subtree_disqualification.
20118 (subtree_mark_written_and_enqueue): New function.
20119 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
20120 RHS has been disqualified and re-queue LHS if necessary. Apart
20121 from that, ignore disqualified RHS.
20123 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20125 * config/s390/s390.c (s390_emit_epilogue): Disable early return
20126 address fetch for z10 or later.
20128 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20130 * config/arc/arc.md (tst_movb): Add guard when splitting.
20132 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20134 * config/arc/arc.c (arc_can_eliminate): Test against
20135 arc_frame_pointer_needed.
20137 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20139 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
20140 to prevent store reordering.
20141 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
20142 (type): Add block type.
20143 (stack_tie): Define special instruction to be used in
20146 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20148 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
20149 constraint. It is not valid for the pattern.
20150 (noncommutative_binary_comparison): Likewise.
20152 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20154 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
20157 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20159 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
20160 be used by the reg-alloc.
20162 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20164 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
20165 reg-alloc when having mul64 or mul32x16 instructions.
20166 (mulsidi3): Likewise.
20167 (umulsidi3): Likewise.
20168 (mulsi32x16): New pattern.
20169 (mulsi64): Likewise.
20170 (mulsidi64): Likewise.
20171 (umulsidi64): Likewise.
20172 (MUL32x16_REG): Define.
20173 (mul64_600): Use MUL32x16_REG.
20174 (mac64_600): Likewise.
20175 (umul64_600): Likewise.
20176 (umac64_600): Likewise.
20178 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20180 * config/arc/arc.md (mulsi3_700): Make it commutative.
20182 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
20184 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
20186 (*sign_extendsidi2_insn): Likewise for movstosw.
20188 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20190 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
20191 the type of the input discriminant value. Convert the
20192 discriminant value of signedness vary.
20194 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
20196 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
20197 Add to -Wall section.
20199 2017-06-01 Richard Biener <rguenther@suse.de>
20201 PR middle-end/66313
20202 * fold-const.c (fold_plusminus_mult_expr): If the factored
20203 factor may be zero use a wrapping type for the inner operation.
20204 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
20205 and handle moved defs.
20206 (process_assignment): Properly guard the unary op case. Return a
20207 tri-state indicating that moving the stmt before the call may allow
20208 to continue. Pass through to_move.
20209 (find_tail_calls): Handle moving unrelated defs before
20212 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
20215 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
20216 splitter result in the canonical way.
20218 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
20220 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
20221 also for 32bit target. Update insn attributes.
20222 (zero-extendsidi2 splitter): Allow all registers for operand 1.
20224 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
20226 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
20227 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
20228 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
20229 (_mm_maskz_min_ss): New intrinsics.
20231 2017-05-31 Martin Liska <mliska@suse.cz>
20233 * tree-vect-loop.c (vect_create_epilog_for_reduction):
20234 Change comment style to one we normally use.
20235 (vectorizable_reduction): Likewise.
20236 (vectorizable_induction): Likewise.
20237 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20238 (vectorizable_call): Likewise.
20239 (vectorizable_simd_clone_call): Likewise.
20240 (vectorizable_conversion): Likewise.
20241 (vectorizable_assignment): Likewise.
20242 (vectorizable_shift): Likewise.
20243 (vectorizable_operation): Likewise.
20244 (vectorizable_store): Likewise.
20245 (vectorizable_load): Likewise.
20246 * tree-vectorizer.h: Likewise.
20248 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
20250 * passes.c (emergency_dump_function): New.
20251 * tree-pass.h (emergency_dump_function): Declare.
20252 * plugin.c (plugins_internal_error_function): Remove.
20253 * plugin.h (plugins_internal_error_function): Remove declaration.
20254 * toplev.c (internal_error_function): New static function. Use it...
20255 (general_init): ...here.
20257 2017-05-31 Graham Markall <graham.markall@embecosm.com>
20259 * config/arc/arc.c (arc_print_operand): Handle constant operands.
20260 (arc_rtx_costs): Add costs for new patterns.
20261 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
20262 * config/arc/predicates.md: Add _1_2_3_operand predicate.
20264 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
20266 * tree-ssa-strlen.c (get_next_strinfo): New function.
20267 (get_stridx_plus_constant): Use it.
20268 (zero_length_string): Likewise.
20269 (adjust_related_strinfos): Likewise.
20270 (adjust_last_stmt): Likewise.
20272 2017-05-31 Richard Biener <rguenther@suse.de>
20275 * config/i386/i386.c (ix86_expand_builtin): Remove assert
20276 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
20278 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
20280 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
20281 loop_vinfo argument and use of dependence distance vectors.
20282 Check instead whether the two references differ only in their
20283 initial value and assume that they have the same alignment if the
20284 difference is a multiple of the vector alignment.
20285 (vect_analyze_data_refs_alignment): Update call accordingly.
20287 2017-05-31 Martin Liska <mliska@suse.cz>
20290 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
20292 2017-05-31 Bin Cheng <bin.cheng@arm.com>
20294 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
20295 (create_intersect_range_checks): Move from ...
20296 * tree-data-ref.c (create_intersect_range_checks_index)
20297 (create_intersect_range_checks): ... to here.
20298 (create_runtime_alias_checks): New function factored from ...
20299 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
20300 here. Call above function.
20301 * tree-data-ref.h (create_runtime_alias_checks): New function.
20303 2017-05-31 Bin Cheng <bin.cheng@arm.com>
20305 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
20306 segment length for dr_b and compute it in wide_int.
20308 2017-05-31 Richard Biener <rguenther@suse.de>
20310 PR tree-optimization/80906
20311 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
20312 and pass through iv_map.
20313 (copy_bb_and_scalar_dependences): Adjust.
20314 (translate_pending_phi_nodes): Likewise.
20315 (copy_loop_close_phi_args): Handle code-generating IVs instead
20318 2017-05-30 David Malcolm <dmalcolm@redhat.com>
20320 * diagnostic-color.c (color_dict): Add "type-diff".
20321 (parse_gcc_colors): Update comment.
20322 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
20323 -fdiagnostics-show-template-tree and -fno-elide-type.
20324 (GCC_COLORS): Add type-diff to example.
20326 (-fdiagnostics-show-template-tree): New.
20327 (-fno-elide-type): New.
20328 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
20329 the pp_format_decoder callback. Call any m_format_postprocessor's
20331 (pretty_printer::pretty_printer): Initialize
20332 m_format_postprocessor.
20333 (pretty_printer::~pretty_printer): Delete any
20334 m_format_postprocessor.
20335 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
20336 (class format_postprocessor): New class.
20337 (struct pretty_printer::format_decoder): Document the new parameters.
20338 (struct pretty_printer::m_format_postprocessor): New field.
20339 * tree-diagnostic.c (default_tree_printer): Update for new
20340 bool and const char ** params.
20341 * tree-diagnostic.h (default_tree_printer): Likewise.
20343 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
20345 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
20346 (lwa_operand): Delete rs6000_gen_cell_microcode test.
20347 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20348 rs6000_gen_cell_microcode code.
20349 (rs6000_final_prescan_insn): Delete.
20350 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
20351 "warn-cell-microcode" entries.
20352 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
20353 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
20354 throughout. Change cc_reg_not_micro_cr0_operand to
20355 cc_reg_not_cr0_operand throughout.
20356 (*extendhi<mode>2_noload): Delete.
20357 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
20358 (mwarn-cell-microcode): Delete.
20359 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
20360 -mgen-cell-microcode and -mwarn-cell-microcode.
20362 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
20365 * config/i386/constraints.md (Yd): New constraint.
20367 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
20368 and (?Yd, r) alternatives. Update insn attributes.
20369 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
20370 and (?*Yd, r) alternatives. Update insn attributes.
20371 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
20373 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
20375 * gimplify.c (gimplify_modify_expr): Don't create a
20376 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
20379 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
20381 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
20383 2017-05-30 Richard Biener <rguenther@suse.de>
20385 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
20386 and reduc_def fields.
20387 (STMT_VINFO_REDUC_TYPE): New define.
20388 (STMT_VINFO_REDUC_DEF): Likewise.
20389 (vect_force_simple_reduction): Adjust prototype.
20390 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
20391 (vect_is_simple_reduction): Remove check_reduction argument.
20392 (vect_force_simple_reduction): Adjust and set
20393 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
20394 (vectorizable_reduction): Do not re-do reduction analysis
20395 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
20396 * tree-parloops.c (gather_scalar_reductions): Adjust.
20398 2017-05-30 Richard Biener <rguenther@suse.de>
20400 PR middle-end/80901
20401 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
20404 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
20406 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
20407 Introduce unknown_misalignment parameter and remove vf.
20408 (vect_peeling_hash_get_lowest_cost):
20409 Pass unknown_misalignment parameter.
20410 (vect_enhance_data_refs_alignment):
20411 Fix unsupportable data ref treatment.
20413 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20415 * tree-vect-data-refs.c (vect_get_data_access_cost):
20416 Workaround for SLP handling.
20417 (vect_enhance_data_refs_alignment):
20418 Compute costs for doing no peeling at all, compare to the best
20419 peeling costs so far and avoid peeling if cheaper.
20421 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20423 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
20424 Return peeling info and set costs to zero for unlimited cost
20426 (vect_enhance_data_refs_alignment): Also inspect all datarefs
20427 with unknown misalignment. Compute and costs for unknown
20428 misalignment, compare them to the costs for known misalignment
20429 and choose the cheapest for peeling.
20431 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20433 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
20434 (vect_get_peeling_costs_all_drs): Create function.
20435 (vect_peeling_hash_get_lowest_cost):
20436 Use vect_get_peeling_costs_all_drs.
20437 (vect_peeling_supportable): Create function.
20438 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
20440 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20442 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
20443 DR_HAS_NEGATIVE_STEP.
20444 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
20445 (vect_enhance_data_refs_alignment): Use.
20446 (vect_duplicate_ssa_name_ptr_info): Use.
20447 * tree-vectorizer.h (dr_misalignment): Use.
20448 (known_alignment_for_access_p): Use.
20450 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20453 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
20455 (has_section_name): New function.
20457 2017-05-30 Martin Liska <mliska@suse.cz>
20460 * auto-profile.c (get_function_decl_from_block): Fix
20463 2017-05-30 Richard Biener <rguenther@suse.de>
20465 PR middle-end/80876
20466 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
20468 2017-05-30 Martin Liska <mliska@suse.cz>
20470 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
20471 * dumpfile.h (struct dump_file_info): Remove ctors.
20473 2017-05-30 Martin Liska <mliska@suse.cz>
20475 * predict.def: Fix GNU coding style.
20477 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
20479 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
20480 Mark 'to' argument with ATTRIBUTE_UNUSED.
20482 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
20484 * config/xtensa/xtensa.c (xtensa_emit_call): Use
20485 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
20486 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
20489 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
20491 * doc/install.texi (Options specification): Restore entry of
20492 --enable-sjlj-exceptions.
20494 2017-05-27 Michael Eager <eager@eagercon.com>
20497 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
20499 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
20501 * config/microblaze/microblaze.h
20502 (FIXED_REGISTERS): Update in macro.
20503 (CALL_USED_REGISTERS): Update in macro.
20505 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
20507 * doc/install.texi: Add links to macOS binary distributions.
20509 2017-05-27 Jakub Jelinek <jakub@redhat.com>
20513 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20515 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20517 2017-05-26 Martin Liska <mliska@suse.cz>
20519 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
20521 2017-05-26 Martin Liska <mliska@suse.cz>
20523 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
20524 always leading ';; '.
20525 (dump_bb_info): Likewise.
20526 (brief_dump_cfg): Likewise.
20527 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
20528 * dumpfile.c: Remove usage of TDF_VERBOSE.
20529 * dumpfile.h (enum dump_kind): Likewise.
20530 (dump_gimple_bb_header): Do not use TDF_COMMENT.
20531 * print-tree.c (debug_verbose): Remove.
20532 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
20533 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
20534 * tree-diagnostic.c (default_tree_printer): Replace
20535 TDF_DIAGNOSTIC with TDF_SLIM.
20537 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20539 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
20540 in parameter loop, rather than loop_vinfo.
20541 (create_intersect_range_checks): Ditto.
20542 (vect_create_cond_for_alias_checks): Update call to above functions.
20544 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20546 PR tree-optimization/80815
20547 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
20548 for merging runtime alias checks. Handle negative DR_STEPs.
20550 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20552 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
20554 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
20555 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
20556 out code pruning runtime alias checks.
20557 * tree-data-ref.c (prune_runtime_alias_test_list): New function
20558 factored out from above.
20559 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20561 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20563 (prune_runtime_alias_test_list): New decalaration.
20565 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20567 * tree-vect-data-refs.c (compare_tree): Rename and move ...
20568 * tree-data-ref.c (data_ref_compare_tree): ... to here.
20569 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
20570 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
20571 (operator==, comp_dr_with_seg_len_pair): Ditto.
20572 (vect_prune_runtime_alias_test_list): Ditto.
20574 2017-05-26 Martin Liska <mliska@suse.cz>
20577 * params.def: Bound partial-inlining-entry-probability param.
20579 2017-05-26 Marek Polacek <polacek@redhat.com>
20582 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
20585 2017-05-26 Richard Biener <rguenther@suse.de>
20587 PR tree-optimization/80842
20588 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
20591 2017-05-26 Richard Biener <rguenther@suse.de>
20593 PR tree-optimization/80844
20594 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
20596 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
20598 * doc/md.texi (Machine Constraints): Update x86 family
20599 machine constraints section to match 'config/i386/constraints.md'.
20601 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
20603 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
20605 2017-05-25 Nathan Sidwell <nathan@acm.org>
20607 * doc/invoke.texi (--enable-languages): Update documentation.
20609 2017-05-25 Martin Liska <mliska@suse.cz>
20611 * dumpfile.c: Add TDF_FOLDING.
20612 * dumpfile.h (enum dump_kind): Likewise.
20613 * genmatch.c (dt_simplify::gen_1): Use it.
20615 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20617 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
20619 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20621 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20622 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
20624 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20626 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
20627 * match.pd (X == C): Rewrite it here.
20628 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
20629 with_certain_nonzero_bits2): New predicates.
20630 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
20632 2017-05-24 Nathan Sidwell <nathan@acm.org>
20634 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
20637 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
20640 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
20642 * config/powerpcspe: New port. Files are copied from the rs6000
20643 port, with "rs6000" in filenames replaced by "powerpcspe".
20645 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
20647 PR rtl-optimization/80754
20648 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
20650 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
20652 * config/sparc/sparc.md (length): Return the correct value for -mflat
20653 sibcalls to match output_sibcall.
20655 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
20659 * config/rs6000/rs6000.c (struct machine_function): Add new field
20661 (rs6000_get_separate_components): Init that field, use it.
20662 (rs6000_components_for_bb): Use the field.
20664 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
20666 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
20668 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
20670 PR middle-end/80823
20671 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
20673 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20676 * config/s390/s390.c (s390_check_qrst_address): Check incoming
20677 address against address_operand predicate.
20678 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
20680 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
20682 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
20683 parameters passed indirectly.
20685 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
20687 * config/i386/i386.md (*movdi_internal): Remove SSE4
20688 alternative 18 (?r, *v). Update insn attributes.
20689 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
20690 Update insn attributes.
20691 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
20692 Update insn attributes.
20693 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
20694 alternative 1 (r, v). Remove isa attribute.
20695 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
20696 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
20697 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
20699 2017-05-23 Tom de Vries <tom@codesourcery.com>
20701 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
20704 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20706 * cgraphunit.c (symbol_table::process_new_functions): Update.
20707 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
20708 (inline_generate_summary): Rename to ...
20709 (ipa_fn_summary_generate): ... this one.
20710 (inline_read_summary): Rename to ...
20711 (ipa_fn_summary_read): ... this one.
20712 (inline_write_summary): Rename to ...
20713 (ipa_fn_summary_write): ... this one.
20714 (inline_free_summary): Rename to ...
20715 (ipa_free_fn_summary): ... this one.
20716 (pass_data_local_fn_summary, pass_local_fn_summary,
20717 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
20718 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
20719 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
20720 make_pass_ipa_fn_summary): New.
20721 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
20722 inline_write_summary, inline_free_summary): Remove.
20723 (ipa_free_fn_summary) : New.
20724 * ipa-inline.c (ipa_inline): Update.
20725 (pass_ipa_inline): Do not generate summaries.
20726 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
20728 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
20729 and add pass_ipa_fn_summary.
20730 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
20732 (make_pass_inline_parameters): Remove.
20734 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
20736 * omp-low.c (struct omp_context): Remove "default_kind" member.
20739 * omp-offload.c (execute_oacc_device_lower): Remove the
20740 parallelism dimensions function attributes for unparallelized
20741 OpenACC kernels constructs.
20743 2017-05-23 Martin Liska <mliska@suse.cz>
20745 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
20747 (cgraph_edge::make_speculative): Likewise.
20748 (cgraph_edge::resolve_speculation): Likewise.
20749 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
20750 (cgraph_node::dump): Likewise.
20751 * cgraph.h: Likewise.
20752 * cgraphunit.c (analyze_functions): Likewise.
20753 (symbol_table::compile): Likewise.
20754 * ipa-cp.c (print_all_lattices): Likewise.
20755 (determine_versionability): Likewise.
20756 (initialize_node_lattices): Likewise.
20757 (ipcp_verify_propagated_values): Likewise.
20758 (estimate_local_effects): Likewise.
20759 (update_profiling_info): Likewise.
20760 (create_specialized_node): Likewise.
20761 (perhaps_add_new_callers): Likewise.
20762 (decide_about_value): Likewise.
20763 (decide_whether_version_node): Likewise.
20764 (identify_dead_nodes): Likewise.
20765 (ipcp_store_bits_results): Likewise.
20766 * ipa-devirt.c (dump_targets): Likewise.
20767 (ipa_devirt): Likewise.
20768 * ipa-icf.c (sem_item::dump): Likewise.
20769 (sem_function::equals): Likewise.
20770 (sem_variable::equals): Likewise.
20771 (sem_item_optimizer::read_section): Likewise.
20772 (sem_item_optimizer::execute): Likewise.
20773 (congruence_class::dump): Likewise.
20774 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
20775 (dump_inline_summary): Likewise.
20776 (estimate_node_size_and_time): Likewise.
20777 (inline_analyze_function): Likewise.
20778 * ipa-inline-transform.c (inline_call): Likewise.
20779 * ipa-inline.c (report_inline_failed_reason): Likewise.
20780 (want_early_inline_function_p): Likewise.
20781 (edge_badness): Likewise.
20782 (update_edge_key): Likewise.
20783 (inline_small_functions): Likewise.
20784 * ipa-profile.c (ipa_profile): Likewise.
20785 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
20786 (ipa_make_edge_direct_to_target): Likewise.
20787 (remove_described_reference): Likewise.
20788 (ipa_impossible_devirt_target): Likewise.
20789 (propagate_controlled_uses): Likewise.
20790 (ipa_print_node_params): Likewise.
20791 (ipcp_transform_function): Likewise.
20792 * ipa-pure-const.c (pure_const_read_summary): Likewise.
20793 (propagate_pure_const): Likewise.
20794 * ipa-reference.c (generate_summary): Likewise.
20795 (read_write_all_from_decl): Likewise.
20796 (propagate): Likewise.
20797 (ipa_reference_read_optimization_summary): Likewise.
20798 * ipa-utils.c (ipa_merge_profiles): Likewise.
20799 * ipa.c (walk_polymorphic_call_targets): Likewise.
20800 (symbol_table::remove_unreachable_nodes): Likewise.
20801 (ipa_single_use): Likewise.
20802 * passes.c (execute_todo): Likewise.
20803 * predict.c (drop_profile): Likewise.
20804 * symtab.c (symtab_node::get_dump_name): New function.
20805 (symtab_node::dump_name): Likewise.
20806 (symtab_node::dump_asm_name): Likewise.
20807 (symtab_node::dump_references): Likewise.
20808 (symtab_node::dump_referring): Likewise.
20809 (symtab_node::dump_base): Likewise.
20810 (symtab_node::debug_symtab): Likewise.
20811 * tree-sra.c (convert_callers_for_node): Likewise.
20812 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
20813 * value-prof.c (init_node_map): Likewise.
20815 2017-05-23 Martin Liska <mliska@suse.cz>
20817 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
20818 and symtab_node::debug_symtab to symbol_table::debug.
20819 * cgraphunit.c (analyze_functions): Use the renamed function.
20820 (symbol_table::compile): Likewise.
20821 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
20822 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
20823 * passes.c (execute_todo): Likewise.
20824 * symtab.c (symbol_table::dump): New function.
20825 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
20827 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20829 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
20830 that nonconst implies exec.
20832 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20834 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
20835 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20836 (inline_edge_summary_vec): Turn into ...
20837 (ipa_call_summaries): ... this one.
20838 (redirect_to_unreachable, edge_set_predicate,
20839 evaluate_properties_for_edge, inline_summary_alloc,
20840 reset_ipa_call_summary, reset_inline_summary,
20841 inline_summary_t::duplicate): Update.
20842 (inline_edge_duplication_hook): Turn to ...
20843 (ipa_call_summary_t::duplicate): ... this one.
20844 (inline_edge_removal_hook): Turn to ...
20845 (ipa_call_summary_t::remove): ... this one.
20846 (dump_inline_edge_summary): Turn to ...
20847 (dump_ipa_call_summary): ... this one.
20848 (estimate_function_body_sizes): Update.
20849 (inline_update_callee_summaries): Update.
20850 (remap_edge_change_prob): Update.
20851 (remap_edge_summaries): Update.
20852 (inline_merge_summary): Update.
20853 (do_estimate_edge_time): Update.
20854 (inline_generate_summary): Update.
20855 (inline_read_section): Update.
20856 (inline_read_summary): Update.
20857 (inline_free_summary): Update.
20858 * ipa-inline.c (can_inline_edge_p): Update.
20859 (compute_inlined_call_time): Update.
20860 (want_inline_small_function_p): Update.
20861 (edge_badness): Update.
20862 (early_inliner): Update.
20863 * ipa-inline.h (inline_edge_summary): Turn to ...
20864 (ipa_call_summary): ... this one.
20865 (ipa_call_summary_t): New class.
20866 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20867 (ipa_call_summaries): New.
20868 (inline_edge_summary): Remove.
20869 (estimate_edge_growth): Update.
20870 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20871 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20872 * ipa-split.c (execute_split_functions): Update.
20873 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20875 2017-05-23 Tom de Vries <tom@codesourcery.com>
20877 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
20878 attributes): Document rdrand effective target.
20880 2017-05-23 Tom de Vries <tom@codesourcery.com>
20882 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
20883 attributes): Sort alphabetically.
20885 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
20887 * config/avr/genmultilib.awk: Use gsub instead of gensub.
20889 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
20892 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
20893 V2DF/V2DI splat into two separate patterns, one that handles
20894 registers, and the other that only handles memory. Drop support
20895 for splatting from a GPR on ISA 2.07 and then splitting the
20896 splat into direct move and splat.
20897 (vsx_splat_<mode>_reg): Likewise.
20898 (vsx_splat_<mode>_mem): Likewise.
20900 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
20902 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
20904 2017-05-22 Jakub Jelinek <jakub@redhat.com>
20906 PR middle-end/80809
20907 * omp-low.c (finish_taskreg_remap): New function.
20908 (finish_taskreg_scan): If unit size of ctx->record_type
20909 is non-constant, unshare the size expression and replace
20910 decls in it with possible outer var refs.
20912 PR middle-end/80809
20913 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
20914 GOVD_SHARED rather than GOVD_PRIVATE with it.
20915 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
20916 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
20918 PR middle-end/80853
20919 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
20920 as last argument to build_outer_var_ref for pointer bases of array
20921 section reductions.
20923 2017-05-19 Martin Sebor <msebor@redhat.com>
20925 * print-tree.c (print_node): Print DECL_READ_P flag.
20927 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20929 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
20930 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
20931 * cgraph.c: Likewise.
20932 * cgraphunit.c: Likewise.
20933 * gengtype.c: Likewise.
20934 * ipa-cp.c: Likewise.
20935 * ipa-devirt.c: Likewise.
20936 * ipa-icf.c: Likewise.
20937 * ipa-predicate.c: Likewise.
20938 * ipa-profile.c: Likewise.
20939 * ipa-prop.c: Likewise.
20940 * ipa-split.c: Likewise.
20942 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
20943 edge_predicate_pool, dump_inline_hints,
20944 inline_summary::account_size_time, redirect_to_unreachable,
20945 edge_set_predicate, set_hint_predicate,
20946 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
20947 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
20948 inline_summary_t::remove, remap_hint_predicate_after_duplication,
20949 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
20950 ipa_call_summary_t::remove, initialize_growth_caches,
20951 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
20952 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
20953 mark_modified, unmodified_parm_1, unmodified_parm,
20954 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
20955 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
20956 compute_bb_predicates, will_be_nonconstant_expr_predicate,
20957 will_be_nonconstant_predicate, record_modified_bb_info,
20958 get_minimal_bb, record_modified, param_change_prob,
20959 phi_result_unknown_predicate, predicate_for_phi_result,
20960 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
20961 estimate_function_body_sizes, compute_inline_parameters,
20962 compute_inline_parameters_for_curren, pass_data_inline_parameters,
20963 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
20964 inline_update_callee_summaries, remap_edge_change_prob,
20965 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
20966 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
20967 inline_analyze_function, inline_summary_t::insert,
20968 inline_generate_summary, read_ipa_call_summary, inline_read_section,
20969 inline_read_summary, write_ipa_call_summary, inline_write_summary,
20970 inline_free_summary): Move to ipa-fnsummary.h
20971 (predicate_t): Remove.
20972 * ipa-fnsummary.c: New file.
20973 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
20974 (enum inline_hints_vals, inline_hints, agg_position_info,
20975 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
20976 inline_summaries, ipa_call_summary, ipa_call_summary_t,
20977 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
20978 dump_inline_summary, dump_inline_hints, inline_generate_summary,
20979 inline_read_summary, inline_write_summary, inline_free_summary,
20980 inline_analyze_function, initialize_inline_failed,
20981 inline_merge_summary, inline_update_overall_summary,
20982 compute_inline_parameters): Move to ipa-fnsummary.h
20983 * ipa-fnsummary.h: New file.
20984 * ipa-inline-transform.h: Include ipa-inline.h.
20985 * ipa-inline.c: LIkewise.
20987 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20989 * ipa-inline.c (edge_badness): Use inlined_time instead of
20990 inline_summaries->get.
20992 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20994 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
20996 2017-05-22 Nathan Sidwell <nathan@acm.org>
20998 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
20999 (fdump-lang): Document 'raw' option.
21000 * dumpfile.h (TDI_tu): Delete.
21001 * dumpfile.c (dump_files): Remove translation-unit.
21002 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
21004 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
21006 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
21007 command option from $(AWK) call.
21008 * config/avr/genmultilib.awk: Simplify and rewrite so that it
21009 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
21010 [FORMAT]: Remove handling of variable.
21011 * config/avr/t-multilib: Regenerate.
21013 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21015 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
21017 (dump_inline_summary): Do not print self_time.
21018 (estimate_function_body_sizes): Do not set self_time.
21019 (compute_inline_parameters): Likewise.
21020 (inline_read_section, inline_write_summary): Do not stream self_time.
21021 * ipa-inline.h (inline_summary): Drop self_time.
21023 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21025 * ipa-inline-analysis.c (account_size_time): Rename to ...
21026 (inline_summary::account_size_time): ... this one.
21027 (reset_ipa_call_summary): Turn to ...
21028 (ipa_call_summary::reset): ... this one.
21029 (reset_inline_summary): Turn to ...
21030 (inline_summary::reset): ... this one.
21031 (inline_summary_t::remove): Update.
21032 (inline_summary_t::duplicate): Update.
21033 (ipa_call_summary_t::remove): Update.
21034 (dump_inline_summary): Update.
21035 (estimate_function_body_sizes): Update.
21036 (compute_inline_parameters): Update.
21037 (estimate_node_size_and_time): Update.
21038 (inline_merge_summary): Update.
21039 (inline_update_overall_summary): Update.
21040 (inline_read_section): Update.
21041 (inline_write_summary): Update.
21042 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
21043 add account_size_time and reset member functions.
21044 (ipa_call_summary): Add reset function.
21045 * ipa-predicate.h (predicate::operator &): Constify.
21047 2017-05-22 Richard Biener <rguenther@suse.de>
21049 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
21051 2017-05-19 Jason Merrill <jason@redhat.com>
21053 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
21055 2017-05-19 Marek Polacek <polacek@redhat.com>
21058 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
21059 TYPE_OVERFLOW_WRAPS checks.
21061 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
21063 * tree-core.h (enum omp_clause_default_kind): Add
21064 "OMP_CLAUSE_DEFAULT_PRESENT".
21065 * tree-pretty-print.c (dump_omp_clause): Handle it.
21066 * gimplify.c (enum gimplify_omp_var_data): Add
21067 "GOVD_MAP_FORCE_PRESENT".
21068 (gimplify_adjust_omp_clauses_1): Map it to
21069 "GOMP_MAP_FORCE_PRESENT".
21070 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
21072 * gimplify.c (oacc_default_clause): Clarify.
21074 2017-05-19 Nathan Sidwell <nathan@acm.org>
21076 LANG_HOOK_REGISTER_DUMPS
21077 * toplev.c (general_init): Call register dump lang hook.
21078 * doc/invoke.texi: Document -fdump-lang option family.
21079 * dumpfile.c (dump_files): Remove class dump here.
21080 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
21081 * dumpfile.h (tree_dump_index): Remove TDI_class.
21082 * langhooks-def.h (lhd_register_dumps): Declare.
21083 (LANG_HOOKS_REGISTER_DUMPS): Define.
21084 (LANG_HOOKS_INITIALIZER): Add it.
21085 * langhooks.c (lhd_register_dumps): Define.
21086 * langhooks.h (struct lang_hooks): Add register_dumps.
21088 2017-05-19 Nathan Sidwell <nathan@acm.org>
21090 * context.h (context::set_passes): New.
21091 * context.c (context::context): Do not create pass manager.
21092 * toplev.c (general_init): Create pass manager here.
21094 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
21096 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
21097 use this splitter if two add or or instructions would also work for
21098 the constant we want to generate.
21100 2017-05-19 Richard Biener <rguenther@suse.de>
21103 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
21104 predicate evaluation.
21106 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21108 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
21110 * ipa-inline.c (want_inline_small_function_p): Do not cast to
21113 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21115 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21116 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21117 (inline_edge_summary_vec): Turn into ...
21118 (ipa_call_summaries): ... this one.
21119 (redirect_to_unreachable, edge_set_predicate,
21120 evaluate_properties_for_edge, inline_summary_alloc,
21121 reset_ipa_call_summary, reset_inline_summary,
21122 inline_summary_t::duplicate): Update.
21123 (inline_edge_duplication_hook): Turn to ...
21124 (ipa_call_summary_t::duplicate): ... this one.
21125 (inline_edge_removal_hook): Turn to ...
21126 (ipa_call_summary_t::remove): ... this one.
21127 (dump_inline_edge_summary): Turn to ...
21128 (dump_ipa_call_summary): ... this one.
21129 (estimate_function_body_sizes): Update.
21130 (inline_update_callee_summaries): Update.
21131 (remap_edge_change_prob): Update.
21132 (remap_edge_summaries): Update.
21133 (inline_merge_summary): Update.
21134 (do_estimate_edge_time): Update.
21135 (inline_generate_summary): Update.
21136 (inline_read_section): Update.
21137 (inline_read_summary): Update.
21138 (inline_free_summary): Update.
21139 * ipa-inline.c (can_inline_edge_p): Update.
21140 (compute_inlined_call_time): Update.
21141 (want_inline_small_function_p): Update.
21142 (edge_badness): Update.
21143 (early_inliner): Update.
21144 * ipa-inline.h (inline_edge_summary): Turn to ...
21145 (ipa_call_summary): ... this one.
21146 (ipa_call_summary_t): New class.
21147 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21148 (ipa_call_summaries): New.
21149 (inline_edge_summary): Remove.
21150 (estimate_edge_growth): Update.
21151 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21152 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21153 * ipa-split.c (execute_split_functions): Update.
21154 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21156 2017-05-19 Richard Biener <rguenther@suse.de>
21158 PR middle-end/80764
21159 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
21161 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
21163 * config/rs6000/rs6000.c (struct machine_function): Add field
21164 fpr_is_wrapped_separately.
21165 (rs6000_get_separate_components): Use 64 components. Handle the
21166 new FPR components.
21167 (rs6000_components_for_bb): Handle the FPR components.
21168 (rs6000_emit_prologue_components): Handle the FPR components.
21169 (rs6000_emit_epilogue_components): Handle the FPR components.
21170 (rs6000_set_handled_components): Handle the FPR components.
21171 (rs6000_emit_prologue): Don't output prologue code for those FPRs
21172 that are already separately shrink-wrapped.
21173 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
21174 that are already separately shrink-wrapped.
21176 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21179 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
21182 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
21183 (define_peephole2 for Altivec d-form load): Add peepholes to catch
21184 cases where the register allocator uses a move and an offsettable
21185 memory operation to/from a FPR register on ISA 2.06/2.07.
21186 (define_peephole2 for Altivec d-form store): Likewise.
21188 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
21191 * config/i386/mmx.md (*mov<mode>_internal): Enable
21192 alternatives 11, 12, 13 and 14 also for 32bit targets.
21193 Remove alternatives 15, 16, 17 and 18.
21194 * config/i386/sse.md (vec_concatv2di): Change
21195 alternative (!x, *y) to (x, ?!*Yn).
21197 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
21199 * dumpfile.h (enum dump_kind): Remove stray comma.
21201 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
21203 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
21204 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
21205 predicate::num_conditions
21206 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
21207 (CHANGED): turn into predicate::changed.
21208 (agg_position_info): Move to ipa-predicate.h
21209 (add_condition, predicate::add_clause, predicate::operator &=,
21210 predicate::or_with, predicate::evaluate, predicate::probability,
21211 dump_condition, dump_clause, predicate::dump,
21212 predicate::remap_after_duplication, predicate::remap_after_inlining,
21213 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
21214 (evaluate_conditions_for_known_args): Update.
21215 (set_cond_stmt_execution_predicate): Update.
21216 * ipa-inline.h: Include ipa-predicate.h
21217 (condition, inline_param_summary, conditions, agg_position_info,
21218 predicate): Move to ipa-predicate.h
21219 * ipa-predicate.c: New file.
21220 * ipa-predicate.h: New file.
21222 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
21224 * final.c (leaf_function_p): Check we are not in a sequence.
21226 2017-05-18 Martin Liska <mliska@suse.cz>
21228 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
21229 * dumpfile.c (dump_register): Use new enum dump_kind.
21230 (get_dump_file_name): Likewise.
21231 (dump_enable_all): Likewise.
21232 (dump_switch_p_1): Likewise.
21233 (enable_rtl_dump_file): Remove usage of TDF_RTL.
21234 * dumpfile.h (enum dump_kind): New enum type.
21235 (struct dump_file_info): Create constructor and
21236 format fields and comments.
21237 * passes.c (pass_manager::register_one_dump_file):
21239 * statistics.c (statistics_early_init): Likewise.
21240 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
21241 TDF_TREE with TDF_SLIM.
21242 (gather_memory_references_ref): Likewise.
21244 2017-05-18 Martin Liska <mliska@suse.cz>
21246 * vec.h (struct vnull): Use it.
21248 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
21250 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
21251 (true_predicate, false_predicate, true_predicate_p,
21252 false_predicate_p): Remove.
21253 (single_cond_predicate, not_inlined_predicate): Turn to member function
21255 (add_condition): Update.
21256 (add_clause): Turn to...
21257 (predicate::add_clause): ... this one; update; allow passing NULL
21259 (and_predicates): Turn to ...
21260 (predicate::operator &=): ... this one.
21261 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
21262 (or_predicates): Turn to ...
21263 (predicate::or_with): ... this one.
21264 (evaluate_predicate): Turn to ...
21265 (predicate::evaluate): ... this one.
21266 (predicate_probability): Turn to ...
21267 (predicate::probability): ... this one.
21268 (dump_condition): Update.
21269 (dump_predicate): Turn to ...
21270 (predicate::dump): ... this one.
21271 (account_size_time): Update.
21272 (edge_set_predicate): Update.
21273 (set_hint_predicate): UPdate.
21274 (evaluate_conditions_for_known_args): Update.
21275 (evaluate_properties_for_edge): Update.
21276 (remap_predicate_after_duplication): Turn to...
21277 (predicate::remap_after_duplication): ... this one.
21278 (remap_hint_predicate_after_duplication): Update.
21279 (inline_summary_t::duplicate): UPdate.
21280 (dump_inline_edge_summary): Update.
21281 (dump_inline_summary): Update.
21282 (set_cond_stmt_execution_predicate): Update.
21283 (set_switch_stmt_execution_predicate): Update.
21284 (compute_bb_predicates): Update.
21285 (will_be_nonconstant_expr_predicate): Update.
21286 (will_be_nonconstant_predicate): Update.
21287 (phi_result_unknown_predicate): Update.
21288 (predicate_for_phi_result): Update.
21289 (array_index_predicate): Update.
21290 (estimate_function_body_sizes): Update.
21291 (estimate_node_size_and_time): Update.
21292 (estimate_ipcp_clone_size_and_time): Update.
21293 (remap_predicate): Rename to ...
21294 (predicate::remap_after_inlining): ... this one.
21295 (remap_hint_predicate): Update.
21296 (inline_merge_summary): Update.
21297 (inline_update_overall_summary): Update.
21298 (estimate_size_after_inlining): Update.
21299 (read_predicate): Rename to ...
21300 (predicate::stream_in): ... this one.
21301 (read_inline_edge_summary): Update.
21302 (write_predicate): Rename to ...
21303 (predicate::stream_out): ... this one.
21304 (write_inline_edge_summary): Update.
21305 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
21306 (clause_t): Turn to uint32_t
21307 (predicate): Turn to class; implement constructor and operators
21309 (size_time_entry): Update.
21310 (inline_summary): Update.
21311 (inline_edge_summary): Update.
21313 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
21315 * fold-const.c (fold_binary_loc): Move transformation...
21316 * match.pd (C - X CMP X): ... here.
21318 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
21320 * config/sparc/sparc.c (sparc_option_override): Set function
21321 alignment for -mcpu=niagara7 to 64 to match the I$ line.
21322 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
21324 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
21326 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
21328 2017-05-18 Marek Polacek <polacek@redhat.com>
21331 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
21332 (pass_ubsan::execute): Call gimple_assign_single_p instead of
21333 gimple_assign_load_p.
21335 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
21337 PR middle-end/80692
21338 * real.c (do_compare): Give decimal_do_compare preference over
21339 comparing just the signs.
21341 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
21343 * doc/md.texi (Canonicalization of Instructions): Describe the
21344 canonical form of instructions that inherently set a condition
21347 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
21349 PR middle-end/80775
21350 * tree-cfg.c: Move deletion of unreachable case statements to after
21351 the merging of consecutive case labels.
21353 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
21355 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
21356 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
21357 restoring of callee-saved registers.
21359 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
21361 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
21362 * config/visium/visium.c (single_set_and_flags): Likewise.
21363 * config/visium/visium.md (Substitutions): Likewise.
21365 2017-05-17 Martin Liska <mliska@suse.cz>
21367 * cfg.c: Introduce dump_flags_t type and
21368 use it instead of int type.
21370 * cfghooks.c: Likewise.
21371 * cfghooks.h (struct cfg_hooks): Likewise.
21372 * cfgrtl.c: Likewise.
21373 * cfgrtl.h: Likewise.
21374 * cgraph.c (cgraph_node::get_body): Likewise.
21375 * coretypes.h: Likewise.
21376 * domwalk.c: Likewise.
21377 * domwalk.h: Likewise.
21378 * dumpfile.c (struct dump_option_value_info): Likewise.
21379 (dump_enable_all): Likewise.
21380 (dump_switch_p_1): Likewise.
21381 (opt_info_switch_p): Likewise.
21382 * dumpfile.h (enum tree_dump_index): Likewise.
21383 (struct dump_file_info): Likewise.
21384 * genemit.c: Likewise.
21385 * generic-match-head.c: Likewise.
21386 * gengtype.c (open_base_files): Likewise.
21387 * gimple-pretty-print.c: Likewise.
21388 * gimple-pretty-print.h: Likewise.
21389 * graph.c (print_graph_cfg): Likewise.
21390 * graphite-scop-detection.c (dot_all_sese): Likewise.
21391 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
21392 * loop-unroll.c (report_unroll): Likewise.
21393 * passes.c (pass_manager::register_one_dump_file): Likewise.
21394 * print-tree.c: Likewise.
21395 * statistics.c: Likewise.
21396 * tree-cfg.c: Likewise.
21397 * tree-cfg.h: Likewise.
21398 * tree-dfa.c: Likewise.
21399 * tree-dfa.h: Likewise.
21400 * tree-dump.c (dump_function): Likewise.
21401 * tree-dump.h (struct dump_info): Likewise.
21402 * tree-pretty-print.c: Likewise.
21403 * tree-pretty-print.h: Likewise.
21404 * tree-ssa-live.c: Likewise.
21405 * tree-ssa-live.h: Likewise.
21406 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
21407 * tree-vect-loop.c: Likewise.
21408 * tree-vect-slp.c: Likewise.
21410 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
21411 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21413 PR tree-optimization/80457
21414 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
21415 of all arguments to a statement as scalar_to_vec operations.
21416 (vectorizable_call): Adjust call to vect_model_simple_cost for
21418 (vectorizable_conversion): Likewise.
21419 (vectorizable_assignment): Likewise.
21420 (vectorizable_shift): Likewise.
21421 (vectorizable_operation): Likewise.
21422 (vectorizable_comparison): Likewise.
21423 (vect_is_simple_cond): Record the def types for operands.
21424 (vectorizable_condition): Likewise, call vect_model_simple_cost.
21425 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
21426 for statement argument count.
21428 2017-05-16 Carl Love <cel@us.ibm.com>
21430 * config/rs6000/rs6000-c: Add support for built-in functions
21431 vector unsigned long long vec_bperm (vector unsigned long long,
21432 vector unsigned char)
21433 vector signed long long vec_mule (vector signed int,
21435 vector unsigned long long vec_mule (vector unsigned int,
21436 vector unsigned int)
21437 vector signed long long vec_mulo (vector signed int,
21439 vector unsigned long long vec_mulo (vector unsigned int,
21440 vector unsigned int)
21441 vector signed char vec_sldw (vector signed char,
21442 vector signed char,
21444 vector unsigned char vec_sldw (vector unsigned char,
21445 vector unsigned char,
21447 vector signed short vec_sldw (vector signed short,
21448 vector signed short,
21450 vector unsigned short vec_sldw (vector unsigned short,
21451 vector unsigned short,
21453 vector signed int vec_sldw (vector signed int,
21456 vector unsigned int vec_sldw (vector unsigned int,
21457 vector unsigned int,
21459 vector signed long long vec_sldw (vector signed long long,
21460 vector signed long long,
21462 vector unsigned long long vec_sldw (vector unsigned long long,
21463 vector unsigned long long,
21465 * config/rs6000/rs6000-c: Add support for built-in functions
21466 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
21467 * config/rs6000/altivec.h: Add defintion for vec_sldw.
21468 * doc/extend.texi: Update the built-in documentation for the
21469 new built-in functions.
21471 2017-05-16 Marek Polacek <polacek@redhat.com>
21475 * tree.c (save_expr): Don't fold the expression.
21477 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
21479 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
21480 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
21481 and (?*y,m). Update insn attributes.
21483 2017-05-16 Martin Liska <mliska@suse.cz>
21485 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
21486 flags argument of print_gimple_stmt, print_gimple_expr,
21487 print_generic_stmt and print_generic_expr.
21488 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
21489 * coretypes.h: Likewise.
21490 * except.c (dump_eh_tree): Likewise.
21491 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
21492 * gimple-pretty-print.h: Likewise.
21493 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
21494 (backprop::push_to_worklist): Likewise.
21495 (backprop::pop_from_worklist): Likewise.
21496 (backprop::process_use): Likewise.
21497 (backprop::intersect_uses): Likewise.
21498 (note_replacement): Likewise.
21499 * gimple-ssa-store-merging.c
21500 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
21501 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
21502 (pass_store_merging::execute): Likewise.
21503 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
21504 (ssa_base_cand_dump_callback): Likewise.
21505 (dump_incr_vec): Likewise.
21506 (replace_refs): Likewise.
21507 (replace_mult_candidate): Likewise.
21508 (create_add_on_incoming_edge): Likewise.
21509 (create_phi_basis): Likewise.
21510 (insert_initializers): Likewise.
21511 (all_phi_incrs_profitable): Likewise.
21512 (introduce_cast_before_cand): Likewise.
21513 (replace_one_candidate): Likewise.
21514 * gimplify.c (gimplify_expr): Likewise.
21515 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
21516 (set_rename): Likewise.
21517 (rename_uses): Likewise.
21518 (copy_loop_phi_nodes): Likewise.
21519 (add_close_phis_to_merge_points): Likewise.
21520 (copy_loop_close_phi_args): Likewise.
21521 (copy_cond_phi_args): Likewise.
21522 (graphite_copy_stmts_from_block): Likewise.
21523 (translate_pending_phi_nodes): Likewise.
21524 * graphite-poly.c (print_pdr): Likewise.
21525 (dump_gbb_cases): Likewise.
21526 (dump_gbb_conditions): Likewise.
21527 (print_scop_params): Likewise.
21528 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
21529 (build_cross_bb_scalars_use): Likewise.
21530 (gather_bbs::before_dom_children): Likewise.
21531 * hsa-dump.c (dump_hsa_immed): Likewise.
21532 * ipa-cp.c (print_ipcp_constant_value): Likewise.
21533 (get_replacement_map): Likewise.
21534 * ipa-inline-analysis.c (dump_condition): Likewise.
21535 (estimate_function_body_sizes): Likewise.
21536 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
21537 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
21538 * ipa-prop.c (ipa_dump_param): Likewise.
21539 (ipa_print_node_jump_functions_for_edge): Likewise.
21540 (ipa_modify_call_arguments): Likewise.
21541 (ipa_modify_expr): Likewise.
21542 (ipa_dump_param_adjustments): Likewise.
21543 (ipa_dump_agg_replacement_values): Likewise.
21544 (ipcp_modif_dom_walker::before_dom_children): Likewise.
21545 * ipa-pure-const.c (check_stmt): Likewise.
21546 (pass_nothrow::execute): Likewise.
21547 * ipa-split.c (execute_split_functions): Likewise.
21548 * omp-offload.c (dump_oacc_loop_part): Likewise.
21549 (dump_oacc_loop): Likewise.
21550 * trans-mem.c (tm_log_emit): Likewise.
21551 (tm_memopt_accumulate_memops): Likewise.
21552 (dump_tm_memopt_set): Likewise.
21553 (dump_tm_memopt_transform): Likewise.
21554 * tree-cfg.c (gimple_verify_flow_info): Likewise.
21555 (print_loop): Likewise.
21556 * tree-chkp-opt.c (chkp_print_addr): Likewise.
21557 (chkp_gather_checks_info): Likewise.
21558 (chkp_get_check_result): Likewise.
21559 (chkp_remove_check_if_pass): Likewise.
21560 (chkp_use_outer_bounds_if_possible): Likewise.
21561 (chkp_reduce_bounds_lifetime): Likewise.
21562 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
21563 (chkp_mark_completed_bounds): Likewise.
21564 (chkp_register_incomplete_bounds): Likewise.
21565 (chkp_mark_invalid_bounds): Likewise.
21566 (chkp_maybe_copy_and_register_bounds): Likewise.
21567 (chkp_build_returned_bound): Likewise.
21568 (chkp_get_bound_for_parm): Likewise.
21569 (chkp_build_bndldx): Likewise.
21570 (chkp_get_bounds_by_definition): Likewise.
21571 (chkp_generate_extern_var_bounds): Likewise.
21572 (chkp_get_bounds_for_decl_addr): Likewise.
21573 * tree-chrec.c (chrec_apply): Likewise.
21574 * tree-data-ref.c (dump_data_reference): Likewise.
21575 (dump_subscript): Likewise.
21576 (dump_data_dependence_relation): Likewise.
21577 (analyze_overlapping_iterations): Likewise.
21578 * tree-inline.c (expand_call_inline): Likewise.
21579 (tree_function_versioning): Likewise.
21580 * tree-into-ssa.c (dump_defs_stack): Likewise.
21581 (dump_currdefs): Likewise.
21582 (dump_names_replaced_by): Likewise.
21583 (dump_update_ssa): Likewise.
21584 (update_ssa): Likewise.
21585 * tree-object-size.c (pass_object_sizes::execute): Likewise.
21586 * tree-parloops.c (build_new_reduction): Likewise.
21587 (try_create_reduction_list): Likewise.
21588 (ref_conflicts_with_region): Likewise.
21589 (oacc_entry_exit_ok_1): Likewise.
21590 (oacc_entry_exit_single_gang): Likewise.
21591 * tree-pretty-print.h: Likewise.
21592 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
21593 (get_scalar_evolution): Likewise.
21594 (add_to_evolution): Likewise.
21595 (get_loop_exit_condition): Likewise.
21596 (analyze_evolution_in_loop): Likewise.
21597 (analyze_initial_condition): Likewise.
21598 (analyze_scalar_evolution): Likewise.
21599 (instantiate_scev): Likewise.
21600 (number_of_latch_executions): Likewise.
21601 (gather_chrec_stats): Likewise.
21602 (final_value_replacement_loop): Likewise.
21603 (scev_const_prop): Likewise.
21604 * tree-sra.c (dump_access): Likewise.
21605 (disqualify_candidate): Likewise.
21606 (create_access): Likewise.
21607 (reject): Likewise.
21608 (maybe_add_sra_candidate): Likewise.
21609 (create_access_replacement): Likewise.
21610 (analyze_access_subtree): Likewise.
21611 (analyze_all_variable_accesses): Likewise.
21612 (sra_modify_assign): Likewise.
21613 (initialize_constant_pool_replacements): Likewise.
21614 (find_param_candidates): Likewise.
21615 (decide_one_param_reduction): Likewise.
21616 (replace_removed_params_ssa_names): Likewise.
21617 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
21618 * tree-ssa-copy.c (dump_copy_of): Likewise.
21619 (copy_prop_visit_cond_stmt): Likewise.
21620 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
21621 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
21622 (record_equivalences_from_stmt): Likewise.
21623 * tree-ssa-dse.c (compute_trims): Likewise.
21624 (delete_dead_call): Likewise.
21625 (delete_dead_assignment): Likewise.
21626 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
21627 (forward_propagate_into_cond): Likewise.
21628 (pass_forwprop::execute): Likewise.
21629 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
21630 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
21632 (move_computations_worker): Likewise.
21633 (execute_sm): Likewise.
21634 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
21635 (remove_exits_and_undefined_stmts): Likewise.
21636 (remove_redundant_iv_tests): Likewise.
21637 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
21638 (adjust_iv_update_pos): Likewise.
21639 * tree-ssa-math-opts.c (bswap_replace): Likewise.
21640 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
21641 (value_replacement): Likewise.
21642 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
21643 * tree-ssa-pre.c (print_pre_expr): Likewise.
21644 (get_representative_for): Likewise.
21645 (create_expression_by_pieces): Likewise.
21646 (insert_into_preds_of_block): Likewise.
21647 (eliminate_insert): Likewise.
21648 (eliminate_dom_walker::before_dom_children): Likewise.
21649 (eliminate): Likewise.
21650 (remove_dead_inserted_code): Likewise.
21651 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
21652 * tree-ssa-reassoc.c (get_rank): Likewise.
21653 (eliminate_duplicate_pair): Likewise.
21654 (eliminate_plus_minus_pair): Likewise.
21655 (eliminate_not_pairs): Likewise.
21656 (undistribute_ops_list): Likewise.
21657 (eliminate_redundant_comparison): Likewise.
21658 (update_range_test): Likewise.
21659 (optimize_range_tests_var_bound): Likewise.
21660 (optimize_vec_cond_expr): Likewise.
21661 (rewrite_expr_tree): Likewise.
21662 (rewrite_expr_tree_parallel): Likewise.
21663 (linearize_expr): Likewise.
21664 (break_up_subtract): Likewise.
21665 (linearize_expr_tree): Likewise.
21666 (attempt_builtin_powi): Likewise.
21667 (attempt_builtin_copysign): Likewise.
21668 (transform_stmt_to_copy): Likewise.
21669 (transform_stmt_to_multiply): Likewise.
21670 (dump_ops_vector): Likewise.
21671 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
21672 (print_scc): Likewise.
21673 (set_ssa_val_to): Likewise.
21674 (visit_reference_op_store): Likewise.
21675 (visit_use): Likewise.
21676 (sccvn_dom_walker::before_dom_children): Likewise.
21677 (run_scc_vn): Likewise.
21678 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
21680 (expr_hash_elt::print): Likewise.
21681 (const_and_copies::pop_to_marker): Likewise.
21682 (const_and_copies::record_const_or_copy_raw): Likewise.
21683 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
21684 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
21685 (dump_predicates): Likewise.
21686 (find_uninit_use): Likewise.
21687 (warn_uninitialized_phi): Likewise.
21688 (pass_late_warn_uninitialized::execute): Likewise.
21689 * tree-ssa.c (verify_vssa): Likewise.
21690 (verify_ssa): Likewise.
21691 (maybe_optimize_var): Likewise.
21692 * tree-vrp.c (dump_value_range): Likewise.
21693 (dump_all_value_ranges): Likewise.
21694 (dump_asserts_for): Likewise.
21695 (register_edge_assert_for_2): Likewise.
21696 (vrp_visit_cond_stmt): Likewise.
21697 (vrp_visit_switch_stmt): Likewise.
21698 (vrp_visit_stmt): Likewise.
21699 (vrp_visit_phi_node): Likewise.
21700 (simplify_cond_using_ranges_1): Likewise.
21701 (fold_predicate_in): Likewise.
21702 (evrp_dom_walker::before_dom_children): Likewise.
21703 (evrp_dom_walker::push_value_range): Likewise.
21704 (evrp_dom_walker::pop_value_range): Likewise.
21705 (execute_early_vrp): Likewise.
21707 2017-05-16 Richard Biener <rguenther@suse.de>
21709 * dwarf2out.c (loc_list_from_tree_1): Do not create
21710 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
21712 2017-05-16 Richard Biener <rguenther@suse.de>
21714 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
21716 (note_variable_value_in_expr): If we resolved the decl ref
21717 do not push to the stack.
21719 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
21721 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
21722 operations in fast-math mode.
21723 (vaddq_f16): Likewise.
21724 (vmul_f16): Likewise.
21725 (vmulq_f16): Likewise.
21726 (vsub_f16): Likewise.
21727 (vsubq_f16): Likewise.
21728 * config/arm/neon.md (add<mode>3): New.
21730 (fma:<VH:mode>3): New. Also remove outdated comment.
21733 2017-05-16 Martin Liska <mliska@suse.cz>
21737 * ipa-devirt.c (warn_types_mismatch): Fix typo.
21738 (odr_types_equivalent_p): Likewise.
21740 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
21742 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
21744 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
21747 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
21748 non-interunit SSE move alternatives with '?'.
21749 (zero-extendsidi peephole2): New peephole to skip intermediate
21750 general register in SSE zero-extend sequence.
21752 2017-05-15 Jeff Law <law@redhat.com>
21754 * reorg.c (relax_delay_slots): Create a new variable to hold
21755 the temporary target rather than clobbering TARGET_LABEL.
21757 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
21758 missing argument to extract_bit_field call.
21759 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
21761 2017-05-15 Martin Liska <mliska@suse.cz>
21764 * gcc.c (process_command): Do not allow empty argument of -o option.
21766 2017-05-15 Renlin Li <renlin.li@arm.com>
21768 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
21769 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
21770 * config/aarch64/constraints.md (Usf): Add long call check.
21771 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
21772 (call_value): Likewise.
21773 (sibcall): Likewise.
21774 (sibcall_value): Likewise.
21776 (call_value_insn): New.
21777 (sibcall_insn): Update rtx pattern.
21778 (sibcall_value_insn): Likewise.
21779 (call_internal): Remove.
21780 (call_value_internal): Likewise.
21781 (sibcall_internal): Likewise.
21782 (sibcall_value_internal): Likewise.
21783 (call_reg): Likewise.
21784 (call_symbol): Likewise.
21785 (call_value_reg): Likewise.
21786 (call_value_symbol): Likewise.
21788 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
21791 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
21793 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
21795 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
21796 compatible with CCGOCmode and with CCZmode.
21798 2017-05-14 Martin Sebor <msebor@redhat.com>
21800 PR middle-end/77671
21801 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
21802 (gimple_fold_builtin_snprintf): Same.
21803 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
21804 (gimple_fold_builtin_snprintf): Same.
21805 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
21806 of character types.
21807 (is_call_safe): New function.
21808 (try_substitute_return_value): Call it.
21809 (try_simplify_call): New function.
21810 (pass_sprintf_length::handle_gimple_call): Call it.
21812 2017-05-14 Martin Sebor <msebor@redhat.com>
21814 PR middle-end/80669
21815 * builtins.c (expand_builtin_stpncpy): Simplify.
21817 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
21819 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
21820 * config/i386/i386.h
21821 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
21822 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
21823 (struct machine_function): Add new members call_ms2sysv,
21824 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
21825 (struct machine_frame_state): New fields sp_realigned and
21826 sp_realigned_offset.
21827 * config/i386/i386.c
21828 (enum xlogue_stub): New enum.
21829 (enum xlogue_stub_sets): New enum.
21830 (class xlogue_layout): New class.
21831 (struct ix86_frame): New fields stack_realign_allocate_offset,
21832 stack_realign_offset and outlined_save_offset. Modify comments to
21833 detail stack layout when using out-of-line stubs.
21834 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
21835 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
21836 -mcall-ms2sysv-xlogues.
21837 (stub_managed_regs): New static variable.
21838 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
21839 registers managed by out-of-line stub.
21840 (disable_call_ms2sysv_xlogues): New function.
21841 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
21842 m->call_ms2sysv when appropriate and compute frame layout for
21844 (sp_valid_at, fp_valid_at): New inline functions.
21845 (choose_basereg): New function.
21846 (choose_baseaddr): Add align parameter, use choose_basereg and modify
21848 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
21849 Use align parameter of choose_baseaddr to generated aligned SSE movs
21851 (pro_epilogue_adjust_stack): Modify to track
21852 machine_frame_state::sp_realigned.
21853 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
21854 (ix86_nsaved_sseregs): Likewise.
21855 (ix86_emit_save_regs): Likewise.
21856 (ix86_emit_save_regs_using_mov): Likewise.
21857 (ix86_emit_save_sse_regs_using_mov): Likewise.
21858 (get_scratch_register_on_entry): Likewise.
21859 (gen_frame_set): New function.
21860 (gen_frame_load): Likewise.
21861 (gen_frame_store): Likewise.
21862 (emit_outlined_ms2sysv_save): Likewise.
21863 (emit_outlined_ms2sysv_restore): Likewise.
21864 (ix86_expand_prologue): Modify stack re-alignment code and call
21865 emit_outlined_ms2sysv_save when appropriate.
21866 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
21867 parameter rtx_insn *insn, which allows the function to be used to only
21868 generate the notes.
21869 (ix86_expand_epilogue): Modify validity checks of frame and stack
21870 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
21871 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
21872 * config/i386/predicates.md
21873 (save_multiple): New predicate.
21874 (restore_multiple): Likewise.
21875 * config/i386/sse.md
21876 (save_multiple<mode>): New pattern.
21877 (save_multiple_realign<mode>): Likewise.
21878 (restore_multiple<mode>): Likewise.
21879 (restore_multiple_and_return<mode>): Likewise.
21880 (restore_multiple_leave_return<mode>): Likewise.
21881 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
21883 2017-05-14 Julia Koval <julia.koval@intel.com>
21885 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
21886 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
21887 (__builtin_ia32_xsetbv): New builtins.
21888 * config/i386/i386.c (ix86_expand_special_args_builtin):
21890 (ix86_expand_builtin): Special expand for new intrinsics.
21891 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
21892 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
21893 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
21895 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21897 * cfganal.c (inverted_post_order_compute): Change argument type
21899 * cfganal.h (inverted_post_order_compute): Adjust prototype.
21900 * df-core.c (rest_of_handle_df_initialize): Adjust.
21901 (rest_of_handle_df_finish): Likewise.
21902 (df_analyze_1): Likewise.
21903 (df_analyze): Likewise.
21904 (loop_inverted_post_order_compute): Change argument to be a vec *.
21905 (df_analyze_loop): Adjust.
21906 (df_get_n_blocks): Likewise.
21907 (df_get_postorder): Likewise.
21908 * df.h (struct df_d): Change field to be a vec.
21909 * lcm.c (compute_laterin): Adjust.
21910 (compute_available): Likewise.
21911 * lra-lives.c (lra_create_live_ranges_1): Likewise.
21912 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
21913 * tree-ssa-pre.c (compute_antic): Likewise.
21915 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21917 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
21918 (depth_first_search::depth_first_search): Change structure init
21919 function to this constructor.
21920 (depth_first_search::add_bb): Rename function to this member.
21921 (depth_first_search::execute): Likewise.
21922 (flow_dfs_compute_reverse_finish): Adjust.
21924 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21926 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
21927 (longest_simple_path): Likewise.
21928 * shrink-wrap.c (spread_components): Likewise.
21929 (disqualify_problematic_components): Likewise.
21930 (emit_common_heads_for_components): Likewise.
21931 (emit_common_tails_for_components): Likewise.
21932 (insert_prologue_epilogue_for_components): Likewise.
21934 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21936 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
21939 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21941 * df-core.c (df_set_blocks): Start using auto_bitmap.
21942 (df_compact_blocks): Likewise.
21943 * df-problems.c (df_rd_confluence_n): Likewise.
21944 * df-scan.c (df_insn_rescan_all): Likewise.
21945 (df_process_deferred_rescans): Likewise.
21946 (df_update_entry_block_defs): Likewise.
21947 (df_update_exit_block_uses): Likewise.
21948 (df_entry_block_bitmap_verify): Likewise.
21949 (df_exit_block_bitmap_verify): Likewise.
21950 (df_scan_verify): Likewise.
21951 * lra-constraints.c (lra_constraints): Likewise.
21952 (undo_optional_reloads): Likewise.
21953 (lra_undo_inheritance): Likewise.
21954 * lra-remat.c (calculate_gen_cands): Likewise.
21955 (do_remat): Likewise.
21956 * lra-spills.c (assign_spill_hard_regs): Likewise.
21957 (spill_pseudos): Likewise.
21958 * tree-ssa-pre.c (bitmap_set_and): Likewise.
21959 (bitmap_set_subtract_values): Likewise.
21961 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21963 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
21964 management with auto_bitmap.
21965 (fix_inter_tick): Likewise.
21966 (fix_recovery_deps): Likewise.
21967 * ira.c (add_store_equivs): Likewise.
21968 (find_moveable_pseudos): Likewise.
21969 (split_live_ranges_for_shrink_wrap): Likewise.
21970 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
21971 (rtx_reuse_manager::seen_def_p): Likewise.
21972 (rtx_reuse_manager::set_seen_def): Likewise.
21973 * print-rtl.h (class rtx_reuse_manager): Likewise.
21975 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21977 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
21979 (migrate_btr_def): Likewise.
21980 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
21981 * df-core.c (loop_post_order_compute): Likewise.
21982 (loop_inverted_post_order_compute): Likewise.
21983 * hsa-common.h: Likewise.
21984 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
21985 * init-regs.c (initialize_uninitialized_regs): Likewise.
21986 * ipa-inline.c (resolve_noninline_speculation): Likewise.
21987 (inline_small_functions): Likewise.
21988 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
21989 * ira.c (combine_and_move_insns): Likewise.
21990 (build_insn_chain): Likewise.
21991 * loop-invariant.c (find_invariants): Likewise.
21992 * lower-subreg.c (propagate_pseudo_copies): Likewise.
21993 * predict.c (tree_predict_by_opcode): Likewise.
21994 (predict_paths_leading_to): Likewise.
21995 (predict_paths_leading_to_edge): Likewise.
21996 (estimate_loops_at_level): Likewise.
21997 (estimate_loops): Likewise.
21998 * shrink-wrap.c (try_shrink_wrapping): Likewise.
21999 (spread_components): Likewise.
22000 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
22001 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
22002 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
22003 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
22004 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
22005 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
22006 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
22007 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
22008 (mark_threaded_blocks): Likewise.
22009 (thread_through_all_blocks): Likewise.
22010 * tree-ssa.c (verify_ssa): Likewise.
22011 (execute_update_addresses_taken): Likewise.
22012 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
22014 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22016 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
22018 (post_order_compute): Likewise.
22019 (inverted_post_order_compute): Likewise.
22020 (pre_and_rev_post_order_compute_fn): Likewise.
22022 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22024 * genrecog.c (int_set::int_set): Explicitly construct our
22025 auto_vec base class.
22026 * vec.h (auto_vec::auto_vec): New constructor.
22028 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22030 * bitmap.h (class auto_bitmap): New constructor taking
22031 bitmap_obstack * argument.
22033 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22035 * bitmap.h (class auto_bitmap): Change type of m_bits to
22036 bitmap_head, and adjust ctor / dtor and member operators.
22038 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
22040 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
22041 when returned register mode doesn't match original mode.
22043 2017-05-12 Jeff Law <law@redhat.com>
22044 Jakub Jelinek <jakub@redhat.com>
22046 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
22047 we look for cc setter after the compare-elim changes.
22048 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
22049 within the vector to match what compare-elim now expects.
22050 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
22051 (xorsi3_flags, one_cmplsi2_flags): Likewise.
22053 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
22054 after the compare-elim changes.
22055 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
22056 the vector to match what compare-elim now expects.
22057 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
22058 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
22059 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
22060 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
22061 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
22063 * config/visium/visium.c (single_set_and_flags): Fix where
22064 we look for cc setter after the compare-elim changes.
22065 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
22066 with the vector to match what compare-elim now expects.
22067 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
22068 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
22069 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
22070 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
22071 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
22072 (neg<mode>2_insn_set_overflow): Likewise.
22074 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
22076 PR middle-end/79794
22077 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
22078 maybe_expand_insn call, set ops[0].target. If still set after call,
22079 set alt_rtl. Add extra arg to recursive calls.
22080 (extract_bit_field): Add alt_rtl argument. Pass to
22082 * expmed.h (extract_bit_field): Fix prototype.
22083 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
22084 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
22085 to extract_bit_field_calls.
22086 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
22087 Pass alt_rtl to extract_bit_field calls.
22088 * calls.c (store_unaligned_arguments_into_psuedos)
22089 load_register_parameters): Pass extra NULL to extract_bit_field calls.
22090 * optabs.c (maybe_legitimize_operand): Clear op->target when call
22092 * optabs.h (struct expand_operand): Add target bitfield.
22094 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22096 * compare-elim.c (try_eliminate_compare): Canonicalize
22097 operation with embedded compare to
22098 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
22099 (set (reg) (operation)].
22101 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
22103 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22106 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
22107 cost of adding a carry flag for ADC instruction.
22108 [case MINUS]: Ignore the cost of subtracting a carry flag
22109 for SBB instruction.
22111 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
22113 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
22115 * config/rs6000/bmiintrin.h: New file.
22116 * config/rs6000/bmi2intrin.h: New file.
22117 * config/rs6000/x86intrin.h: New file.
22119 2017-05-12 Jeff Law <law@redhat.com>
22121 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
22124 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
22126 PR middle-end/80707
22127 * tree-cfg.c: Remove cfg edges of unreachable case statements.
22129 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22131 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22132 early expansion of vector divide builtins.
22133 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
22134 builtins identified as having unsigned arguments.
22136 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22138 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
22139 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
22140 expansion of vector logical operations (and, andc, or, xor,
22143 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22145 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
22146 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
22148 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22150 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22151 early GIMPLE expansion of vector multiplies.
22153 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22155 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
22156 TARGET_HAVE_MOVT conditional.
22157 (movt splitter): Likewise.
22159 2017-05-12 Richard Biener <rguenther@suse.de>
22161 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
22162 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
22163 Fold all stmts not inplace.
22165 2017-05-12 Richard Biener <rguenther@suse.de>
22167 PR tree-optimization/80713
22168 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
22169 inserted_exprs bit for not removed stmts.
22171 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
22173 PR middle-end/69921
22174 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
22175 parallelized" attribute for parallelized OpenACC kernels.
22176 * omp-offload.c (execute_oacc_device_lower): Use it.
22178 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
22179 Set "oacc kernels" attribute.
22180 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
22181 parameter. Adjust all users.
22182 (oacc_fn_attrib_kernels_p): Remove function.
22183 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
22184 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
22185 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
22186 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
22187 assert "oacc kernels" attribute is set.
22189 2017-05-11 Carl Love <cel@us.ibm.com>
22191 * config/rs6000/rs6000-c: Add support for built-in functions
22192 vector unsigned char vec_popcnt (vector signed char)
22193 vector unsigned char vec_popcnt (vector unsigned char)
22194 vector unsigned short vec_popcnt (vector signed short)
22195 vector unsigned short vec_popcnt (vector unsigned short)
22196 vector unsigned int vec_popcnt (vector signed int)
22197 vector unsigned int vec_popcnt (vector unsigned int)
22198 vector unsigned long long vec_popcnt (vector signed long long)
22199 vector unsigned long long vec_popcnt (vector unsigned long long)
22200 vector signed long long vec_slo (vector signed long long,
22201 vector signed char)
22202 vector signed long long vec_slo (vector signed long long,
22203 vector unsigned char)
22204 vector unsigned long long vec_slo (vector unsigned long long,
22205 vector signed char)
22206 vector unsigned long long vec_slo (vector unsigned long long,
22207 vector unsigned char)
22208 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
22209 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
22210 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
22211 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
22212 * doc/extend.texi: Update the built-in documentation file for the
22213 new built-in functions.
22215 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22217 * attribs.h (sorted_attr_string): Move machine independent
22218 functions for target clone support from the i386 port to common
22219 code. Rename ix86_function_versions to common_function_versions.
22220 Rename make_name to make_unique_name.
22221 (common_function_versions): Likewise.
22222 (make_unique_name): Likewise.
22223 (make_dispatcher_decl): Likewise.
22224 (is_function_default_version): Likewise.
22225 * attribs.c (attr_strcmp): Likewise.
22226 (sorted_attr_string): Likewise.
22227 (common_function_versions): Likewise.
22228 (make_unique_name): Likewise.
22229 (make_dispatcher_decl): Likewise.
22230 (is_function_default_version): Likewise.
22231 * config/i386/i386.c (attr_strcmp): Likewise.
22232 (sorted_attr_string): Likewise.
22233 (ix86_function_versions): Likewise.
22234 (make_name): Likewise.
22235 (make_dispatcher_decl): Likewise.
22236 (is_function_default_version): Likewise.
22237 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
22239 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22242 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22243 Account for direct move costs for vec_construct of integer
22246 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
22249 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
22250 (UNSPEC_STX_ATOMIC): Ditto.
22251 (loaddi_via_sse): New insn.
22252 (storedi_via_sse): Ditto.
22253 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
22254 Update corresponding peephole2 patterns.
22255 (atomic_storedi_fpu): Ditto.
22257 2017-05-11 Julia Koval <julia.koval@intel.com>
22259 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
22260 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
22262 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
22263 (__builtin_ia32_rsqrt14ss_mask): New builtins.
22264 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
22266 2017-05-11 Nathan Sidwell <nathan@acm.org>
22268 * graphite-poly.c: Include dumpfile.h.
22270 * dumpfle.h (dump_function): Declare here ...
22271 * tree-dump.h (dump_function): ... not here.
22272 * dumpfile.c: #include tree-cfg.h.
22273 (dump_function): Move here from ...
22274 * tree-dump.c (dump_function): ... here.
22275 * gimplify.c: #include splay-tree.h, not tree-dump.h.
22276 * graphite-poly.c: Don't include tree-dump.h.
22277 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
22278 * print-tree.c: Likewise.
22279 * stor-layout.c: Likewise.
22280 * tree-nested.c: Likewise.
22282 * dumpfile.c (dump_start): Use TDF_FLAGS.
22283 (dump_enable_all): Fix TDF_KIND check thinko.
22285 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
22287 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22288 array entries to represent two legal parameterizations of the
22289 overloaded __builtin_cmpb function, as represented by the
22290 P6_OV_BUILTIN_CMPB constant.
22291 (altivec_resolve_overloaded_builtin): Add special case handling
22292 for the __builtin_cmpb function, as represented by the
22293 P6_OV_BUILTIN_CMPB constant.
22294 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
22295 (BU_P6_64BIT_2): New macro.
22296 (BU_P6_OVERLOAD_2): New macro
22297 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
22298 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
22299 (CMPB): Add overload support to represent both 32-bit and 64-bit
22300 compare-bytes function.
22301 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
22302 support for TARGET_CMPB.
22303 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
22304 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22305 documentation of the __builtin_cmpb overloaded built-in function.
22307 2017-05-11 Richard Biener <rguenther@suse.de>
22309 PR tree-optimization/80705
22310 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
22311 bases are not vectorizable.
22313 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22315 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
22316 when counting register pressure.
22318 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22320 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
22321 (iv_ca_more_deps): Renamed to ...
22322 (iv_ca_compare_deps): ... this.
22323 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
22325 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22327 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
22329 (determine_group_iv_costs): ... here.
22330 (find_inv_vars_cb): Record inv var if it's not recorded before.
22332 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22334 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
22335 (get_shiftadd_cost): Ditto.
22337 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22339 * tree-ssa-address.c: Include header file.
22340 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
22342 (add_to_parts): Refactor.
22343 (addr_to_parts): New parameter. Update use of move_hint_to_base.
22344 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
22347 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22349 PR tree-optimization/53090
22350 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
22352 (extract_cond_operands): Detect condition with IV on both sides
22353 and return COMP_IV_EXPR_2.
22354 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
22355 (rewrite_use_compare): Simplify by removing call to function
22356 extract_cond_operands.
22358 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22360 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
22361 (extract_cond_operands): Detect condition comparing against non-
22362 invariant bound and return appropriate enum value.
22363 (find_interesting_uses_cond): Update use of extract_cond_operands.
22364 Handle its return value accordingly.
22365 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
22367 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22369 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
22370 nonlinear iv_use computation in loop invariant sensitive way.
22372 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22374 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
22375 (find_iv_candidates): Call relate_compare_use_with_all_cands.
22377 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22379 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
22380 (dump_cand): Support iv_cand.inv_exprs.
22381 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
22383 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
22386 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22388 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
22390 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
22391 as local function. Include necessary header files.
22392 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
22394 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22396 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
22398 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22400 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
22401 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
22402 RSHIFT_EXPR and BIT_NOT_EXPR.
22404 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22406 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
22407 (adjust_setup_cost): New parameter supporting round up adjustment.
22408 (struct address_cost_data): Delete.
22409 (force_expr_to_var_cost): Don't bound cost with spill_cost.
22410 (split_address_cost, ptr_difference_cost): Delete.
22411 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
22412 (struct ainc_cost_data): New struct.
22413 (get_address_cost_ainc): New function.
22414 (get_address_cost, get_computation_cost): Reimplement.
22415 (determine_group_iv_cost_address): Record inv_expr for all uses of
22417 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
22418 (iv_ca_has_deps): Reimplemented to ...
22419 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
22421 (iv_ca_extend): Call iv_ca_more_deps.
22423 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22425 * tree-ssa-address.c (struct mem_address): Move to header file.
22426 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
22427 * tree-ssa-address.h (struct mem_address): Move from C file.
22428 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
22430 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22432 * tree-affine.h (aff_combination_type): New interface.
22433 (aff_combination_zero_p): Remove static.
22434 (aff_combination_const_p): New interface.
22435 (aff_combination_singleton_var_p): New interfaces.
22437 2017-05-11 Richard Biener <rguenther@suse.de>
22439 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
22440 Skip unreachable blocks and destinations.
22441 (eliminate): Move stmt removal and fixup ...
22442 (fini_eliminate): ... here. Skip inserted exprs.
22443 (pass_pre::execute): Move fini_pre after fini_eliminate.
22444 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
22445 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
22446 PRE to get rid of dead code that has invalid SSA form and
22447 split critical edges again.
22449 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22451 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22453 2017-05-11 Richard Biener <rguenther@suse.de>
22455 * passes.c (execute_function_todo): Verify loops if they are
22456 said to be up-to-date.
22457 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
22458 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
22460 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
22463 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
22464 handle calling assemble_external ourself.
22467 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
22468 modes with zero size. Enhance comment.
22470 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22472 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
22473 built-ins for vec_xl and vec_xst with short and char pointer
22476 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
22478 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
22479 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
22480 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
22481 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
22482 (_mm_maskz_min_round_ss): New intrinsics.
22483 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
22484 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
22485 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
22486 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
22487 (__builtin_ia32_minss_mask_round): New builtins.
22488 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22489 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22490 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
22492 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
22493 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
22495 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
22498 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
22500 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
22501 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
22502 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
22503 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
22504 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
22505 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
22506 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
22507 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22508 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
22509 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
22510 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
22511 (__builtin_ia32_mulss_mask_round): New builtins.
22512 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22513 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22514 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
22516 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
22517 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
22519 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
22522 2017-05-10 Julia Koval <julia.koval@intel.com>
22524 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
22525 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
22526 (_mm256_setr_m128i): New intrinsics.
22528 2017-05-10 Julia Koval <julia.koval@intel.com>
22530 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
22531 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
22532 (_mm_maskz_rcp14_ss): New intrinsics.
22533 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
22534 (__builtin_ia32_rcp14ss_mask): New builtins.
22535 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
22537 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
22539 PR tree-optimization/51513
22540 * tree-cfg.c (gimple_seq_unreachable_p): New function.
22541 (assert_unreachable_fallthru_edge_p): Use it.
22542 (group_case_labels_stmt): Likewise.
22543 * tree-cfg.h: Prototype it.
22544 * stmt.c: Include cfghooks.h and tree-cfg.h.
22545 (emit_case_dispatch_table) <gap_label>: New local variable.
22546 Use it to fill dispatch table gaps.
22547 Test for default_label before updating probabilities.
22548 (expand_case) <default_label>: Remove unneeded initialization.
22549 Test for unreachable default case statement and remove its edge.
22550 Set default_label accordingly.
22551 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
22553 2017-05-10 Carl Love <cel@us.ibm.com>
22555 * config/rs6000/rs6000-c: Add support for built-in functions
22556 vector signed char vec_neg (vector signed char)
22557 vector signed short int vec_neg (vector short int)
22558 vector signed int vec_neg (vector signed int)
22559 vector signed long long vec_neg (vector signed long long)
22560 vector float vec_neg (vector float)
22561 vector double vec_neg (vector double)
22562 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
22564 * config/rs6000/altivec.h: Add define for vec_neg
22565 * doc/extend.texi: Update the built-in documentation for the
22566 new built-in functions.
22568 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22570 PR tree-optimization/77644
22571 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
22573 2017-05-10 Nathan Sidwell <nathan@acm.org>
22575 * dumpfile.h (TDI_lang_all): New.
22576 (TDF_KIND): New. Renumber others
22577 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
22579 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
22581 (get_dump_file_name): Adjust suffix generation.
22582 (dump_enable_all): Use TDF_KIND.
22583 * doc/invoke.texi (-fdump-lang-all): Document.
22585 * dumpfile.h: Tabify.
22587 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
22590 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
22591 Move member access before delete.
22593 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
22595 * tree-inline.c (expand_call_inline): Split block at stmt
22598 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
22601 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
22602 are now unused after splitting mov{sf,sd}_hardfloat.
22603 (f32_lr2): Likewise.
22604 (f32_lm): Likewise.
22605 (f32_lm2): Likewise.
22606 (f32_li): Likewise.
22607 (f32_li2): Likewise.
22608 (f32_lv): Likewise.
22609 (f32_sr): Likewise.
22610 (f32_sr2): Likewise.
22611 (f32_sm): Likewise.
22612 (f32_sm2): Likewise.
22613 (f32_si): Likewise.
22614 (f32_si2): Likewise.
22615 (f32_sv): Likewise.
22616 (f32_dm): Likewise.
22617 (f32_vsx): Likewise.
22618 (f32_av): Likewise.
22619 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
22620 For movsf, order stores so the VSX stores occur before the GPR
22621 store which encourages the register allocator to use a traditional
22622 FPR instead of a GPR. For movsd, order the stores so that the GPR
22623 store comes before the VSX stores to allow the power6 to work.
22624 This is due to the power6 not having a 32-bit integer store
22625 instruction from a FPR.
22626 (movsf_hardfloat): Likewise.
22627 (movsd_hardfloat): Likewise.
22629 2017-05-09 Martin Sebor <msebor@redhat.com>
22631 PR translation/80280
22632 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
22635 PR translation/80280
22636 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
22637 data member added in r247778.
22638 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
22640 2017-05-09 Nathan Sidwell <nathan@acm.org>
22642 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
22644 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
22647 2017-05-09 Marek Polacek <polacek@redhat.com>
22649 * doc/invoke.texi: Fix typo.
22651 2017-05-09 Richard Biener <rguenther@suse.de>
22653 * tree-vrp.c (vrp_val_is_max): Adjust comment.
22654 (vrp_val_is_min): Likewise.
22655 (set_value_range_to_value): Likewise.
22656 (set_value_range_to_nonnegative): Likewise.
22657 (gimple_assign_nonzero_p): Likewise.
22658 (gimple_stmt_nonzero_p): Likewise.
22659 (vrp_int_const_binop): Likewise. Remove unreachable case.
22660 (adjust_range_with_scev): Adjust comments.
22661 (compare_range_with_value): Likewise.
22662 (extract_range_from_phi_node): Likewise.
22663 (test_for_singularity): Likewise.
22665 2017-05-09 Richard Biener <rguenther@suse.de>
22667 * tree-vrp.c (get_single_symbol): Add assert that we don't
22668 get overflowed constants as invariant part.
22669 (compare_values_warnv): Add comment before the TREE_NO_WARNING
22670 checks. Use wi::cmp instead of recursing for integer constants.
22671 (compare_values): Just ignore whether we assumed undefined
22672 overflow instead of failing the compare.
22673 (extract_range_for_var_from_comparison_expr): Add comment before the
22674 TREE_NO_WARNING sets.
22675 (test_for_singularity): Likewise.
22676 (extract_range_from_comparison): Do not disable optimization
22677 when we assumed undefined overflow.
22678 (extract_range_basic): Remove init of unused var.
22680 2017-05-09 Richard Biener <rguenther@suse.de>
22682 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
22683 (extract_range_from_multiplicative_op_1): Adjust.
22684 (extract_range_from_binary_expr_1): Use int_const_binop.
22686 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
22689 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
22690 rs6000_store_data_bypass_p in seven define_bypass directives and
22691 in several comments.
22692 * config/rs6000/rs6000-protos.h: Add prototype for
22693 rs6000_store_data_bypass_p function.
22694 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
22695 function implements slightly different (rs6000-specific) semantics
22696 than store_data_bypass_p, returning false rather than aborting
22697 with assertion error when arguments do not satisfy the
22698 requirements of store data bypass.
22699 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
22700 rs6000_store_data_bypass_p.
22702 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
22704 * config/xtensa/xtensa-protos.h
22705 (xtensa_initial_elimination_offset): New declaration.
22706 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22707 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
22708 macro definition, add case for FRAME_POINTER_REGNUM when
22709 FRAME_GROWS_DOWNWARD.
22710 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
22711 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
22712 xtensa_initial_elimination_offset.
22714 2017-05-08 Nathan Sidwell <nathan@acm.org>
22716 * doc/invoke.texi: Alphabetize -fdump options.
22718 2017-05-08 Martin Sebor <msebor@redhat.com>
22720 PR translation/80280
22721 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
22723 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
22725 * target.def (compute_frame_layout): New optional target hook.
22726 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
22727 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
22728 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
22730 * reload1.c (verify_initial_elim_offsets): Likewise.
22731 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
22732 (use_simple_return_p): Call arm_compute_frame_layout if needed.
22733 (arm_get_frame_offsets): Split up into this ...
22734 (arm_compute_frame_layout): ... and this function.
22736 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
22738 * config/aarch64/constraints.md (Usa): New constraint.
22739 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
22741 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
22743 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
22744 with_multilib_list after it has been checked.
22746 2017-05-08 Richard Biener <rguenther@suse.de>
22748 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
22749 (bitmap_set_subtract_values): Likewise.
22751 2017-05-08 Richard Biener <rguenther@suse.de>
22753 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
22754 (gimple_assign_nonzero): ... this and remove strict_overflow_p
22756 (gimple_stmt_nonzero_warnv_p): Rename to ...
22757 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
22759 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
22760 (extract_range_basic): Adjust, do not disable propagation on
22761 strict overflow sensitive simplification.
22762 (vrp_visit_cond_stmt): Likewise.
22764 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
22766 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
22767 body size unconditionally.
22769 2017-05-07 Jeff Law <law@redhat.com>
22772 2017-05-06 Jeff Law <law@redhat.com>
22773 PR tree-optimization/78496
22774 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
22777 PR tree-optimization/78496
22778 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
22779 (simplify_stmt_using_ranges): Call it.
22780 (vrp_dom_walker::before_dom_children): Extract equivalences
22781 from an ASSERT_EXPR with an equality comparison against a
22784 2017-05-06 Jeff Law <law@redhat.com>
22786 PR tree-optimization/78496
22787 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
22790 PR tree-optimization/78496
22791 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
22792 (simplify_stmt_using_ranges): Call it.
22793 (vrp_dom_walker::before_dom_children): Extract equivalences
22794 from an ASSERT_EXPR with an equality comparison against a
22797 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
22799 * lra-constraints.c (lra_copy_reg_equiv): New function.
22800 (split_reg): Use it to copy equivalence information from the
22801 original register to the spill register.
22803 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
22805 PR rtl-optimization/75964
22806 * simplify-rtx.c (simplify_const_relational_operation): Remove
22807 invalid handling of comparisons of integer ABS.
22809 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
22811 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
22812 initialize to zero.
22813 (init_regs): Remove declaration.
22814 (function_arg_advance_32): Initialize error_p as boolean variable.
22816 2017-05-05 Nathan Sidwell <nathan@acm.org>
22818 * store-motion.c (remove_reachable_equiv_notes): Reformat long
22819 lines. Use for (;;).
22821 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22823 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
22824 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
22825 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
22826 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
22827 VF=2 that require versioning.
22829 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22831 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
22834 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22836 * diagnostic.h (diagnostic_override_option_index): Convert from
22837 macro to inline function.
22839 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22841 * diagnostic.c (last_module_changed_p): New function.
22842 (set_last_module): New function.
22843 (diagnostic_report_current_module): Convert macro usage to
22844 the above functions.
22845 * diagnostic.h (diagnostic_context::last_module): Strengthen
22846 from const line_map * to const line_map_ordinary *.
22847 (diagnostic_last_module_changed): Delete macro.
22848 (diagnostic_set_last_module): Delete macro.
22850 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22852 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
22853 with diagnostic_report_diagnostic.
22854 (diagnostic_n_impl_richloc): Likewise.
22855 * diagnostic.h (report_diagnostic): Delete macro.
22856 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
22857 with diagnostic_report_diagnostic.
22858 * substring-locations.c (format_warning_va): Likewise.
22860 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22862 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
22863 save/restor of format_spec. Move option-printing code to...
22864 (print_option_information): ...this new function, and
22865 reimplement by simply printing to the pretty_printer,
22866 rather than appending to the format string.
22868 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22870 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
22871 handling logic into...
22872 (update_effective_level_from_pragmas): ...this new function.
22874 2017-05-04 Andrew Waterman <andrew@sifive.com>
22876 * config/riscv/riscv.opt (mstrict-align): New option.
22877 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
22878 (SLOW_UNALIGNED_ACCESS): Define.
22879 (riscv_slow_unaligned_access): Declare.
22880 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
22882 (riscv_slow_unaligned_access): New variable.
22883 (rocket_tune_info): Set slow_unaligned_access to true.
22884 (optimize_size_tune_info): Set slow_unaligned_access to false.
22885 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
22886 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
22887 (riscv_option_override): Set riscv_slow_unaligned_access.
22888 * doc/invoke.texi: Add -mstrict-align to RISC-V.
22890 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
22892 * config/riscv/riscv.md: Unify indentation.
22894 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22899 * config/rs6000/rs6000.md (u code attribute): Add FIX and
22901 (extendsi<mode>2): Add support for doing sign extension via
22902 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
22903 don't have ISA 3.0 instructions.
22904 (extendsi<mode>2 splitter): Likewise.
22905 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
22906 generate the normal insns since SImode can now go in vector
22907 registers. Disallow the special UNSPECs needed for previous
22908 machines to hide SImode being used. Add new insns
22909 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
22910 (fix_trunc<mode>si2_stfiwx): Likewise.
22911 (fix_trunc<mode>si2_internal): Likewise.
22912 (fixuns_trunc<mode>si2): Likewise.
22913 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22914 (fctiw<u>z_<mode>_smallint): Likewise.
22915 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
22916 of floating point to 32-bit integer from doing a direct move to
22917 the GPR registers to do a store.
22918 (fctiwz_<mode>): Break long line.
22920 2017-05-05 Bin Cheng <bin.cheng@arm.com>
22922 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
22923 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
22924 (addr_list, addr_offset_valid_p): New.
22925 (split_address_groups): Check offset validity with above function.
22926 (gt-tree-ssa-loop-ivopts.h): Include header file.
22928 2017-05-05 Nathan Sidwell <nathan@acm.org>
22930 * config.gcc (arm*-*-*): Add missing 'fi'.
22932 2017-05-05 Steve Ellcey <sellcey@cavium.com>
22934 * doc/invoke.texi (-fopt-info): Explicitly say order of options
22935 included in -fopt-info does not matter.
22936 * doc/optinfo.texi (-fopt-info): Fix description of default
22937 behavour. Explicitly say order of options included in -fopt-info
22940 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22942 * config.gcc: Allow combinations of aprofile and rmprofile values for
22943 --with-multilib-list.
22944 * config/arm/t-multilib: New file.
22945 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
22946 variables. Remove setting of ISA and floating-point ABI in
22947 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
22948 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
22949 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
22950 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
22952 * config/arm/t-rmprofile: Likewise except for the matches changes.
22953 * doc/install.texi (--with-multilib-list): Document the combination of
22954 aprofile and rmprofile values and warn about pitfalls in doing that.
22956 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22958 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
22959 (movdi_aarch64): Likewise.
22961 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22963 PR tree-optimization/80632
22964 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
22966 (build_arrays): Initialize it for virtual phis.
22967 (fix_phi_nodes): Use it for virtual phis.
22969 PR tree-optimization/80558
22970 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
22971 [x, y] op z into [x op, y op z] for op & or | if conditions
22974 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
22975 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22978 * config/arm/arm.md (use_literal_pool): Remove.
22979 (64-bit immediate split): No longer takes cost into consideration
22980 if arm_disable_literal_pool is enabled.
22981 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
22982 used when arm_disable_literal_pool is enabled.
22983 (arm_max_const_double_inline_cost): Remove use of
22984 arm_disable_literal_pool.
22985 (push_minipool_fix): Add assert.
22986 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
22987 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
22988 (no_literal_pool_sf_immediate): New.
22990 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22992 PR tree-optimization/80613
22993 * tree-ssa-dce.c (propagate_necessity): Remove cases for
22994 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
22996 2017-05-05 Richard Biener <rguenther@suse.de>
22998 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
23000 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
23002 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
23003 of this flag from insn conditions due to removal from r247495.
23005 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23007 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
23009 (arm_early_store_addr_dep_ptr): Likewise.
23010 * config/arm/aarch-common-protos.h
23011 (arm_early_load_addr_dep_ptr): Add prototype.
23012 (arm_early_store_addr_dep_ptr): Likewise.
23013 * config/arm/cortex-a53.md: Add new bypasses.
23015 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23017 * tree.c (next_type_uid): Change type to unsigned.
23018 (type_hash_canon): Decrement back next_type_uid if
23019 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
23020 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
23023 2017-05-04 Martin Sebor <msebor@redhat.com>
23025 * builtins.c: Fix a trivial typo in a comment.
23027 PR middle-end/79234
23028 * builtins.c (check_sizes): Adjust to handle reading past the end.
23029 Avoid printing excessive upper bound of ranges. Use %E to print
23030 tree nodes instead of converting them to %wu.
23031 (expand_builtin_memchr): New function.
23032 (compute_dest_size): Rename...
23033 (compute_objsize): ...to this.
23034 (expand_builtin_memcpy): Adjust.
23035 (expand_builtin_mempcpy): Adjust.
23036 (expand_builtin_strcat): Adjust.
23037 (expand_builtin_strcpy): Adjust.
23038 (check_strncat_sizes): Adjust.
23039 (expand_builtin_strncat): Adjust.
23040 (expand_builtin_strncpy): Adjust and simplify.
23041 (expand_builtin_memset): Adjust.
23042 (expand_builtin_bzero): Adjust.
23043 (expand_builtin_memcmp): Adjust.
23044 (expand_builtin): Handle memcmp.
23045 (maybe_emit_chk_warning): Check strncat just once.
23047 2017-05-04 Martin Sebor <msebor@redhat.com>
23049 PR preprocessor/79214
23050 PR middle-end/79222
23051 PR middle-end/79223
23052 * builtins.c (check_sizes): Add inlining context and issue
23053 warnings even when -Wno-system-headers is set.
23054 (check_strncat_sizes): Same.
23055 (expand_builtin_strncat): Same.
23056 (expand_builtin_memmove): New function.
23057 (expand_builtin_stpncpy): Same.
23058 (expand_builtin): Handle memmove and stpncpy.
23060 2017-05-04 Bin Cheng <bin.cheng@arm.com>
23062 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
23063 which is not used any more.
23065 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23067 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
23069 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23071 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
23072 (cortexa53_tunings): Likewise.
23073 (cortexa57_tunings): Likewise.
23074 (cortexa72_tunings): Likewise.
23075 (cortexa73_tunings): Likewise.
23077 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23079 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
23080 Set loop alignment to 8.
23082 2017-05-04 Martin Sebor <msebor@redhat.com>
23084 PR translation/80280
23085 * builtins.c (expand_builtin_object_size): Add missing quoting to
23086 %D and like directives.
23087 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
23088 (hsa_type_for_tree_type): Same.
23089 (verify_function_arguments): Same.
23090 * symtab.c (symbol_table::change_decl_assembler_name): Same.
23091 * varasm.c (get_section): Same.
23094 2017-05-04 Martin Sebor <msebor@redhat.com>
23096 PR translation/80280
23097 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
23099 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23101 * config/aarch64/aarch64.c (generic_addrcost_table):
23102 Change HI/TI mode setting.
23104 2017-05-04 Martin Jambor <mjambor@suse.cz>
23106 PR tree-optimization/80622
23107 * tree-sra.c (comes_initialized_p): New function.
23108 (build_accesses_from_assign): Only set write lazily when
23109 comes_initialized_p is false.
23110 (analyze_access_subtree): Use comes_initialized_p.
23111 (propagate_subaccesses_across_link): Assert !comes_initialized_p
23112 instead of testing for PARM_DECL.
23114 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23116 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
23117 constraint on operand 0 to allow more general addressing modes.
23118 Adjust output template.
23119 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
23121 * config/aarch64/aarch64-protos.h
23122 (aarch64_address_valid_for_prefetch_p): Declare prototype.
23123 * config/aarch64/constraints.md (Dp): New address constraint.
23124 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
23127 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
23129 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
23130 update use of estimate_ipcp_clone_size_and_time.
23131 (estimate_local_effects): Update use of
23132 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
23133 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
23134 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
23135 Return nonspecialized time.
23137 2017-05-04 Richard Biener <rguenther@suse.de>
23139 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
23140 for the last VUSE which def dominates the PHI. Directly call
23142 (get_continuation_for_phi_1): Remove.
23144 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
23146 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
23147 to explain the use of truncating division. Cap the number of
23148 iterations to the maximum given by nb_iterations_upper_bound,
23151 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23153 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
23154 * configure: Regenerate.
23155 * config.in: Regenerate.
23156 * config/i386/driver-mingw32.c: new file.
23157 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
23158 * config.host: Link driver-mingw32.o on MinGW host.
23159 * doc/install.texi: Document new --enable-mingw-wildcard configure
23162 2017-05-04 Marek Polacek <polacek@redhat.com>
23164 PR tree-optimization/80612
23165 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
23167 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23168 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
23170 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
23171 (movt splitter): Likewise.
23172 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
23173 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
23174 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
23175 block for Thumb-1 with MOVT.
23176 (thumb2_legitimate_address_p): Move code block ...
23177 (can_avoid_literal_pool_for_label_p): ... into this new function.
23178 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
23180 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
23181 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
23182 "M-profile targets with the MOVT instruction".
23184 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23186 * config/arm/arm-builtins.c (arm_init_builtins): Rename
23187 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
23188 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
23190 2017-05-04 Martin Liska <mliska@suse.cz>
23192 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
23193 variable cond_code.
23195 2017-05-04 Richard Biener <rguenther@suse.de>
23197 * tree.c (array_at_struct_end_p): Handle arrays at struct
23198 end with flexarrays more conservatively. Refactor and treat
23199 arrays of arrays or aggregates more strict. Fix
23200 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
23201 * tree.c (array_at_struct_end_p): Adjust prototype.
23202 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
23203 * gimple-fold.c (get_range_strlen): Likewise.
23204 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
23206 2017-05-04 Richard Biener <rguenther@suse.de>
23208 PR tree-optimization/31130
23209 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
23211 (supports_overflow_infinity): Likewise.
23212 (is_negative_overflow_infinity): Likewise.
23213 (is_positive_overflow_infinity): Likewise.
23214 (is_overflow_infinity): Likewise.
23215 (stmt_overflow_infinity): Likewise.
23216 (overflow_infinity_range_p): Likewise.
23217 (usable_range_p): Remove as always returning true.
23218 (make_overflow_infinity): Remove.
23219 (negative_overflow_infinity): Likewise.
23220 (positive_overflow_infinity): Likewise.
23221 (avoid_overflow_infinity): Likewise.
23222 (set_value_range): Adjust accordingly.
23223 (set_value_range_to_nonnegative): Likewise, remove now unused
23224 overflow_infinity arg.
23225 (vrp_operand_equal_p): Adjust.
23226 (update_value_range): Likewise.
23227 (range_int_cst_singleton_p): Likewise.
23228 (operand_less_p): Likewise.
23229 (compare_values_warnv): Likewise.
23230 (extract_range_for_var_from_comparison_expr): Likewise.
23231 (vrp_int_const_binop): Likewise.
23232 (zero_nonzero_bits_from_vr): Likewise.
23233 (extract_range_from_multiplicative_op_1): Likewise.
23234 (extract_range_from_binary_expr_1): Likewise.
23235 (extract_range_from_unary_expr): Likewise.
23236 (extract_range_from_comparison): Likewise.
23237 (extract_range_basic): Likewise.
23238 (adjust_range_with_scev): Likewise.
23239 (compare_ranges): Likewise.
23240 (compare_range_with_value): Likewise.
23241 (dump_value_range): Likewise.
23242 (test_for_singularity): Likewise, remove strict_overflow_p parameter
23244 (simplify_cond_using_ranges): Adjust.
23246 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23248 * brig-builtins.def: Added a builtin for class_f64.
23249 * builtin-types.def: Added a builtin type needed by class_f64.
23251 2017-05-03 Jason Merrill <jason@redhat.com>
23253 * timevar.def: Add TV_CONSTEXPR.
23255 2017-05-03 David Malcolm <dmalcolm@redhat.com>
23257 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
23259 2017-05-03 Martin Jambor <mjambor@suse.cz>
23261 * ipa-prop.c (ipa_update_after_lto_read): Removed.
23262 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
23263 * ipa-cp.c (ipcp_propagate_stage): Do not call
23264 ipa_update_after_lto_read.
23265 * ipa-inline.c (ipa_inline): Likewise.
23267 2017-05-03 Martin Jambor <mjambor@suse.cz>
23269 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
23270 tag. Added a default constructor and a destructor.
23271 (ipa_edge_args_sum_t): New class;
23272 (ipa_edge_args_sum): Declare.
23273 (ipa_edge_args_vector): Remove declaration.
23274 (IPA_EDGE_REF): Use ipa_edge_args_sum.
23275 (ipa_free_edge_args_substructures): Remove declaration.
23276 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
23277 (ipa_edge_args_info_available_for_edge_p): Likewise.
23278 * ipa-prop.c (ipa_edge_args_vector): Removed.
23279 (edge_removal_hook_holder): Likewise.
23280 (edge_duplication_hook_holder): Likewise.
23281 (ipa_edge_args_sum): New variable.
23282 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
23283 ipa_edge_args_vector.
23284 (ipa_free_edge_args_substructures): Likewise.
23285 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
23286 ipa_edge_args_vector.
23287 (ipa_edge_removal_hook): Turned into method
23288 ipa_edge_args_sum_t::remove.
23289 (ipa_edge_duplication_hook): Turned into method
23290 ipa_edge_args_sum_t::duplicate.
23291 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
23292 registering edge hooks.
23293 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
23294 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
23295 ipa_edge_args_sum instead of ipa_edge_args_vector.
23296 * ipa-profile.c (ipa_profile): Likewise.
23298 2017-05-03 Martin Jambor <mjambor@suse.cz>
23300 * symbol-summary.h (function_summary): New method exists.
23301 (function_summary::symtab_removal): Deallocate through release.
23302 (call_summary): New class.
23303 (gt_ggc_mx): New overload.
23304 (gt_pch_nx): Likewise.
23305 (gt_pch_nx): Likewise.
23307 2017-05-03 Jeff Law <law@redhat.com>
23309 PR tree-optimization/78496
23310 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
23311 from simplify_cond_using_ranges. Split off code to walk
23312 backwards through casts into ...
23313 (simplify_cond_using_ranges_2): New function.
23314 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
23315 (execute_vrp): After identifying jump threads, call
23316 simplify_cond_using_ranges_2.
23318 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
23321 * ipa-inline.h (inline_summary): Add ctor.
23322 (create_ggc): Do not use ggc_cleared_alloc.
23324 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
23325 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23327 * gcc.c (handle_braces): Support escaping in switch matching
23329 * doc/invoke.texi (Spec Files): Document it.
23330 Remove superfluous @code markup in items.
23332 2017-05-03 David Malcolm <dmalcolm@redhat.com>
23334 * diagnostic-show-locus.c (struct column_range): New struct.
23335 (get_affected_columns): New function.
23336 (get_printed_columns): New function.
23337 (struct correction): New struct.
23338 (correction::ensure_capacity): New function.
23339 (correction::ensure_terminated): New function.
23340 (struct line_corrections): New struct.
23341 (line_corrections::~line_corrections): New dtor.
23342 (line_corrections::add_hint): New function.
23343 (layout::print_trailing_fixits): Reimplement in terms of the new
23345 (selftest::test_overlapped_fixit_printing): New function.
23346 (selftest::diagnostic_show_locus_c_tests): Call it.
23348 2017-05-03 Nathan Sidwell <nathan@acm.org>
23350 Canonicalize canonical type hashing
23351 * tree.h (type_hash_canon_hash): Declare.
23352 * tree.c (type_hash_list, attribute_hash_list): Move into
23353 type_hash_canon_hash.
23354 (build_type_attribute_qual_variant): Break out hash code calc into
23355 type_hash_canon_hash.
23356 (type_hash_canon_hash): New. Generic type hash computation.
23357 (build_range_type_1, build_array_type_1, build_function_type,
23358 build_method_type_directly, build_offset_type, build_complex_type,
23359 make_vector_type): Call it.
23361 2017-05-03 Richard Biener <rguenther@suse.de>
23363 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
23364 When all DRs have unknown misaligned do not always peel
23365 when there is a store but apply the same costing model as if
23366 there were only loads.
23368 2017-05-03 Richard Biener <rguenther@suse.de>
23371 PR tree-optimization/80492
23372 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
23373 compare_base_decls returning dont-know properly.
23375 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
23377 * config/arm/iterators.md (CCSI): New mode iterator.
23378 (arch): New mode attribute.
23379 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
23380 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
23381 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
23382 code iterator for success result mode.
23383 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
23384 the corresponding new insn generators.
23386 2017-05-03 Bin Cheng <bin.cheng@arm.com>
23389 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23390 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23392 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
23394 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
23395 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
23396 (DDR_A): Wrap DDR argument in brackets.
23397 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
23398 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
23399 (DDR_REVERSED_P): Likewise.
23401 2017-05-03 Jakub Jelinek <jakub@redhat.com>
23403 PR tree-optimization/79472
23404 * tree-switch-conversion.c (struct switch_conv_info): Add
23405 contiguous_range and default_case_nonstandard fields.
23406 (collect_switch_conv_info): Compute contiguous_range and
23407 default_case_nonstandard fields, don't clear final_bb if
23408 contiguous_range and only the default case doesn't have the required
23410 (check_all_empty_except_final): Set default_case_nonstandard instead
23411 of failing if contiguous_range and the default case doesn't have empty
23413 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
23414 and only the default case doesn't have the required constants. Skip
23416 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
23417 if default_case_nonstandard.
23418 (build_constructors): Build constant 1 just once. Assert that default
23419 values aren't inserted in between cases if contiguous_range. Skip
23421 (build_arrays): Skip virtual phis.
23422 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
23423 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
23424 Handle virtual phis.
23425 (gen_inbound_check): Handle default_case_nonstandard case.
23426 (process_switch): Adjust check_final_bb caller. Call
23427 gather_default_values with the first non-default case instead of
23428 default case if default_case_nonstandard.
23430 2017-05-02 Nathan Sidwell <nathan@acm.org>
23432 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
23433 check. Fix formatting.
23435 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
23437 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
23438 errors when comparing specialized and unspecialized times.
23440 2017-05-02 David Malcolm <dmalcolm@redhat.com>
23442 * diagnostic-show-locus.c
23443 (layout::should_print_annotation_line_p): Make private.
23444 (layout::print_annotation_line): Make private.
23445 (layout::annotation_line_showed_range_p): Make private.
23446 (layout::show_ruler): Make private.
23447 (layout::print_source_line): Make private. Pass in line and
23448 line_width, rather than calling location_get_source_line. Drop
23450 (layout::print_leading_fixits): New method.
23451 (layout::print_any_fixits): Rename to...
23452 (layout::print_trailing_fixits): ...this, and make private.
23453 Don't print newline fixits.
23454 (diagnostic_show_locus): Move logic for printing one row into...
23455 (layout::print_line): ...this new function. Move the
23456 location_get_source_line call and error-handling from
23457 print_source_line to here. Call print_leading_fixits, and rename
23458 print_any_fixits to print_trailing_fixits.
23459 (selftest::test_fixit_insert_containing_newline): Update now that
23460 newlines are partially supported.
23461 (selftest::test_fixit_insert_containing_newline_2): New test.
23462 (selftest::test_fixit_replace_containing_newline): Update comments.
23463 (selftest::diagnostic_show_locus_c_tests): Call the new test.
23464 * edit-context.c (class added_line): New class.
23465 (class edited_line): Describe newline handling in comment.
23466 (edited_line::actually_edited_p): New method.
23467 (edited_line::print_content): Delete redundant decl.
23468 (edited_line::m_predecessors): New field.
23469 (edited_file::print_content): Call edited_line::print_content.
23470 (edited_file::print_diff): Update to support newlines.
23471 (edited_file::print_diff_hunk): Likewise.
23472 (edited_file::print_run_of_changed_lines): New function.
23473 (edited_file::print_diff_line): Convert to...
23474 (print_diff_line): ...this.
23475 (edited_file::get_effective_line_count): New function.
23476 (edited_line::edited_line): Initialize new field m_predecessors.
23477 (edited_line::~edited_line): Clean up m_predecessors.
23478 (edited_line::apply_fixit): Handle newlines.
23479 (edited_line::get_effective_line_count): New function.
23480 (edited_line::print_content): New function.
23481 (edited_line::print_diff_lines): New function.
23482 (selftest::test_applying_fixits_insert_containing_newline): New
23484 (selftest::test_applying_fixits_replace_containing_newline): New
23486 (selftest::insert_line): New function.
23487 (selftest::test_applying_fixits_multiple_lines): Add example of
23489 (selftest::edit_context_c_tests): Call the new tests.
23491 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23493 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
23494 parameter cand. Update dump information.
23495 (get_computation_cost): Update uses.
23497 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23499 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
23500 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
23501 (get_computation_at, rewrite_use_address): Update use of
23502 get_computation_aff.
23504 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23506 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
23507 (get_computation): Delete.
23508 (get_computation_cost): Implement like get_computation_cost_at.
23509 Use get_computation_at.
23510 (get_computation_cost_at): Delete.
23511 (rewrite_use_nonlinear_expr): Use get_computation_at.
23512 (rewrite_use_compare, remove_unused_ivs): Ditto.
23514 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23516 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
23518 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23520 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
23521 (ivopts_global_cost_for_size): Rename parameter and update uses.
23522 (iv_ca_recount_cost): Update uses.
23523 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
23524 candidates seperately in n_invs and n_cands.
23525 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
23527 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23529 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
23530 (find_inv_vars_cb): New.
23531 (find_depends): Renamed to ...
23532 (find_inv_vars): ... this.
23533 (add_candidate_1, force_var_cost): Call find_inv_vars.
23534 (split_address_cost, determine_group_iv_cost_cond): Ditto.
23536 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23538 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
23539 inv_vars. Add inv_exprs.
23540 (struct iv_cand): Rename depends_on to inv_vars.
23541 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
23542 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
23543 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
23544 (dump_cand): Dump inv_vars.
23545 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
23546 (record_invariant, find_depends, add_candidate_1): Ditto.
23547 (set_group_iv_cost, force_var_cost): Ditto.
23548 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
23549 (get_computation_cost_at, get_computation_cost): Ditto.
23550 (determine_group_iv_cost_generic): Ditto.
23551 (determine_group_iv_cost_address): Ditto.
23552 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
23553 (determine_group_iv_costs): Ditto.
23554 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
23555 (iv_ca_set_remove_invariants): Renamed to ...
23556 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
23557 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
23558 (iv_ca_set_add_invariants): Renamed to ...
23559 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
23560 (iv_ca_set_cp): Use iv_ca_set_add_invs.
23561 (iv_ca_has_deps): Support inv_vars and inv_exprs.
23562 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
23563 (create_new_ivs): Remove useless dump.
23565 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23567 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
23569 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
23570 (iv_ca_set_no_cp, create_new_iv): Ditto.
23572 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23574 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23576 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23578 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
23579 function tree_check2.
23581 2017-05-02 Martin Liska <mliska@suse.cz>
23583 * doc/gcov.texi: Add missing preposition.
23584 * gcov.c (function_info::function_info): Properly fill up
23585 all member variables.
23587 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23589 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
23591 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23593 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
23595 2017-05-02 Martin Liska <mliska@suse.cz>
23598 * lto-streamer-in.c (lto_read_tree_1): Remove
23599 LTO_STREAMER_DEBUG.
23600 * lto-streamer.c (struct tree_hash_entry): Likewise.
23601 (struct tree_entry_hasher): Likewise.
23602 (tree_entry_hasher::hash): Likewise.
23603 (tree_entry_hasher::equal): Likewise.
23604 (lto_streamer_init): Likewise.
23605 (lto_orig_address_map): Likewise.
23606 (lto_orig_address_get): Likewise.
23607 (lto_orig_address_remove): Likewise.
23608 * lto-streamer.h: Likewise.
23609 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
23610 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
23612 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
23614 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
23615 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
23616 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
23617 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
23618 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
23619 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
23620 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
23621 (mm_maskz_sub_ss): New intrinsics.
23622 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23623 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23624 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
23625 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
23626 (__builtin_ia32_subss_mask_round): New builtins.
23627 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23628 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23629 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
23631 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
23632 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23634 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23637 2017-05-02 Martin Jambor <mjambor@suse.cz>
23639 PR tree-optimization/78687
23640 * tree-sra.c (access): New field parent.
23641 (process_subtree_disqualification): New function.
23642 (disqualify_candidate): Call it.
23643 (build_accesses_from_assign): Reset write flag if creating an
23645 (build_access_subtree): Fill in parent field and also prpagate
23646 down grp_write flag.
23647 (create_artificial_child_access): New parameter set_grp_write, set
23648 grp_write to its value.
23649 (propagate_subaccesses_across_link): Also propagate grp_write flag
23651 (propagate_all_subaccesses): Push the closest parent back to work
23652 queue if add_access_to_work_queue returned true.
23654 2017-05-02 Richard Biener <rguenther@suse.de>
23656 * common.opt (fstrict-overflow): Alias negative to fwrapv.
23657 * doc/invoke.texi (fstrict-overflow): Remove all traces of
23658 -fstrict-overflow documentation.
23659 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
23660 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
23661 flag_strict_overflow.
23662 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
23663 * lto-opts.c (lto_write_options): Do not stream it.
23664 * lto-wrapper.c (merge_and_complain): Do not handle it.
23665 * opts.c (default_options_table): Do not set -fstrict-overflow.
23666 (finish_options): Likewise do not clear it when sanitizing.
23667 * simplify-rtx.c (simplify_const_relational_operation): Do not
23668 test flag_strict_overflow.
23670 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
23672 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
23673 using enabled attribute.
23674 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
23675 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
23676 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
23677 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
23678 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
23679 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
23680 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
23681 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
23682 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
23683 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
23685 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
23687 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
23689 2017-05-02 Richard Biener <rguenther@suse.de>
23691 PR tree-optimization/80591
23693 2017-04-10 Richard Biener <rguenther@suse.de>
23695 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23698 2017-05-02 Richard Biener <rguenther@suse.de>
23700 PR tree-optimization/80549
23701 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
23702 (cleanup_tree_cfg_noloop): Create forwarders to known loop
23703 headers if they do not have a preheader.
23705 2017-05-02 Martin Liska <mliska@suse.cz>
23708 * common.opt: Fix typo.
23709 * doc/invoke.texi: Likewise.
23711 2017-05-01 Jan Beulich <jbeulich@suse.com>
23713 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
23714 swapping, add (x,x,m,x,n) alternative.
23716 2017-05-01 Nathan Sidwell <nathan@acm.org>
23718 * calls.c (combine_pending_stack_adjustment_and_call): Remove
23719 unnecessary unadjusted_alignment check.
23721 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23724 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
23726 * gimplify.c (gimplify_cilk_detach): New function.
23727 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
23728 * tree-core.h: Document EXPR_CILK_SPAWN.
23729 * tree.h (EXPR_CILK_SPAWN): Define.
23731 2017-05-01 David Malcolm <dmalcolm@redhat.com>
23733 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
23734 to use new fixit_hint representation, using the "replace" logic.
23735 (get_line_span_for_fixit_hint): Likewise.
23736 (layout::print_any_fixits): Likewise.
23737 (selftest::test_one_liner_many_fixits): Rename to...
23738 (selftest::test_one_liner_many_fixits_1): ...this, and update
23739 comment and expected output to reflect that the multiple fix-it
23740 hints are now consolidated into one insertion.
23741 (selftest::test_one_liner_many_fixits_2): New test.
23742 (selftest::test_diagnostic_show_locus_one_liner): Update for
23744 (selftest::test_fixit_consolidation): Update for fix-it API
23746 * diagnostic.c (print_parseable_fixits): Likewise.
23747 * edit-context.c (edited_line::m_line_events): Convert from
23748 auto_vec <line_event *> to auto_vec <line_event>.
23749 (class line_event): Convert from abstract base class to a concrete
23750 class, taking over the role of replace_event.
23751 (class insert_event): Delete.
23752 (class replace_event): Rename to class line_event. Convert to
23754 (edit_context::add_fixits): Reimplement.
23755 (edit_context::apply_insert): Delete.
23756 (edit_context::apply_replace): Rename to...
23757 (edit_context::apply_fixit): ...this. Convert to half-open range.
23758 (edited_file::apply_insert): Delete.
23759 (edited_file::apply_replace): Rename to...
23760 (edited_file::apply_fixit): ...this.
23761 (edited_line::~edited_line): Drop deletion of events.
23762 (edited_line::apply_insert): Delete.
23763 (edited_line::apply_replace): Rename to...
23764 (edited_line::apply_fixit): ...this. Convert to half-open range.
23765 Update for change to type of m_line_events.
23766 * edit-context.h (edit_context::apply_insert): Delete.
23767 (edit_context::apply_replace): Rename to...
23768 (edit_context::apply_fixit): ...this.
23770 2017-05-01 Martin Sebor <msebor@redhat.com>
23772 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
23775 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
23778 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
23779 __get_cpuid_max returns 0.
23780 (__get_cpuid_count): Ditto.
23782 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
23784 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
23785 replacement expression is another instance of one of its arguments.
23787 2017-05-01 Jakub Jelinek <jakub@redhat.com>
23790 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
23791 check for stack push/pop autoinc.
23792 * config/i386/i386.c (ix86_agi_dependent): Return false
23793 if the only reason why modified_in_p returned true is that
23794 addr is SP based and set_insn is a push or pop.
23796 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
23798 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
23801 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
23804 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
23805 (account_size_time): Use two predicates - exec_pred and
23807 (evaluate_conditions_for_known_args): Compute both clause and
23809 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
23810 (inline_summary_t::duplicate): Update.
23811 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
23813 (compute_inline_parameters): Likewise.
23814 (estimate_edge_size_and_time): Update caluclation of time.
23815 (estimate_node_size_and_time): Compute both time and nonspecialized
23817 (estimate_ipcp_clone_size_and_time): Update.
23818 (inline_merge_summary): Update.
23819 (do_estimate_edge_time): Update.
23820 (do_estimate_edge_size): Update.
23821 (do_estimate_edge_hints): Update.
23822 (inline_read_section, inline_write_summary): Stream both new predicates.
23823 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
23825 (compute_inlined_call_time): Cleanup.
23826 (big_speedup_p): Update.
23827 (edge_badness): Update.
23828 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
23829 (size_time_entry): Replace predicate by exec_predicate and
23830 nonconst_predicate.
23831 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
23832 (estimate_edge_time): Return also nonspec_time.
23833 (reset_edge_growth_cache): Update.
23835 2017-04-29 Jakub Jelinek <jakub@redhat.com>
23837 PR rtl-optimization/80491
23838 * ifcvt.c (noce_process_if_block): When looking for x setter
23839 with missing else_bb, don't check only the insn right before
23840 cond_earliest, but look for the last insn that x is modified in
23841 within the same bb.
23843 PR rtl-optimization/80491
23844 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
23846 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
23848 PR tree-optimization/80487
23849 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
23851 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23853 PR tree-optimization/79697
23854 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
23855 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
23856 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
23858 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
23859 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
23861 2017-04-28 Martin Sebor <msebor@redhat.com>
23863 PR tree-optimization/80523
23864 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
23865 (init_target_to_host_charmap, target_to_host, target_strtol10): New
23867 (maybe_warn, format_directive, parse_directive): Use new functions.
23868 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
23870 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
23872 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
23874 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
23876 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
23877 target_header_dir): Set correctly.
23878 * configure: Regenerated.
23879 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23880 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23881 instead of SYSTEM_HEADER_DIR.
23883 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
23885 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
23886 (estimate_local_effects): Likewise.
23887 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
23888 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
23889 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
23890 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
23891 do_estimate_edge_time, estimate_edge_time): Likewise.
23892 * ipa-inline-analysis.c (estimate_node_size_and_time,
23893 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
23894 (estimate_time_after_inlining): Remove.
23896 2017-04-28 Martin Liska <mliska@suse.cz>
23898 * doc/gcov.texi: Enhance documentation of gcov.
23900 2017-04-28 Martin Liska <mliska@suse.cz>
23902 * doc/gcov.texi: Sort options in alphabetic order.
23903 * doc/gcov-dump.texi: Likewise.
23904 * doc/gcov-tool.texi: Likewise.
23905 * gcov.c (print_usage): Likewise.
23906 * gcov-dump.c (print_usage): Likewise.
23907 * gcov-tool.c (print_merge_usage_message): Likewise.
23908 (print_rewrite_usage_message): Likewise.
23909 (print_overlap_usage_message): Likewise.
23911 2017-04-28 Martin Liska <mliska@suse.cz>
23913 PR gcov-profile/53915
23914 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
23916 2017-04-28 Martin Liska <mliska@suse.cz>
23918 PR gcov-profile/79891
23919 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
23920 is marked by compiler as living on a line.
23921 (get_cycles_count): Remove usage of the union.
23922 (output_intermediate_file): Likewise.
23923 (find_source): Fix GNU coding style.
23924 (accumulate_line_counts): Remove old non-all block mode.
23925 (output_lines): Remove usage of the union.
23926 * profile.c (output_location): Include all BBs, even if
23927 belonging to a same line (and file) as a previous BB.
23929 2017-04-28 Martin Liska <mliska@suse.cz>
23931 * gcov.c (process_args): Handle new argument 'w'.
23932 (read_graph_file): Assign ID to BBs.
23933 (output_branch_count): Display BB # if verbose flag is set.
23934 (output_lines): Likewise for arcs.
23935 (print_usage): Add '--verbose' option help.
23936 * doc/gcov.texi: Document --verbose (-w) option.
23938 2017-04-28 Martin Liska <mliska@suse.cz>
23940 * gcov.c (struct block_location_info): New struct.
23941 (process_file): Fill up the new structure.
23942 (read_graph_file): Replace usage of encoding by the newly added
23944 (add_line_counts): Likewise.
23945 (accumulate_line_counts): Remove usage of the union.
23946 (function_info::function_info): New function.
23947 (function_info::~function_info): Likewise.
23948 (process_file): Call delete instead of release_function.
23949 (release_function): Release the function.
23950 (release_structures): Call delete instead of release_function.
23951 (solve_flow_graph): Replace usage of num_blocks.
23952 (find_exception_blocks): Likewise.
23953 (output_lines): Fix GNU coding style.
23955 2017-04-28 Martin Liska <mliska@suse.cz>
23958 * coverage.c (coverage_remove_note_file): New function.
23959 * coverage.h: Declare the function.
23960 * toplev.c (finalize): Clean if an error has been seen.
23962 2017-04-28 Martin Liska <mliska@suse.cz>
23964 PR gcov-profile/80031
23965 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
23966 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
23967 * gcov.c (read_graph_file): Read just number of blocks.
23968 * profile.c (branch_prob): Do not stream 0 flags per a basic
23971 2017-04-28 Martin Liska <mliska@suse.cz>
23973 * gcov-dump.c (tag_*): Add new argument to declarations.
23974 (dump_gcov_file): Likewise.
23975 (tag_blocks): Add and use new argument depth.
23976 (tag_arcs): Likewise.
23977 (tag_lines): Likewise.
23978 (tag_counters): Likewise.
23979 (tag_summary): Likewise.
23980 (dump_working_sets): Use depth to do a proper indentation.
23982 2017-04-28 Jakub Jelinek <jakub@redhat.com>
23985 * cgraph.h (symtab_node::debug_symtab): No longer inline.
23986 * symtab.c (symtab_node::debug_symtab): Move definition here.
23988 2017-04-28 Richard Biener <rguenther@suse.de>
23990 * lto-streamer.h (LTO_major_version): Bump to 7.
23992 2017-04-28 Richard Biener <rguenther@suse.de>
23994 * tree-vrp.c (assert_info): New struct.
23995 (add_assert_info): New helper.
23996 (register_edge_assert_for_2): Refactor to add asserts to a vector
23998 (register_edge_assert_for_1): Likewise.
23999 (register_edge_assert_for): Likewise.
24000 (finish_register_edge_assert_for): New helper actually registering
24001 asserts where live on edge.
24002 (find_conditional_asserts): Adjust.
24003 (find_switch_asserts): Likewise.
24004 (evrp_dom_walker::try_find_new_range): Generalize.
24005 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
24007 2017-04-27 Marek Polacek <polacek@redhat.com>
24010 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
24011 arg10 and arg11 to itype.
24013 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
24015 * doc/extend.texi (Object Size Checking): Improve grammar.
24017 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
24020 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
24021 that the logic for permitting reciprocal estimates matches that
24024 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24027 * tree.c (type_cache_hasher::equal): Only compare
24028 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
24029 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
24030 non-aggregate element types.
24031 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
24032 about the flag on ARRAY_TYPEs in the comment, formatting fix.
24034 2017-04-27 Richard Biener <rguenther@suse.de>
24036 PR middle-end/80533
24037 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
24038 stripping ARRAY_REFs from MEM_EXPR make sure we're not
24039 keeping a reference to a trailing array.
24041 2017-04-27 Richard Biener <rguenther@suse.de>
24043 PR middle-end/80539
24044 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
24045 being in loop-closed SSA form conservatively.
24046 (chrec_fold_multiply_poly_poly): Likewise.
24048 2017-04-27 Tamar Christina <tamar.christina@arm.com>
24050 PR middle-end/79665
24051 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
24052 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
24054 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24057 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
24058 (aarch64_function_arg_alignment): Return unsigned int again, but still
24059 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
24060 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24061 Don't emit -Wpsabi note.
24062 (aarch64_function_arg_boundary): Likewise.
24063 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24066 2017-04-26 Nathan Sidwell <nathan@acm.org>
24068 * tree.h (crc32_unsigned_n): Declare.
24069 (crc32_unsigned, crc32_unsigned): Make inline.
24070 * tree.c (crc32_unsigned_bits): Replace with ...
24071 (crc32_unsigned_n): ... this.
24072 (crc32_unsigned, crc32_byte): Remove.
24073 (crc32_string): Remove unnecessary braces.
24075 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24077 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
24078 * ipa-inline-analysis.c (MAX_TIME): Remove.
24079 (account_size_time): Use sreal for time.
24080 (dump_inline_summary): Update.
24081 (estimate_function_body_sizes): Update.
24082 (estimate_edge_size_and_time): Update.
24083 (estimate_calls_size_and_time): Update.
24084 (estimate_node_size_and_time): Update.
24085 (inline_merge_summary): Update.
24086 (inline_update_overall_summary): Update.
24087 (estimate_time_after_inlining): Update.
24088 (inline_read_section): Update.
24089 (inline_write_summary): Update.
24090 * ipa-inline.c (compute_uninlined_call_time): Update.
24091 (compute_inlined_call_time): Update.
24092 (recursive_inlining): Update.
24093 (inline_small_functions): Update.
24094 (dump_overall_stats): Update.
24095 * ipa-inline.h: Include sreal.h.
24096 (size_time_entry): Turn time to sreal.
24097 (inline_summary): Turn self_time nad time to sreal.
24099 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24101 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
24103 (sreal::stream_out, sreal::stream_in): New.
24104 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
24106 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24108 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
24111 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
24114 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
24116 (dimode_scalar_chain::compute_convert_gain): Ditto.
24117 (dimode_scalar_chain::make_vector_copies): Ditto.
24118 (dimode_scalar_chain::convert_reg): Ditto.
24119 (dimode_scalar_chain::convert_insn): Ditto.
24120 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
24121 (VI248_AVX512BW_1): New mode iterator.
24122 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
24123 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
24126 2017-04-25 Martin Sebor <msebor@redhat.com>
24128 PR tree-optimization/80497
24129 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
24130 constants are representable in HOST_WIDE_INT.
24131 (parse_directive): Ditto.
24133 2017-04-25 Martin Sebor <msebor@redhat.com>
24136 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
24137 (new_zero_array): Adjust signature.
24138 (dom_info::dom_init): Used unsigned rather that size_t.
24139 (dom_info::dom_info): Same.
24141 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24142 Jakub Jelinek <jakub@redhat.com>
24145 * config/arm/arm.c: Include gimple.h.
24146 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24147 returns negative, increment ncrn only if it returned positive.
24148 (arm_needs_doubleword_align): Return int instead of bool,
24149 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
24150 members, but if there is any such non-FIELD_DECL
24151 > PARM_BOUNDARY aligned decl, return -1 instead of false.
24152 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24153 returns negative, increment nregs only if it returned positive.
24154 (arm_setup_incoming_varargs): Likewise.
24155 (arm_function_arg_boundary): Emit -Wpsabi note if
24156 arm_needs_doubleword_align returns negative, return
24157 DOUBLEWORD_ALIGNMENT only if it returned positive.
24159 2017-04-25 Marek Polacek <polacek@redhat.com>
24162 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
24163 first argument to type.
24165 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
24168 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
24169 type checks to test for compatibility instead of equality.
24171 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24172 Jakub Jelinek <jakub@redhat.com>
24175 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
24177 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
24178 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
24179 the alignment computation, but return their maximum in warn_alignment.
24180 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24181 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
24183 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
24184 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24187 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24189 * config/arc/simdext.md (dmpyh): Fix typo.
24191 2017-04-25 Richard Biener <rguenther@suse.de>
24193 PR tree-optimization/80492
24194 * alias.c (compare_base_decls): Handle registers with asm
24195 specification conservatively.
24196 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24197 compare_base_decls returning dont-know properly.
24199 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24201 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
24202 (legitimate_offset_address_p): New function.
24203 (arc_legitimate_address_p): Use above function.
24205 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24207 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
24209 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24211 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
24212 ACCH registers whenever they are available.
24214 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24216 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
24217 double regs fix when not used.
24219 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24221 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
24223 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
24224 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
24226 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24228 * config/arc/arc.c (arc_output_addsi): Check for h-register class
24229 when emitting short ADD instructions.
24231 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24233 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
24235 (cmpsi_cc_c_insn): Likewise.
24236 (cbranchsi4_scratch): Compute proper instruction length using
24237 compact_hreg_operand.
24238 * config/arc/predicates.md (compact_hreg_operand): New predicate.
24240 2017-04-25 Richard Biener <rguenther@suse.de>
24242 PR middle-end/80509
24243 * passes.c (pass_manager::pass_manager): Initialize
24244 m_name_to_pass_map.
24246 2017-04-25 Richard Biener <rguenther@suse.de>
24248 PR tree-optimization/79201
24249 * tree-ssa-sink.c (statement_sink_location): Handle calls.
24251 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24254 * config/s390/vector.md: Split MEM->GPR vector moves for
24255 non-s_operand addresses.
24257 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24260 * config/s390/predicates.md (reload_const_wide_int_operand): New
24262 * config/s390/s390.md ("movti"): Remove d/P alternative.
24263 ("movti_bigconst"): New pattern definition.
24265 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24268 * s390-protos.h (s390_expand_cs_hqi): Removed.
24269 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
24270 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
24271 modes as well as CCZ1mode and CCZmode.
24272 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
24273 signature of s390_emit_compare_and_swap.
24274 (s390_expand_cs_hqi): Likewise, make static.
24275 (s390_expand_cs_tdsi): Generate an explicit compare before trying
24276 compare-and-swap, in some cases.
24277 (s390_expand_cs): Wrapper function.
24278 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
24280 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
24281 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
24282 patterns for small and large integers. Forbid symref memory operands.
24283 Move expander to s390.c. Require cc register.
24284 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
24285 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
24286 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
24287 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
24288 symref memory operands. Remove CC mode and call s390_match_ccmode
24290 ("atomic_exchange<mode>"): Allow and implement all integer modes.
24292 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24294 * config/s390/s390.md (define_peephole2): New peephole to help
24295 combining the load-and-test pattern with volatile memory.
24297 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24299 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
24300 with CCZmode for TARGET_Z196.
24302 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24304 PR rtl-optimization/80501
24305 * combine.c (make_compound_operation_int): Set subreg_code to SET
24306 even for AND with mask of the sign bit of mode.
24308 PR rtl-optimization/80500
24309 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
24310 sum's initial value.
24312 2017-04-25 Julian Brown <julian@codesourcery.com>
24313 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24315 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
24317 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
24319 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
24321 2017-04-25 Julian Brown <julian@codesourcery.com>
24322 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24324 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
24325 (thunderx2t99_sha): New Reservation.
24327 2017-04-25 Julian Brown <julian@codesourcery.com>
24328 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24330 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
24331 type for 1-element load.
24333 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
24335 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
24337 2017-04-24 Martin Jambor <mjambor@suse.cz>
24339 PR tree-optimization/80293
24340 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
24341 char arrays not totally scalarizable if it is false.
24342 (analyze_all_variable_accesses): Pass correct value in the new
24343 parameter. Add a statistics counter.
24345 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
24347 PR middle-end/79931
24348 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
24350 2017-04-24 Richard Biener <rguenther@suse.de>
24352 PR tree-optimization/80494
24353 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
24354 out for complex types.
24356 2017-04-24 Richard Biener <rguenther@suse.de>
24358 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
24359 * tree-ssa-sccvn.c (print_scc): Print SCC size.
24360 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
24361 (DFS): Adjust and never fail.
24362 (sccvn_dom_walker::fail): Remove.
24363 (sccvn_dom_walker::before_dom_children): Adjust.
24364 (run_scc_vn): Likewise and never fail.
24365 * tree-ssa-pre.c (pass_pre::execute): Adjust.
24366 (pass_fre::execute): Likewise.
24368 2017-04-24 Richard Biener <rguenther@suse.de>
24370 PR tree-optimization/79725
24371 * tree-ssa-sink.c (statement_sink_location): Return whether
24372 failure reason was zero uses. Move that check later.
24373 (sink_code_in_bb): Deal with zero uses by removing the stmt
24376 2017-04-24 Richard Biener <rguenther@suse.de>
24379 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
24380 pointer-based references.
24382 2017-04-24 Richard Biener <rguenther@suse.de>
24385 * pass_manager.h (pass_manager::operator new): Remove.
24386 (pass_manager::operator delete): Likewise.
24387 * passes.c (pass_manager::operator new): Remove.
24388 (pass_manager::operator delete): Likewise.
24389 (pass_manager::pass_manager): Zero individual pass members.
24391 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
24394 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
24395 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
24396 Check "XEXP (src, 1)" operand here.
24397 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
24398 Check "XEXP (src, 1)" operand here.
24399 (dimode_scalar_chain::make_vector_copies): Detect count register
24400 of a shift instruction. Zero extend count register from QImode
24401 to DImode to satisfy vector shift pattern count operand predicate.
24402 Substitute vector shift count operand with a DImode copy.
24403 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
24406 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
24408 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
24409 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
24410 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
24411 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
24412 (UNSPEC_NOREX_MEM): Remove definition.
24414 2017-04-21 Richard Biener <rguenther@suse.de>
24416 PR tree-optimization/79547
24417 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
24418 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
24419 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
24420 without any constraints.
24422 2017-04-21 Richard Biener <rguenther@suse.de>
24424 PR tree-optimization/78847
24425 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
24427 2017-04-21 Richard Biener <rguenther@suse.de>
24429 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
24430 (build_distinct_type_copy): Likewise.
24431 (build_variant_type_copy): Likewise.
24432 * tree.c (build_qualified_type): Pass down mem-stat info.
24433 (build_distinct_type_copy): Likewise.
24434 (build_variant_type_copy): Likewise.
24436 2017-04-21 Richard Biener <rguenther@suse.de>
24438 PR tree-optimization/80237
24439 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
24441 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
24442 for a simplified result.
24444 2016-04-21 Richard Biener <rguenther@suse.de>
24446 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
24447 sth as strict as a simple_iv but a chrec without symbols and an
24448 operand defined in the loop we are peeling (and not some subloop).
24449 (propagate_constants_for_unrolling): Propagate all constants.
24451 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
24454 * config/i386/i386.c (print_reg): Remove assert for disalowed
24455 regno values, call output_operand_lossage instead.
24457 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
24460 * config/i386/constraints.md (Yc): New register constraint.
24461 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
24462 Use Yc constraint for alternative 2 of operand 0. Remove
24463 preferred_for_speed attribute.
24465 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
24467 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
24468 lastprivate clauses in SIMT case.
24470 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
24472 * doc/invoke.texi (-Wextra-semi): Document new warning option.
24474 2017-04-20 Richard Biener <rguenther@suse.de>
24476 PR tree-optimization/57796
24477 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
24478 as N scalar stores.
24479 (vect_model_load_cost): Cost gathers as N scalar loads.
24481 2017-04-20 Richard Biener <rguenther@suse.de>
24483 * ggc-page.c (ggc_allocated_p): Rename to ...
24484 (safe_lookup_page_table_entry): ... this and return the lookup
24486 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
24488 2017-04-20 Richard Biener <rguenther@suse.de>
24490 PR tree-optimization/80453
24491 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
24492 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
24493 from the conditions.
24494 (vn_phi_eq): Pass them down.
24495 (vn_phi_lookup): Record them.
24496 (vn_phi_insert): Likewise.
24498 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
24500 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
24501 uninitialized variable warning to avoid buffer overrun.
24503 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
24506 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
24507 is suppressed for '{ 0 }' in C.
24509 2017-04-20 Jakub Jelinek <jakub@redhat.com>
24511 * BASE-VER: Set to 8.0.0.
24513 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
24515 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
24516 priority .init_array and .fini_array section with SECTION_NOTYPE
24519 2017-04-20 Jakub Jelinek <jakub@redhat.com>
24521 PR middle-end/80423
24522 * tree.h (build_array_type): Add typeless_storage default argument.
24523 * tree.c (type_cache_hasher::equal): Also compare
24524 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
24525 (build_array_type): Add typeless_storage argument, set
24526 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
24528 (build_nonshared_array_type): Adjust build_array_type_1 caller.
24529 (build_array_type): Likewise. Add typeless_storage argument.
24531 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24532 Jakub Jelinek <jakub@redhat.com>
24534 PR tree-optimization/80426
24535 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
24536 operation on symbolic operands, also compute the overflow for the
24537 invariant part when the operation degenerates into a negation.
24539 2017-04-19 Jakub Jelinek <jakub@redhat.com>
24542 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
24543 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
24546 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
24548 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
24551 * config/avr/avr.c (tree.h): Include it.
24552 (cgraph.h): Include it.
24553 (avr_encode_section_info): Don't warn for uninitialized progmem
24554 variable if it's just an alias.
24556 2017-04-19 Richard Biener <rguenther@suse.de>
24559 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
24560 when needed by AutoPGO.
24562 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
24565 * doc/lto.texi: Remove an extra 'that'.
24567 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
24569 PR rtl-optimization/80429
24570 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
24571 are only used in debug insns.
24573 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24574 Vladimir Makarov <vmakarov@redhat.com>
24576 * config/sparc/predicates.md (input_operand): Add comment. Return
24577 true for any memory operand when LRA is in progress.
24578 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
24580 2017-04-18 Jeff Law <law@redhat.com>
24583 * mips.md ({return,simple_return}_internal): Do not overwrite
24586 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24588 PR tree-optimization/80443
24589 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
24590 instead of adding 1, subtract -1 and similarly instead of subtracting
24593 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
24595 PR rtl-optimization/80357
24596 * haifa-sched.c (tmp_bitmap): New variable.
24597 (model_recompute): Handle duplicate use records.
24598 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
24599 (free_global_sched_pressure_data): Free it.
24601 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
24604 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24605 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24606 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24607 instead of SYSTEM_HEADER_DIR.
24609 2017-04-18 Jeff Law <law@redhat.com>
24611 PR middle-end/80422
24612 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
24613 predecessors after walking up the insn chain.
24615 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24618 * dwarf2out.c (modified_type_die): Try harder not to emit internal
24619 sizetype type into debug info.
24621 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24624 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
24625 unneeded test for TARGET_UPPER_REGS_SF.
24626 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
24628 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24631 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
24632 instead of gsi_after_labels.
24634 2017-04-18 Jeff Law <law@redhat.com>
24636 * regcprop.c (maybe_mode_change): Avoid creating copies of the
24640 2017-04-13 Jeff Law <law@redhat.com>
24641 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
24642 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
24644 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
24647 * config/avr/avr.c (intl.h): Include it.
24648 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
24650 2017-04-18 Martin Liska <mliska@suse.cz>
24652 PR gcov-profile/78783
24653 * gcov-tool.c (gcov_output_files): Validate that destination
24654 file is either removed by the tool or by a user.
24656 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24657 Guy Benyei <guybe@mellanox.com>
24659 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
24660 block, and do not negate it, the stored id is already negative.
24662 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24664 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
24666 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
24669 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
24670 masks of options that should be turned off if the VSX vector
24671 options are turned off.
24672 (OTHER_P8_VECTOR_MASKS): Likewise.
24673 (OTHER_VSX_VECTOR_MASKS): Likewise.
24674 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
24675 rs6000_disable_incompatible_switches to validate no type switches
24677 (rs6000_incompatible_switch): New function to disallow turning on
24678 other vector options if -mno-vsx, -mno-power8-vector, or
24679 -mno-power9-vector are specified.
24681 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24683 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
24685 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24687 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
24688 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
24689 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
24690 (ARG_POINTER_CFA_OFFSET): Likewise.
24692 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24694 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
24695 conditions to take advantage of various optimizations.
24697 2017-04-13 Jeff Law <law@redhat.com>
24699 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
24700 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
24701 (zero_extendsidi2_dext): Likewise.
24703 2017-04-13 Jakub Jelinek <jakub@redhat.com>
24706 * fold-const.c (fold_ternary_loc): Revert
24707 use op0 instead of fold_convert_loc (loc, type, arg0) part of
24710 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
24712 PR rtl-optimization/80343
24713 * lra-remat.c (update_scratch_ops): Assign original hard reg to
24714 new scratch pseudo.
24716 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
24719 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
24720 to ubsan_encode_value.
24722 2017-04-13 Jeff Law <law@redhat.com>
24724 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
24725 appearing in DEBUG_INSNs.
24727 2017-04-13 Martin Liska <mliska@suse.cz>
24729 PR gcov-profile/80413
24730 * gcov-io.c (gcov_write_string): Copy to buffer just when
24731 allocated size is greater than zero.
24733 2017-04-13 Jakub Jelinek <jakub@redhat.com>
24736 * dwarf2out.c (decls_for_scope): Ignore declarations of
24737 current_function_decl in BLOCK_NONLOCALIZED_VARS.
24739 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
24742 * ipa-visibility.c (non_local_p): Fix typos.
24743 (localize_node): When localizing symbol in same comdat group,
24744 dissolve the group only when we know external symbols are going
24746 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
24748 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24750 PR tree-optimization/79390
24751 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
24752 order does not result in usable sequence, retry with reversed operand
24758 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
24759 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
24760 op0 instead of fold_convert_loc (loc, type, arg0).
24762 2017-04-12 Jeff Law <law@redhat.com>
24764 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
24765 has a delay slot in the generated code.
24767 * config/cris/cris.md (cris_preferred_reload_class): Return
24768 GENNONACR_REGS rather than GENERAL_REGS.
24770 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24773 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
24774 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
24775 signedness of the result type.
24777 2017-04-12 Richard Biener <rguenther@suse.de>
24778 Jeff Law <law@redhat.com>
24780 PR tree-optimization/80359
24781 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
24782 trim stores to TARGET_MEM_REFs.
24784 2017-04-12 Richard Biener <rguenther@suse.de>
24786 PR tree-optimization/79390
24787 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
24788 threading case even more.
24790 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
24793 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
24794 for quad_address_p for TImode, instead of just not indexed_address.
24796 2017-04-12 Richard Biener <rguenther@suse.de>
24797 Bernd Edlinger <bernd.edlinger@hotmail.de>
24799 PR middle-end/79671
24800 * alias.c (component_uses_parent_alias_set_from): Handle
24801 TYPE_TYPELESS_STORAGE.
24802 (get_alias_set): Likewise.
24803 * tree-core.h (tree_type_common): Add typeless_storage flag.
24804 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
24805 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
24806 for types containing members with TYPE_TYPELESS_STORAGE.
24807 (place_field): Likewise.
24808 (layout_type): Likewise for ARRAY_TYPE.
24809 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
24810 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
24811 TYPE_TYPELESS_STORAGE.
24812 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
24814 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24817 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
24818 first argument to type.
24820 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24824 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
24825 CONST0_RTX (mode) rather than const0_rtx where appropriate.
24826 (rs6000_expand_binop_builtin): Likewise.
24827 (rs6000_expand_ternop_builtin): Likewise; also add missing
24828 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
24829 vshasigma built-ins.
24830 * doc/extend.texi: Document that vec_xxpermdi's third argument
24831 must be a constant.
24833 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
24835 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
24836 Use shift_const cost parameter when calculating gain of STV shifts.
24838 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
24840 PR rtl-optimization/70478
24841 * lra-constraints.c (process_alt_operands): Check memory for
24842 disfavoring memory insn operand.
24844 2017-04-11 Jakub Jelinek <jakub@redhat.com>
24846 PR middle-end/80100
24847 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
24848 left shift in unsigned HOST_WIDE_INT type.
24850 PR rtl-optimization/80385
24851 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
24852 (not (neg X)) into (plus X -1) for complex or non-integral modes.
24855 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
24856 if they have any depend clauses.
24858 2017-04-11 Martin Liska <mliska@suse.cz>
24861 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
24862 * ipa-split.c (split_function): Create a local comdat symbol
24863 if caller is in a comdat group.
24865 2017-04-11 Martin Liska <mliska@suse.cz>
24868 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
24871 2017-04-11 Martin Sebor <msebor@redhat.com>
24873 PR middle-end/80364
24874 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
24875 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
24877 (directive::set_width, directive::set_precision, format_character):
24879 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
24882 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
24885 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
24886 conflict, set target->arch_name instead of target->cpu_name.
24888 2017-04-11 Richard Biener <rguenther@suse.de>
24890 PR tree-optimization/80374
24891 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
24892 build_zero_cst, remove fold_convertible_p check again.
24894 2017-04-11 Martin Liska <mliska@suse.cz>
24897 * ubsan.c (instrument_object_size): Do not instrument register
24900 2017-04-11 Jakub Jelinek <jakub@redhat.com>
24903 * config/i386/i386-builtin-types.def
24904 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
24905 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
24906 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
24907 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
24908 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
24909 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
24910 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
24911 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
24912 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
24913 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
24914 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
24915 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
24916 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
24917 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
24918 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
24919 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
24920 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
24921 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
24922 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
24923 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
24924 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
24925 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
24926 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
24927 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
24928 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
24929 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
24930 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
24931 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
24932 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
24933 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
24934 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
24935 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
24936 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
24937 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
24938 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
24939 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
24940 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
24941 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
24942 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
24943 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
24944 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
24945 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
24946 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
24947 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
24948 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
24949 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
24951 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
24952 flag to second_arg_count, handle 4 argument function type _COUNT
24953 aliases, handle second_arg_count on second argument rather than last.
24955 2017-04-10 Jeff Law <law@redhat.com>
24957 PR tree-optimization/80374
24958 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
24959 record anything if we can not convert integer_zero_node to the
24962 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24965 * config/rs6000/rs6000.c (rs6000_option_override_internal):
24966 Enhance special handling given to the TARGET_P9_MINMAX option in
24967 relation to certain other options.
24969 2017-04-10 Bin Cheng <bin.cheng@arm.com>
24971 PR tree-optimization/80153
24972 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
24973 remove POINTER_PLUS_EXPR's base part directly, rather than through
24976 2017-04-10 Richard Biener <rguenther@suse.de>
24977 Bin Cheng <bin.cheng@arm.com>
24979 PR tree-optimization/80153
24980 * tree-affine.c (aff_combination_to_tree): Get base pointer from
24981 the first element of pointer type aff_tree. Build result expr in
24983 (add_elt_to_tree): Convert to type unconditionally. Remove other
24984 fold_convert calls.
24985 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
24986 (rewrite_use_nonlinear_expr): Check invariant using iv information.
24988 2017-04-10 Richard Biener <rguenther@suse.de>
24990 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24993 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
24995 PR rtl-optimization/70478
24996 * lra-constraints.c (curr_small_class_check): New.
24997 (update_and_check_small_class_inputs): New.
24998 (process_alt_operands): Update curr_small_class_check. Disfavor
24999 alternative insn memory operands. Check available regs for small
25002 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
25005 * config/mips/mips.opt (-mvirt): Update description.
25006 * doc/invoke.texi (-mvirt): Likewise.
25008 2017-04-10 Richard Biener <rguenther@suse.de>
25010 PR middle-end/80362
25011 * fold-const.c (fold_binary_loc): Look at unstripped ops when
25012 looking for NEGATE_EXPR in -A / -B to A / B folding.
25014 2017-04-10 Martin Liska <mliska@suse.cz>
25016 PR gcov-profile/80224
25017 * gcov.c (print_usage): Fix usage string.
25018 (get_gcov_intermediate_filename): Remove.
25019 (output_gcov_file): Use both for normal and intermediate format.
25020 (generate_results): Do not initialize special file for
25021 intermediate format.
25023 2017-04-10 Richard Biener <rguenther@suse.de>
25025 PR tree-optimization/80304
25026 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
25029 2017-04-10 Nathan Sidwell <nathan@acm.org>
25032 * config/rs6000/rs6000.c (rs6000_vector_type): New.
25033 (rs6000_init_builtins): Use it.
25035 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25037 * config/arm/arm.md (<mrc>): Add mode to SET source.
25038 (<mrrc>): Likewise.
25040 2017-04-10 Richard Biener <rguenther@suse.de>
25042 PR middle-end/80344
25043 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
25045 2017-04-10 Jakub Jelinek <jakub@redhat.com>
25048 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
25049 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
25050 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
25051 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
25052 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
25053 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
25054 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
25055 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
25056 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
25057 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
25058 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
25059 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
25060 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
25061 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
25062 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
25063 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
25064 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
25065 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
25066 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
25067 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
25068 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
25069 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
25070 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
25072 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
25074 PR rtl-optimization/70478
25075 * lra-constraints.c: Reverse the last patch.
25077 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
25079 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
25080 Add comment for WCHAR_T.
25082 2017-04-08 Martin Liska <mliska@suse.cz>
25085 2017-04-07 Martin Liska <mliska@suse.cz>
25088 * ipa-split.c (split_function): Add function part to a same comdat
25091 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25094 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
25096 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
25098 * rs6000/rs6000.c (vec_load_pendulum): Rename...
25099 (vec_pairing): ...to this.
25100 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
25101 (rs6000_sched_init): Adjust for name change.
25102 (struct rs6000_sched_context): Likewise.
25103 (rs6000_init_sched_context): Likewise.
25104 (rs6000_set_sched_context): Likewise.
25106 2017-04-07 Jakub Jelinek <jakub@redhat.com>
25112 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
25114 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
25115 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
25116 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
25118 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
25120 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
25122 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25124 PR rtl-optimization/70703
25125 * ira-color.c (update_conflict_hard_regno_costs): Use
25126 int64_t instead of HOST_WIDE_INT.
25128 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25130 PR rtl-optimization/70478
25131 * lra-constraints.c (process_alt_operands): Disfavor alternative
25132 insn memory operands.
25134 2017-04-07 Jeff Law <law@redhat.com>
25136 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
25137 CALL and NOTE_INSN_CALL_ARG_LOCATION.
25139 2017-04-07 Martin Liska <mliska@suse.cz>
25142 * config/aarch64/aarch64.c (aarch64_process_target_attr):
25143 Show error message instead of an ICE.
25145 2017-04-07 Martin Liska <mliska@suse.cz>
25148 * ipa-split.c (split_function): Add function part to a same comdat
25151 2017-04-07 Richard Biener <rguenther@suse.de>
25153 PR middle-end/80341
25154 * tree.c (get_unwidened): Also handle ! for_type case for
25156 * convert.c (do_narrow): Split out from ...
25157 (convert_to_integer_1): ... here. Do not pass final truncation
25158 type to get_unwidened for TRUNC_DIV_EXPR.
25160 2017-04-07 Richard Biener <rguenther@suse.de>
25162 * tree-affine.c (wide_int_ext_for_comb): Take type rather
25164 (aff_combination_const): Adjust.
25165 (aff_combination_scale): Likewise.
25166 (aff_combination_add_elt): Likewise.
25167 (aff_combination_add_cst): Likewise.
25168 (aff_combination_convert): Likewise.
25169 (add_elt_to_tree): Likewise. Remove unused argument.
25170 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
25172 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
25174 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
25176 * config/arm/arm.c (arm_default_short_enums): Use
25177 ARM_DEFAULT_SHORT_ENUMS.
25178 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
25180 2017-04-06 Jakub Jelinek <jakub@redhat.com>
25183 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
25184 members with redundant out-of-class redeclaration.
25186 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25189 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
25190 * config/i386/i386.md (*zero_extendsidi2):
25191 Add (?*x,*x) and (?*v,*v) alternatives.
25193 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25196 * config/i386/i386.c (ix86_expand_builtin)
25197 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
25198 mode from insn data. Convert operands to insn operand mode.
25199 Copy operands that don't satisfy insn predicate to a register.
25201 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
25203 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
25206 2017-04-06 Richard Biener <rguenther@suse.de>
25208 PR tree-optimization/80334
25209 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
25210 preserve alignment of accesses.
25212 2017-04-06 Richard Biener <rguenther@suse.de>
25214 PR tree-optimization/80262
25215 * tree-sra.c (build_ref_for_offset): Preserve address-space
25217 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
25218 Drop useless address-space information on MEM_REF offsets.
25220 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
25222 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
25224 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25226 PR rtl-optimization/70703
25227 * ira-color.c (update_conflict_hard_regno_costs): Use
25228 HOST_WIDE_INT instead of long.
25230 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
25233 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
25234 not defined for x86_64 target. Add -mmmx target option when __SSE2__
25236 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
25237 for x86_64 target. Handle -m3dnowa option.
25239 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25241 PR rtl-optimization/70703
25242 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
25243 (update_conflict_hard_regno_costs): Use long instead of unsigned
25244 arithmetic for cost calculation.
25246 2017-04-05 Jakub Jelinek <jakub@redhat.com>
25247 Bernd Edlinger <bernd.edlinger@hotmail.de>
25250 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
25253 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
25256 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
25257 ptr_mode with Pmode throughout.
25258 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
25259 into probe_stack_range and use DImode.
25261 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
25264 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
25265 call_eh_return is true.
25267 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25269 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
25270 Initialize last_match_fntype_index.
25272 2017-04-05 Jakub Jelinek <jakub@redhat.com>
25275 * tree-nvr.c: Include internal-fn.h.
25276 (pass_return_slot::execute): Ignore internal calls without
25279 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25280 Richard Biener <rguenther@suse.de>
25283 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
25284 captures used multiple times, except for the last use.
25285 * generic-match-head.c: Include gimplify.h.
25287 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25289 PR tree-optimization/79390
25290 * target.h (struct noce_if_info): Declare.
25291 * targhooks.h (default_noce_conversion_profitable_p): Declare.
25292 * target.def (noce_conversion_profitable_p): New target hook.
25293 * ifcvt.h (struct noce_if_info): New type, moved from ...
25294 * ifcvt.c (struct noce_if_info): ... here.
25295 (noce_conversion_profitable_p): Renamed to ...
25296 (default_noce_conversion_profitable_p): ... this. No longer
25298 (noce_try_store_flag_constants, noce_try_addcc,
25299 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
25300 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
25301 instead of noce_conversion_profitable_p.
25302 * config/i386/i386.c: Include ifcvt.h.
25303 (ix86_option_override_internal): Don't override
25304 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
25305 (ix86_noce_conversion_profitable_p): New function.
25306 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
25307 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
25308 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
25309 * doc/tm.texi: Regenerated.
25311 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25313 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
25316 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25319 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
25320 instructions for small multiply cores.
25322 2017-04-04 Jeff Law <law@redhat.com>
25324 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
25326 (mips_expand_vec_perm_const): Initialize elements in orig_perm
25327 that are not set by the loop over the elements.
25329 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25332 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
25333 int mode, convert_modes it to mode as unsigned, otherwise use
25334 lowpart_subreg to mode rather than SImode.
25335 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
25336 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
25337 Use DImode instead of SImode for the shift count operand.
25338 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
25341 2017-04-04 Richard Biener <rguenther@suse.de>
25343 PR middle-end/80281
25344 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
25345 arithmetic done for the negate or the plus. Simplify.
25346 (A - (-B) -> A + B): Likewise.
25347 * fold-const.c (split_tree): Make sure to not negate pointers.
25349 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
25351 PR rtl-optimization/60818
25352 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
25353 a compare of comparisons with the thing compared if this results
25354 in a different machine mode.
25356 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
25358 * alias.c (base_alias_check): Fix typo in comment.
25359 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
25360 * cgraphunit.c (symbol_table::compile): Likewise.
25361 * collect2.c (maybe_run_lto_and_relink): Likewise.
25362 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
25363 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
25364 * config/avr/avr.c (avr_map_op_t): Likewise.
25365 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
25366 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
25367 * config/epiphany/epiphany.md (movcc): Likewise.
25368 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
25369 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
25371 * config/mips/mips.c (mips_save_restore_reg): Likewise.
25372 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
25373 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
25374 * config/sh/sh.c (sh_rtx_costs): Likewise.
25375 * fold-const.c (fold_truth_andor): Likewise.
25376 * genautomata.c (collapse_flag): Likewise.
25377 * gengtype.h (struct type::u::s): Likewise.
25378 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
25379 * input.c (FORMAT_AMOUNT): Likewise.
25380 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
25381 (known_aggs_to_agg_replacement_list): Likewise.
25382 * ipa-inline-analysis.c: Likewise.
25383 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
25384 * ipa-polymorphic-call.c
25385 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
25386 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
25387 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
25389 * modulo-sched.c (apply_reg_moves): Likewise.
25390 * omp-expand.c (build_omp_regions_1): Likewise.
25391 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
25392 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
25393 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
25394 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
25395 * value-prof.c: Likewise.
25396 * var-tracking.c (val_reset): Likewise.
25398 2017-04-03 Richard Biener <rguenther@suse.de>
25400 PR tree-optimization/80275
25401 * fold-const.c (split_address_to_core_and_offset): Handle
25404 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
25406 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
25407 descriptors is at least equal to that of functions.
25409 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
25411 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
25413 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
25416 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
25417 (mov<IMOD4:mode>): New expander.
25418 (*mov<IMOD4:mode>_internal): New insn and split pattern.
25420 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
25422 PR rtl-optimization/79405
25423 * fwprop.c (propagations_left): New variable.
25424 (forward_propagate_into): Decrement it.
25425 (fwprop_init): Initialize it.
25426 (fw_prop): If the variable has reached zero, stop propagating.
25427 (fwprop_addr): Ditto.
25429 2017-03-31 Jakub Jelinek <jakub@redhat.com>
25432 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
25433 a FUNCTION_DECL, pass it as decl instead of origin to
25436 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
25438 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
25441 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
25444 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
25445 TARGET_VSX_SMALL_INTEGER.
25447 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25449 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25450 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
25452 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
25454 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
25455 extraction from odd-numbered MSA register.
25457 2017-03-31 Jakub Jelinek <jakub@redhat.com>
25459 PR middle-end/80173
25460 * expmed.c (store_bit_field_1): Don't attempt to create
25461 a word subreg out of hard registers wider than word if they
25462 have HARD_REGNO_NREGS of 1 for their mode.
25464 PR middle-end/80163
25465 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
25466 conversions to integer types wider than word and pointer.
25469 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
25470 (rtx_equal_for_cselib_p): Pass 0 to it.
25471 * cselib.c (cselib_hasher::equal): Likewise.
25472 (rtx_equal_for_cselib_1): Add depth argument. If depth
25473 is 128, don't look up VALUE locs and punt. Increment
25474 depth in recursive calls when walking VALUE locs.
25476 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
25478 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
25479 (make_gcov_file_name): Use the canonical path name for generating
25481 (read_line): Fix handling of files with ascii null bytes.
25483 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
25485 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
25486 to initialise a vector register instead
25487 of using a const_int.
25489 2017-03-30 Jakub Jelinek <jakub@redhat.com>
25491 PR translation/80189
25492 * gimplify.c (omp_default_clause): Use %qs instead of %s in
25493 diagnostic messages.
25495 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
25498 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
25499 (dfp_diex_<mode>): Update mode of operand 1.
25500 * doc/extend.texi (dxex, dxexq): Document change to return type.
25501 (diex, diexq): Document change to argument type.
25503 2017-03-30 Martin Jambor <mjambor@suse.cz>
25506 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
25507 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
25508 it reflects the signature changes performed at the callee side.
25509 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
25510 to cgraph_build_function_type_skip_args.
25511 (build_function_decl_skip_args): Adjust call to the above function.
25513 2017-03-30 Jakub Jelinek <jakub@redhat.com>
25516 * config/i386/sse.md
25517 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
25518 register as dest whenever it is a MEM not rtx_equal_p to the
25519 corresponding dup operand, and when forcing into reg move the
25520 reg into the memory afterwards.
25521 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
25522 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
25523 for the force_reg mode.
25524 (avx512vl_vextractf128<mode>): Use register as dest either
25525 always when a MEM, or when it is a MEM not rtx_equal_p to the
25526 corresponding dup operand, or even not when it is a CONST_VECTOR
25527 depending on the mode and lo vs. hi.
25528 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
25530 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
25531 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
25532 Likewise. Require that operands[2] is even.
25533 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
25534 Remove extraneous parens. Require that operands[2] is a multiple
25536 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
25537 operands[0] is a MEM if <mask_applied>, the predicates/constraints
25538 disallow memory then.
25540 2017-03-30 Richard Biener <rguenther@suse.de>
25542 PR tree-optimization/77498
25543 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
25544 to non-constants over backedges.
25546 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
25548 PR rtl-optimization/80233
25549 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
25550 as last_combined_insn. Do not test for BARRIER_P separately.
25552 2017-03-29 Andreas Schwab <schwab@suse.de>
25555 * calls.c (prepare_call_address): Convert funexp to Pmode before
25556 copying to temp reg.
25558 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25560 PR tree-optimization/80158
25561 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
25562 Handle possible future case of more than one alternate
25564 (replace_rhs_if_not_dup): Likewise.
25565 (replace_one_candidate): Likewise.
25567 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
25569 PR rtl-optimization/80193
25570 * ira.c (ira): Do not check allocation for LRA.
25572 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
25574 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
25575 (nvptx_output_simt_exit): Declare.
25576 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
25577 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
25578 (init_softstack_frame): Move initialization of crtl->is_leaf to...
25579 (nvptx_declare_function_name): ...here. Emit declaration of local
25580 memory space buffer for omp_simt_enter insn.
25581 (nvptx_output_unisimt_switch): New.
25582 (nvptx_output_softstack_switch): New.
25583 (nvptx_output_simt_enter): New.
25584 (nvptx_output_simt_exit): New.
25585 * config/nvptx/nvptx.h (struct machine_function): New fields
25586 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
25587 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
25588 (UNSPECV_SIMT_EXIT): Ditto.
25589 (omp_simt_enter_insn): New insn.
25590 (omp_simt_enter): New expansion.
25591 (omp_simt_exit): New insn.
25592 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
25594 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
25595 (expand_GOMP_SIMT_ENTER_ALLOC): New.
25596 (expand_GOMP_SIMT_EXIT): New.
25597 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
25598 (GOMP_SIMT_ENTER_ALLOC): Ditto.
25599 (GOMP_SIMT_EXIT): Ditto.
25600 * target-insns.def (omp_simt_enter): New insn.
25601 (omp_simt_exit): Ditto.
25602 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
25604 (lower_rec_simd_input_clauses): Implement SIMT privatization.
25605 (lower_rec_input_clauses): Likewise.
25606 (lower_lastprivate_clauses): Handle SIMT privatization.
25608 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
25609 (ompdevlow_adjust_simt_enter): New.
25610 (find_simtpriv_var_op): New.
25611 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
25612 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
25614 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
25615 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
25616 (copy_decl_for_dup_finish): Ditto.
25618 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
25620 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
25623 * config/i386/i386.c (ix86_option_override_internal): Always
25624 allow -mpreferred-stack-boundary=3 for 64-bit targets.
25626 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25628 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
25630 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25632 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
25633 mark new edge's irreducible flag accordign to it.
25634 (vect_do_peeling): Check loop preheader edge's irreducible flag
25635 and pass it to function slpeel_add_loop_guard.
25637 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
25639 PR tree-optimization/80218
25640 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25641 Update block frequencies and counts.
25643 2017-03-28 Richard Biener <rguenther@suse.de>
25645 PR tree-optimization/78644
25646 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
25647 of a simplification result we may not use it at all.
25649 2017-03-28 Richard Biener <rguenther@suse.de>
25652 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
25653 without arguments, generate default definition of a SSA name.
25655 2017-03-28 Richard Biener <rguenther@suse.de>
25657 PR middle-end/80222
25658 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
25659 TYPE_REF_CAN_ALIAS_ALL references.
25660 * fold-const.c (fold_indirect_ref_1): Likewise.
25662 2017-03-28 Martin Liska <mliska@suse.cz>
25665 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
25666 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
25668 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
25669 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
25671 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
25672 (EXTRA_SPECS): Define.
25673 (SUBTARGET_EXTRA_SPECS): Likewise.
25674 (SUBTARGET_CPP_SPEC): Likewise.
25675 * config/arc/elf.h (EXTRA_SPECS): Renamed to
25676 SUBTARGET_EXTRA_SPECS.
25677 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
25679 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
25681 * config/arc/simdext.md (vst64_insn): Update pattern.
25682 (vld32wh_insn): Likewise.
25683 (vld32wl_insn): Likewise.
25684 (vld64_insn): Likewise.
25685 (vld32_insn): Likewise.
25687 2017-03-28 Marek Polacek <polacek@redhat.com>
25690 * fold-const.c (fold_comparison): Use protected_set_expr_location
25691 instead of SET_EXPR_LOCATION.
25693 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
25695 * tree.c (add_expr): Avoid name lookup warning.
25697 2017-03-27 Jeff Law <law@redhat.com>
25699 PR tree-optimization/80216
25700 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
25701 function name. Limit recursion depth.
25702 (record_temporary_equivalences): Corresponding changes.
25704 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
25706 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
25709 2017-03-27 Jakub Jelinek <jakub@redhat.com>
25712 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
25714 * cfgcleanup.c (reg_note_cfa_p): New array.
25715 (insns_have_identical_cfa_notes): New function.
25716 (old_insns_match_p): Don't cross-jump in between /f
25717 and non-/f instructions. If both i1 and i2 are frame related,
25718 verify all CFA notes, their order and content.
25720 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
25723 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
25724 HImode and SImode with zero extend to DImode to one insn.
25725 (bswap<mode>2_extenddi): Likewise.
25726 (bswapsi2_extenddi): Likewise.
25727 (bswaphi2_extendsi): Likewise.
25728 (bswaphi2): Combine bswap HImode and SImode into one insn.
25729 Separate memory insns from swapping register.
25730 (bswapsi2): Likewise.
25731 (bswap<mode>2): Likewise.
25732 (bswaphi2_internal): Delete, no longer used.
25733 (bswapsi2_internal): Likewise.
25734 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
25735 store, and gpr<-gpr swap insns.
25736 (bswap<mode>2_store): Likewise.
25737 (bswaphi2_reg): Register only splitter, combine with the splitter.
25738 (bswaphi2 splitter): Likewise.
25739 (bswapsi2_reg): Likewise.
25740 (bswapsi2 splitter): Likewise.
25741 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
25742 the insns into load, store, and register/register insns.
25743 (bswapdi2_ldbrx): Likewise.
25744 (bswapdi2_load): Likewise.
25745 (bswapdi2_store): Likewise.
25746 (bswapdi2_reg): Likewise.
25748 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
25750 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
25751 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
25753 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
25756 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
25758 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25759 special handling for target option conflicts between dform
25760 options (-mpower9-dform, -mpower9-dform-vector,
25761 -mpower9-dform-scalar) and -mno-direct-move.
25763 2017-03-27 Richard Biener <rguenther@suse.de>
25765 PR tree-optimization/80181
25766 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
25768 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25770 * config/arc/predicates.md (move_double_src_operand): Replace the
25771 call to move_double_src_operand with a call to address_operand.
25773 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25775 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
25776 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
25777 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
25779 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25781 * config/arc/predicates.md (long_immediate_loadstore_operand):
25782 Consider scaled addresses cases.
25784 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25786 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
25787 restored when in interrupt.
25788 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
25789 doesn't have delay slot.
25791 2017-03-27 Richard Biener <rguenther@suse.de>
25794 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
25795 inlined thunk clones.
25797 2017-03-27 Jakub Jelinek <jakub@redhat.com>
25800 * asan.c (instrument_derefs): Copy over last operand from
25801 original COMPONENT_REF to the new COMPONENT_REF with
25802 DECL_BIT_FIELD_REPRESENTATIVE.
25803 * ubsan.c (instrument_object_size): Likewise.
25805 2017-03-27 Richard Biener <rguenther@suse.de>
25807 PR tree-optimization/80170
25808 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
25809 sure DR/SCEV didnt fold in constants we do not see when looking
25810 at the reference base alignment.
25812 2017-03-27 Richard Biener <rguenther@suse.de>
25814 PR middle-end/80171
25815 * gimple-fold.c (fold_ctor_reference): Properly guard against
25816 NULL return value from canonicalize_constructor_val.
25818 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
25821 * config/i386/i386.c (ix86_expand_builtin)
25822 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
25823 flags reg setting and flags reg using instructions.
25824 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
25825 clobbering instructions to zero extend op2.
25827 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
25829 * doc/install.texi (Configuration) <--with-aix-soname>:
25830 Update link to AIX ld.
25832 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
25834 PR rtl-optimization/80160
25835 PR rtl-optimization/80159
25836 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
25837 reg_alternate_class into account.
25839 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
25842 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
25843 to consider in curr_insn_transform.
25845 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25847 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
25848 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
25849 and emit_mode_inner.
25851 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25853 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
25854 argument to the overloaded builtin variants. Use the new flag to
25855 deprecate certain builtin variants.
25856 * config/s390/s390-builtin-types.def: Add new builtin types.
25857 * config/s390/s390-builtins.h: Support new flags field for
25858 overloaded builtins.
25859 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
25860 (s390_macro_to_expand): Enable vector float data type.
25861 (s390_cpu_cpp_builtins_internal): Indicate support of the new
25862 builtins by incrementing the __VEC__ version number.
25863 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
25865 (s390_resolve_overloaded_builtin): Emit error messages depending
25866 on the builtin flags.
25867 * config/s390/s390.c (s390_expand_builtin): Support additional
25868 flags argument. Change error message to match the messages
25869 emitted in s390-c.c.
25870 * config/s390/s390.md: New UNSPEC_* constants.
25871 (op_type): Add new instruction types.
25872 * config/s390/vecintrin.h: Add new builtins and test data class
25874 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
25875 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
25876 (VEC_INEXACT, VEC_NOINEXACT): New constants.
25877 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
25878 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
25879 ("vec_mergel<mode>"): V_HW -> VEC_HW.
25881 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
25882 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
25883 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
25884 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
25886 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
25887 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
25888 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
25889 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
25891 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
25892 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
25893 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
25894 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
25895 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
25896 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
25897 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
25899 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
25900 ("vec_scatter_element<V_HW_4:mode>_DI")
25901 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
25902 ("vec_fpint<mode>", "vflls")
25903 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
25904 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
25905 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
25906 ("*vec_cmphe<mode>_cc"): ... these.
25908 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
25909 mode constant instead of magic value.
25911 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25913 * config/s390/s390.c (s390_expand_vec_compare): Support other
25914 vector floating point modes than just V2DF.
25915 (s390_expand_vcond): Likewise.
25916 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
25917 (s390_cannot_change_mode_class): Prevent mode changes between TF
25918 and V1TF in vector registers.
25919 * config/s390/s390.md (DF, SF): New mode attributes.
25920 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
25921 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
25922 SFmode support for VRs.
25923 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
25925 (VFT, VF_HW): New mode iterators.
25926 (vw, sdx): New mode attributes.
25927 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
25928 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
25929 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
25930 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
25931 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
25932 also the new vector floating point modes. Renaming to ...
25934 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
25935 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
25936 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
25937 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
25938 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
25939 ("vec_unordered<mode>"): ... these.
25941 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
25942 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
25943 ("*vec_extendv2df"): New insn definitions.
25945 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25947 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
25948 ("mulditi3_2", "*muldi3_sign"): New patterns.
25949 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
25950 rename the pattern definition.
25952 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25954 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
25956 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
25958 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25960 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
25961 instruction if possible.
25962 * config/s390/vector.md (vec_halfnumelts): New mode
25964 ("*vec_vllezlf<mode>"): New pattern.
25966 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25968 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
25969 ("popcountv4si2", "popcountv2di2"): Rename to ...
25970 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
25971 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
25973 ("popcount<mode>2_vxe"): New pattern.
25975 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25977 * common/config/s390/s390-common.c (processor_flags_table): Add
25979 * config.gcc: Add arch12.
25980 * config/s390/driver-native.c (s390_host_detect_local_cpu):
25981 Default to arch12 for unknown CPU model numbers.
25982 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
25983 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
25984 PROCESSOR_max sanity check.
25985 * config/s390/s390-opts.h (enum processor_type): Add
25987 * config/s390/s390.c (processor_table): Add arch12.
25988 (s390_expand_builtin): Add check for B_VXE flag.
25989 (s390_issue_rate): Add PROCESSOR_ARCH12.
25990 (s390_get_sched_attrmask): Likewise.
25991 (s390_get_unit_mask): Likewise.
25992 (s390_sched_score): Enable z13 scheduling for arch12.
25993 (s390_sched_reorder): Likewise.
25994 (s390_sched_variable_issue): Likewise.
25995 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
25997 (s390_tune_attr): Use z13 scheduling also for arch12.
25998 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
25999 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
26000 (TARGET_VXE_P): New macros.
26001 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
26002 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
26003 * config/s390/s390.opt: Add arch12 as processor_type.
26005 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26007 * config/s390/s390.md
26008 ("fixuns_truncdddi2", "fixuns_trunctddi2")
26009 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
26010 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
26012 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
26013 Rename expanders to ...
26015 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
26016 ("fixuns_truncdddi2_emu"): ... these.
26018 ("fixuns_trunc<mode>si2_emu"): New expander.
26020 ("*fixuns_truncdfdi2_z13"): Rename to ...
26021 ("*fixuns_truncdfdi2_vx"): ... this.
26023 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26025 * config/s390/2964.md: Remove the single element vector compare
26026 instructions which are no longer used.
26027 * config/s390/s390.c (s390_select_ccmode): Remove handling of
26029 (s390_canonicalize_comparison): Remove handling of DFmode
26031 (s390_expand_vec_compare_scalar): Remove function.
26032 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
26033 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
26035 ("*cmp<mode>_ccs"): Add wfcdb instruction.
26037 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26039 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
26041 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
26042 will anyway by matched by mov<mode>_64dfp.
26044 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26046 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
26047 vlef/vstef. Add missing operand to vleif.
26049 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26051 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
26052 pair for all vector types with 64 bit elements.
26053 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
26054 * config/s390/vector.md (V_HW_64): ... here.
26055 (V_128_NOSINGLE): New mode iterator.
26056 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
26057 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
26058 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
26059 ("*vec_load_pairv2di"): Change to ...
26060 ("*vec_load_pair<mode>"): ... this one.
26062 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26064 * config/s390/constraints.md: Add comments.
26065 (jKK): Reject element sizes > 8 bytes.
26066 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
26068 * config/s390/s390.md: Add the s_operand checks formerly in
26069 s390_split_ok_p to various splitters where they are still
26071 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
26072 for 128 bit vectors. Plus two splitters.
26074 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26076 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
26079 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26082 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
26083 error if the boundary argument is not constant.
26085 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26087 PR rtl-optimization/80112
26088 * loop-doloop.c (doloop_condition_get): Don't check condition
26089 if cmp isn't SET with IF_THEN_ELSE src.
26091 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26093 PR tree-optimization/80158
26094 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
26095 replacing a candidate statement, also replace it for the
26096 candidate's alternate interpretation.
26097 (replace_rhs_if_not_dup): Likewise.
26098 (replace_one_candidate): Likewise.
26100 2017-03-24 Richard Biener <rguenther@suse.de>
26102 PR tree-optimization/80167
26103 * graphite-isl-ast-to-gimple.c
26104 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
26106 (translate_isl_ast_to_gimple::get_rename): Likewise.
26108 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26110 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
26111 handling of certain combinations of target options, including the
26112 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
26113 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
26115 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26118 * config/arm/arm.md (*load_multiple): Add reload_completed to
26119 matching condition.
26121 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26122 Richard Biener <rguenth@suse.de>
26124 PR tree-optimization/79908
26125 PR tree-optimization/80136
26126 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
26127 been cast away, gimplify_and_add suffices.
26129 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
26131 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
26133 2017-03-23 Richard Biener <rguenther@suse.de>
26135 PR tree-optimization/80032
26136 * gimplify.c (gimple_push_cleanup): Forced unconditional
26137 cleanups still have to go to the conditional_cleanups
26140 2017-03-22 Jakub Jelinek <jakub@redhat.com>
26142 PR tree-optimization/80072
26143 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
26145 (next_operand_entry_id): Change type to unsigned int.
26146 (sort_by_operand_rank): Make sure to return the right return value
26147 even if unsigned fields are bigger than INT_MAX.
26148 (struct oecount): Change cnt and id type to unsigned int.
26149 (oecount_hasher::equal): Formatting fix.
26150 (oecount_cmp): Make sure to return the right return value
26151 even if unsigned fields are bigger than INT_MAX.
26152 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
26155 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
26156 TREE_READONLY on result if writing it more than once.
26159 * doc/invoke.texi (-fsanitize=thread): Document that with
26160 -fnon-call-exceptions atomics are not able to throw
26164 * tsan.c: Include tree-eh.h.
26165 (instrument_builtin_call): Call maybe_clean_eh_stmt or
26166 maybe_clean_or_replace_eh_stmt where needed.
26167 (instrument_memory_accesses): Add cfg_changed argument.
26168 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
26169 if it returned true.
26170 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
26172 PR rtl-optimization/63191
26173 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
26174 wrapper function, moved the whole old content into ...
26175 (ix86_delegitimize_address_1): ... this. New inline function.
26176 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
26177 true as last argument instead of ix86_delegitimize_address.
26179 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26181 * config/aarch64/aarch64.c (generic_branch_cost): Copy
26182 cortexa57_branch_cost.
26184 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26186 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
26188 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26191 * doc/md.texi (Constraints): Document wA constraint.
26192 * config/rs6000/constraints.md (wA): New.
26193 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
26194 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
26195 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
26196 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
26198 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
26201 * gimplify.c (is_oacc_declared): New function.
26202 (oacc_default_clause): Use it to set default flags for acc declared
26203 variables inside parallel regions.
26204 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
26205 declared variables.
26206 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
26207 declare attribute to any decl as necessary.
26209 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
26212 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
26213 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
26214 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
26215 (arm_arch_lpae): This.
26216 * config/arm/arm.c (arm_arch7ve): Rename into ...
26217 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
26218 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
26221 2017-03-22 Martin Liska <mliska@suse.cz>
26224 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
26225 error message instead of an ICE.
26227 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26229 * doc/extend.texi (6.11 Additional Floating Types): Revise.
26231 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
26233 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
26235 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26238 2017-03-21 Martin Sebor <msebor@redhat.com>
26240 * doc/extend.texi: Use "cannot" instead of "can't."
26241 * doc/hostconfig.texi: Same.
26242 * doc/install.texi: Same.
26243 * doc/invoke.texi: Same.
26244 * doc/loop.texi: Same.
26245 * doc/md.texi: Same.
26246 * doc/objc.texi: Same.
26247 * doc/rtl.texi: Same.
26248 * doc/tm.texi: Same.
26249 * doc/tm.texi.in: Same.
26250 * doc/trouble.texi: Same.
26252 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
26255 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
26256 (collect_checksum_attributes): Set it.
26257 (die_checksum_ordered): Use it.
26259 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26261 PR tree-optimization/79908
26262 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
26263 change: For a VA_ARG whose LHS has been cast away, use
26264 force_gimple_operand to construct the side effects.
26266 2017-03-21 David Malcolm <dmalcolm@redhat.com>
26268 PR translation/80001
26269 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
26270 more amenable to translation.
26271 (oacc_loop_auto_partitions): Likewise.
26273 2017-03-21 Marek Polacek <polacek@redhat.com>
26274 Martin Sebor <msebor@redhat.com>
26276 PR tree-optimization/80109
26277 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
26278 on INTEGRAL_TYPE_P.
26280 2017-03-21 Jakub Jelinek <jakub@redhat.com>
26281 Segher Boessenkool <segher@kernel.crashing.org>
26284 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
26285 check reg_used_between_p between insn and one of succ or succ2
26286 depending on if succ is artificial insn not inserted into insn
26289 2017-03-21 Martin Liska <mliska@suse.cz>
26291 PR gcov-profile/80081
26292 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
26293 * doc/gcc.texi: Include gcov-dump stuff.
26294 * doc/gcov-dump.texi: New file.
26296 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
26298 PR rtl-optimization/79150
26299 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
26300 conditional jump, if the jump is the last insn of the loop.
26302 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26303 Richard Biener <rguenth@suse.de>
26305 PR tree-optimization/79908
26306 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
26307 been cast away, use force_gimple_operand to construct the side
26310 2017-03-21 Martin Liska <mliska@suse.cz>
26312 PR libfortran/79956
26313 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
26316 2017-03-21 Brad Spengler <spender@grsecurity.net>
26319 * plugin.c (htab_hash_plugin): New function.
26320 (add_new_plugin): Use it and adjust.
26321 (parse_plugin_arg_opt): Adjust.
26322 (init_one_plugin): Likewise.
26324 2017-03-21 Richard Biener <rguenther@suse.de>
26326 PR tree-optimization/80032
26327 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
26328 if set force the cleanup to happen unconditionally.
26329 (gimplify_target_expr): Push inserted clobbers with force_uncond
26330 to avoid them being removed by control-dependent DCE.
26332 2017-03-21 Richard Biener <rguenther@suse.de>
26334 PR tree-optimization/80122
26335 * tree-inline.c (copy_bb): Do not expans va-arg packs or
26336 va_arg_pack_len when the inlined call stmt requires pack
26338 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
26340 2017-03-21 Jakub Jelinek <jakub@redhat.com>
26343 * tsan.c (instrument_builtin_call): If the memory model argument
26344 is not a constant, assume it is valid.
26347 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
26350 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
26352 PR rtl-optimization/79910
26353 * combine.c (can_combine_p): Do not allow combining an I0 or I1
26354 if its dest is used by an insn before I2 (other than the combined
26355 insns themselves, which are properly handled already).
26357 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
26360 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
26362 * combine.c (record_used_regs): New static function.
26363 (try_combine): Handle situations where there is an additional
26364 instruction between I2 and I3 which needs to have a LOG_LINK
26368 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
26370 * combine.c (try_combine): Delete redundant i1 test. Call
26371 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
26373 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26376 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
26377 alternatives 13/14.
26379 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26381 PR tree-optimization/80054
26382 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
26383 the optimization if a PHI or any of its arguments is not dominated
26384 by the candidate's basis. Use gphi* rather than gimple* as
26386 (replace_profitable_candidates): Clean up a gimple* variable that
26387 should be a gphi* variable.
26389 2017-03-20 Martin Sebor <msebor@redhat.com>
26392 * doc/extend.texi (attribute constructor): Document present limitation.
26394 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
26397 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
26398 __POWER9_VECTOR__ #ifdef control, change template definition to
26399 use Power9-specific built-in function.
26400 (vec_any_eq): Likewise.
26401 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
26402 to control outcomes from this test.
26403 (vector_ae_<mode>p): For VEC_F modes, likewise.
26405 2017-03-20 Ian Lance Taylor <iant@google.com>
26407 * config/i386/i386.c (ix86_function_regparm): Save an extra
26408 register for -fsplit-stack with DECL_STATIC_CHAIN.
26410 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
26413 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
26414 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
26416 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
26418 * config/riscv/riscv.c (riscv_print_operand): Use "fence
26420 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
26423 2017-03-20 Marek Polacek <polacek@redhat.com>
26426 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
26428 2017-03-20 Richard Biener <rguenther@suse.de>
26430 PR tree-optimization/80113
26431 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
26432 allocate extra SSA name for PHI def.
26433 (add_close_phis_to_outer_loops): Likewise.
26434 (add_close_phis_to_merge_points): Likewise.
26435 (copy_loop_close_phi_args): Likewise.
26436 (copy_cond_phi_nodes): Likewise.
26438 2017-03-20 Martin Liska <mliska@suse.cz>
26440 PR middle-end/79753
26441 * tree-chkp.c (chkp_build_returned_bound): Do not build
26442 returned bounds for a LHS that's not a BOUNDED_P type.
26444 2017-03-20 Martin Liska <mliska@suse.cz>
26448 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
26449 COMPLEX_CST and VECTOR_CST.
26451 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26454 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
26455 target operand. A new splitter adds the clobber statement in case
26456 the target operand is dead anyway.
26458 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
26460 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
26461 to age-old versions of binutils and glibc.
26463 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
26465 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
26467 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
26469 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
26471 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
26473 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
26474 requirement for binutils 2.13.
26476 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
26478 * combine.c (try_combine): Delete redundant i1 test. Call
26479 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
26481 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
26483 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
26484 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
26486 <riscv64-*-elf>: Re-arrange section
26487 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
26488 <riscv32-*-linux>: Likewise.
26489 <riscv64-*-elf>: Likewise
26490 <riscv64-*-linux>: Likewise.
26492 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
26495 * aarch64.opt(verbose-cost-dump): Fix typo.
26497 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
26500 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
26501 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
26503 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
26505 * reload.c (find_reloads): When reloading a nonoffsettable address,
26506 use RELOAD_OTHER for it and its address reloads.
26508 PR rtl-optimization/79910
26509 * combine.c (record_used_regs): New static function.
26510 (try_combine): Handle situations where there is an additional
26511 instruction between I2 and I3 which needs to have a LOG_LINK
26514 2017-03-17 Jeff Law <law@redhat.com>
26516 PR tree-optimization/71437
26517 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
26518 conditional in the hash table first.
26519 (vrp_dom_walker::before_dom_children): Extract condition from
26520 ASSERT_EXPR. Record condition, its inverion and any implied
26521 conditions as well.
26523 2017-03-17 Marek Polacek <polacek@redhat.com>
26524 Markus Trippelsdorf <markus@trippelsdorf.de>
26526 PR tree-optimization/80079
26527 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
26530 2017-03-17 Richard Biener <rguenther@suse.de>
26532 PR middle-end/80075
26533 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
26534 Properly verify the LHS before the RHS possibly claims to be
26536 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
26539 2017-03-17 Martin Jambor <mjambor@suse.cz>
26541 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
26542 (List of -O2 options): Likewise.
26543 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
26546 2017-03-17 Tom de Vries <tom@codesourcery.com>
26548 * gcov-dump.c (print_usage): Print bug_report_url.
26550 2017-03-17 Richard Biener <rguenther@suse.de>
26552 PR middle-end/80050
26553 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
26554 (parser::peek): Likewise.
26556 2017-03-17 Richard Biener <rguenther@suse.de>
26558 PR tree-optimization/80048
26559 * sese.c (free_sese_info): Properly release rename_map and
26560 copied_bb_map elements.
26562 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
26564 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
26565 Add linked-list forward and backlinks. Insert on
26566 construction, remove on destruction.
26567 (class pass_store_merging): Add m_stores_head field.
26568 (pass_store_merging::terminate_and_process_all_chains):
26569 Iterate over m_stores_head list.
26570 (pass_store_merging::terminate_all_aliasing_chains):
26572 (pass_store_merging::execute): Check for debug stmts first.
26573 Push new chains onto the m_stores_head stack.
26575 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
26578 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
26579 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
26580 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
26582 2017-03-16 Jeff Law <law@redhat.com>
26584 PR tree-optimization/71437
26585 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
26586 member function. Implementation moved into after_dom_children
26587 member function and into the threader's thread_outgoing_edges
26589 (dom_opt_dom_walker::after_dom_children): Simplify by moving
26590 some code into new thread_outgoing_edges.
26591 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
26592 definition. Simplify marker handling (do it here). Assume we always
26593 have the available expression and the const/copies tables.
26594 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
26596 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
26597 * tree-vrp.c (equiv_stack): No longer file scoped.
26598 (vrp_dom_walker): New class.
26599 (vrp_dom_walker::before_dom_children): New member function.
26600 (vrp_dom_walker::after_dom_children): Likewise.
26601 (identify_jump_threads): Setup domwalker. Use it rather than
26602 walking edges in a random order by hand. Simplify setup/finalization.
26603 (finalize_jump_threads): Remove.
26604 (vrp_finalize): Do not call identify_jump_threads here.
26605 (execute_vrp): Do it here instead and call thread_through_all_blocks
26608 PR tree-optimization/71437
26609 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
26611 (simplify_stmt_for_jump_threading): Add basic_block argument. All
26613 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
26614 (dom_opt_dom_walker::thread_across_edge): Remove
26615 handle_dominating_asserts argument. All callers changed.
26616 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
26617 changes. Remove calls to lhs_of_dominating_assert. Other
26618 uses of handle_dominating_asserts turn into unconditional code
26619 (simplify_control_stmt_condition_1): Likewise.
26620 (simplify_control_stmt_condition): Likewise.
26621 (thread_through_normal_block, thread_across_edge): Likewise.
26622 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
26623 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
26624 object if it is not an SSA_NAME.
26625 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
26626 before calling into the VRP specific simplifiers.
26627 (identify_jump_threads): Remove handle_dominating_asserts
26630 2017-03-16 Jakub Jelinek <jakub@redhat.com>
26633 * tree-diagnostic.c (default_tree_printer): No longer static.
26634 * tree-diagnostic.h (default_tree_printer): New prototype.
26636 2017-03-16 Tamar Christina <tamar.christina@arm.com>
26638 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
26639 Change ins into fmov.
26641 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26643 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
26644 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
26645 Use h_con constraint for operand 1.
26646 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
26647 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
26649 2017-03-15 Jeff Law <law@redhat.com>
26651 PR tree-optimization/71437
26652 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
26653 (record_temporary_equivalences): Use it.
26655 PR tree-optimization/71437
26656 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
26657 tree-ssa-scopedtables.
26658 (lookup_avail_expr, build_and_record_new_cond): Likewise.
26659 (record_conditions, record_cond, vuse_eq): Likewise.
26660 (record_edge_info): Adjust to API tweak of record_conditions.
26661 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
26662 (record_temporary_equivalences, optimize_stmt): Likewise.
26663 (eliminate_redundant_computations): Likewise.
26664 (record_equivalences_from_stmt): Likewise.
26665 * tree-ssa-scopedtables.c: Include options.h and params.h.
26666 (vuse_eq): New function, moved from tree-ssa-dom.c
26667 (build_and_record_new_cond): Likewise.
26668 (record_conditions): Likewise. Accept vector of conditions rather
26669 than edge_equivalence structure for first argument.
26670 for the first argument.
26671 (avail_exprs_stack::lookup_avail_expr): New member function, moved
26672 from tree-ssa-dom.c.
26673 (avail_exprs_stack::record_cond): Likewise.
26674 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
26675 from tree-ssa-dom.c.
26676 (avail_exprs_stack): Add new member functions lookup_avail_expr
26678 (record_conditions): Declare.
26680 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
26683 * lra-constraints.c (process_alt_operands): Increase reject for
26684 reloading an input/output operand.
26686 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
26689 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
26690 insns to convert from signed/unsigned char/short to IEEE 128-bit
26692 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
26694 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
26697 * config/i386/i386.c (ix86_vector_duplicate_value): Create
26698 subreg of inner mode for values already in registers.
26700 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
26702 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
26703 iteration reg is used after the loop.
26705 2017-03-14 Martin Sebor <msebor@redhat.com>
26707 PR tree-optimization/79800
26708 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
26709 precision in negative-positive range.
26710 (format_floating): Call non-const overload with adjusted precision.
26712 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26715 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
26718 2017-03-14 Martin Sebor <msebor@redhat.com>
26720 PR middle-end/80020
26721 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
26722 * builtins.def (aligned_alloc): Use it.
26725 * Makefile.in (GTFILES): Add calls.c.
26726 * calls.c: Include "gt-calls.h".
26728 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
26730 PR rtl-optimization/79728
26731 * regs.h (struct target_regs): New field
26732 x_contains_allocatable_regs_of_mode.
26733 (contains_allocatable_regs_of_mode): New macro.
26734 * reginfo.c (init_reg_sets_1): Initialize it, and change
26735 contains_reg_of_mode so it includes global regs as well.
26736 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
26737 rather than contains_regs_of_mode.
26739 2017-03-14 Martin Liska <mliska@suse.cz>
26741 * doc/invoke.texi: Document options that can't be combined with
26742 -fcheck-pointer-bounds.
26744 2017-03-14 Martin Liska <mliska@suse.cz>
26746 PR middle-end/79831
26747 * doc/invoke.texi (-Wchkp): Document the option.
26749 2017-03-14 Martin Liska <mliska@suse.cz>
26751 * Makefile.in: Install gcov-dump.
26753 2017-03-14 Martin Liska <mliska@suse.cz>
26755 * multiple_target.c (expand_target_clones): Bail out for
26756 an invalid attribute.
26758 2017-03-14 Richard Biener <rguenther@suse.de>
26760 * alias.c (struct alias_set_entry): Pack properly.
26761 * cfgloop.h (struct loop): Likewise.
26762 * cse.c (struct set): Likewise.
26763 * ipa-utils.c (struct searchc_env): Likewise.
26764 * loop-invariant.c (struct invariant): Likewise.
26765 * lra-remat.c (struct cand): Likewise.
26766 * recog.c (struct change_t): Likewise.
26767 * rtl.h (struct address_info): Likewise.
26768 * symbol-summary.h (function_summary): Likewise.
26769 * tree-loop-distribution.c (struct partition): Likewise.
26770 * tree-object-size.c (struct object_size_info): Likewise.
26771 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
26772 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
26773 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
26774 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
26775 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
26776 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
26777 (struct _stmt_vec_info): Likewise.
26779 2017-03-14 Martin Liska <mliska@suse.cz>
26782 * multiple_target.c (create_dispatcher_calls): Check that
26783 a target can create a function dispatcher.
26785 2017-03-14 Martin Liska <mliska@suse.cz>
26788 * multiple_target.c (expand_target_clones): Drop local.local
26789 flag for default implementation.
26791 2017-03-14 Richard Biener <rguenther@suse.de>
26793 PR tree-optimization/80030
26794 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
26796 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
26798 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
26799 gcc_fallthrough() instead of __attribute__((fallthrough));
26801 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
26803 * doc/gcc.texi: Remove "up" link to (DIR).
26804 * doc/gccint.texi: Ditto.
26806 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
26808 * doc/install.texi (Specific) <avr>: Remove reference to
26811 2017-03-13 Jeff Law <law@redhat.com>
26813 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
26814 attribute rather than comments.
26816 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
26817 match_scratch operand is highest.
26819 2017-03-13 Martin Liska <mliska@suse.cz>
26821 PR middle-end/78339
26822 * ipa-pure-const.c (warn_function_noreturn): If the declarations
26823 is a CHKP clone, use original declaration.
26825 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26827 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
26828 (arc_conditional_register_usage): Use a different allocation order
26829 when optimizing for size.
26830 * common/config/arc/arc-common.c (arc_option_optimization_table):
26831 Section anchors default on when optimizing for size.
26833 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26835 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
26837 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26839 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
26840 * config/arc/arc.md (cpu_facility): Add cd variant.
26841 (*movqi_insn): Add code density variant.
26842 (*movhi_insn): Likewise.
26843 (*movqi_insn): Likewise.
26844 (*addsi3_mixed): Likewise.
26845 (subsi3_insn): Likewise.
26847 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26849 * config/arc/arc.md (movsi_cond_exec): Update constraint.
26851 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26853 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
26854 expressions with MINUS and UNARY ops.
26856 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26859 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
26861 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
26862 between vec_select and vector argument.
26863 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
26864 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
26865 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
26866 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
26867 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
26868 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
26870 2017-03-13 Richard Biener <rguenther@suse.de>
26873 * params.def (vect-max-peeling-for-alignment): Fix typo.
26875 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26877 * doc/install.texi (Specific) <mips-*-*>: Remove description of
26878 issue that only occurred with binutils below 2.18.
26880 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26882 * doc/install.texi (Specific) <cris-axis-elf>: No longer
26883 refer to binutils 2.11/2.12 minimum.
26885 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26887 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
26888 ftp.kernel.org and simplify binutils requirement.
26890 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
26892 * doc/invoke.texi (Warning Options): Fix spelling of link-time
26894 (Optimize Options): Ditto. Also remove redundancy.
26896 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26898 PR translation/79848
26899 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
26901 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
26902 to G_ to avoid double translation.
26904 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26906 PR translation/79923
26907 * auto-profile.c (get_combined_location): Convert leading
26908 character of diagnostics to lower case and remove trailing period.
26909 (read_profile): Likewise for various diagnostics.
26910 * config/arm/arm.c (arm_option_override): Remove trailing period
26911 from various diagnostics.
26912 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
26913 (msp430_expand_delay_cycles): Likewise.
26915 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26918 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
26919 full command-line argument, rather than just "str".
26920 (aarch64_validate_march): Likewise.
26921 (aarch64_validate_mtune): Likewise.
26923 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
26925 PR rtl-optimization/78911
26926 * lra-assigns.c (must_not_spill_p): New function.
26927 (spill_for): Use it.
26929 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26931 PR tree-optimization/79981
26932 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
26933 ATOMIC_COMPARE_EXCHANGE ifn result.
26934 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
26935 IFN_ATOMIC_COMPARE_EXCHANGE.
26937 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26940 * opts.c (parse_sanitizer_options): Add missing question mark to
26941 "did you mean" message.
26943 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26945 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
26947 (VMULEUH_UNS): Likewise.
26948 (VMULOUB_UNS): Likewise.
26949 (VMULOUH_UNS): Likewise.
26950 * config/rs6000/rs6000.c (builtin_function_type): Remove
26951 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
26953 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26956 * read-rtl-function.c (function_reader::read_rtx_operand): Update
26957 x with result of extra_parsing_for_operand_code_0.
26958 (function_reader::extra_parsing_for_operand_code_0): Convert
26959 return type from void to rtx, returning x. When reading
26960 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
26961 larger size containing struct block_symbol.
26963 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
26965 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
26966 -mfloat128-hardware without -m64.
26968 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
26971 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
26972 entries to the case statement that marks unsigned arguments to
26973 overloaded functions.
26975 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26977 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
26978 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
26980 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
26983 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
26984 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
26986 2017-03-10 Martin Liska <mliska@suse.cz>
26990 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
26991 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
26994 2017-03-10 Olivier Hainque <hainque@adacore.com>
26996 * tree-switch-conversion (array_value_type): Start by resetting
26997 candidate type to it's main variant.
26999 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27001 PR rtl-optimization/79909
27002 * combine.c (try_combine): Use simplify_replace_rtx on individual
27003 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
27004 of the whole CALL_INSN_FUNCTION_USAGE.
27006 PR tree-optimization/79972
27007 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
27008 get_range_info on SSA_NAMEs. Formatting fixes.
27010 2017-03-10 Richard Biener <rguenther@suse.de>
27011 Jakub Jelinek <jakub@redhat.com>
27013 PR tree-optimization/77975
27014 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
27015 edge to be constant.
27016 (get_val_for): For constant x return it. Formatting fix.
27017 (loop_niter_by_eval): Avoid pointless looping if the next iteration
27018 would use the same bases as the current one.
27020 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27022 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
27023 instead of vec_select for V1TImode.
27024 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
27026 (VSX_LE_128): Add V1TI to this mode iterator.
27027 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
27028 (*vsx_le_perm_store_<mode>): Likewise.
27029 (pre-reload splitter for VSX stores): Likewise.
27030 (post-reload splitter for VSX stores): Likewise.
27031 (*vsx_xxpermdi2_le_<mode>): Likewise.
27032 (*vsx_lxvd2x2_le_<mode>): Likewise.
27033 (*vsx_stxvd2x2_le_<mode>): Likewise.
27035 2017-03-09 Michael Eager <eager@eagercon.com>
27037 Correct failures with --enable-checking=yes,rtl.
27039 * config/microblaze/microblaze.c (microblaze_expand_shift):
27040 Replace GET_CODE test with CONST_INT_P and INTVAL test with
27041 test for const0_rtx.
27042 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
27043 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
27045 2017-03-09 Richard Biener <rguenther@suse.de>
27047 PR tree-optimization/79977
27048 * graphite-scop-detection.c (scop_detection::merge_sese):
27049 Handle the case of extra exits to blocks dominating the entry.
27051 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
27053 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27056 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
27058 PR rtl-optimization/79949
27059 * lra-constraints.c (process_alt_operands): Check memory when
27060 trying to predict a cycle. Print about the overall increase.
27062 2017-03-09 Richard Biener <rguenther@suse.de>
27064 PR middle-end/79971
27065 * gimple-expr.c (useless_type_conversion_p): Preserve
27066 TYPE_SATURATING for fixed-point types.
27068 2017-03-09 Richard Biener <rguenther@suse.de>
27071 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
27072 alignment of BLKmode params.
27074 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27077 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
27078 (VALL_NO_V2Q): Likewise.
27080 * config/aarch64/aarch64-simd.md
27081 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
27083 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
27084 VALL_NO_V2Q mode iterator.
27085 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
27087 2017-03-09 Martin Liska <mliska@suse.cz>
27089 PR tree-optimization/79631
27090 * tree-chkp-opt.c (chkp_is_constant_addr): Call
27091 tree_int_cst_sign_bit just for INTEGER constants.
27093 2017-03-09 Martin Liska <mliska@suse.cz>
27097 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
27100 2017-03-09 Marek Polacek <polacek@redhat.com>
27103 * tree.c (inchash::add_expr): Handle TREE_VEC.
27105 2017-03-09 Martin Liska <mliska@suse.cz>
27108 (chkp_narrow_size_and_offset): New function.
27109 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
27110 (void chkp_parse_bit_field_ref): New function.
27111 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
27112 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
27114 2017-03-09 Martin Liska <mliska@suse.cz>
27117 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
27118 (chkp_find_bounds_1): Remove gcc_unreachable.
27120 2017-03-09 Jakub Jelinek <jakub@redhat.com>
27123 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
27124 BUILT_IN_SYNC*, determine the access type from the size suffix and
27125 always build a MEM_REF with that type. Handle forgotten
27126 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
27129 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
27130 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
27131 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
27132 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
27133 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
27134 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
27135 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
27136 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
27137 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
27138 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
27139 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
27140 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
27141 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
27142 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
27143 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
27144 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
27145 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
27146 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
27147 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
27148 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
27149 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
27150 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
27151 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
27152 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
27153 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
27154 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
27155 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
27156 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
27157 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
27158 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
27159 definitions outside of __OPTIMIZE__ guarded section.
27162 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
27163 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
27164 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
27165 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
27168 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27170 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
27171 ("vfenez<mode>"): Add missing constraints.
27173 2017-03-08 Martin Sebor <msebor@redhat.com>
27176 * config/nds32/nds32.c (nds32_option_override):
27177 Fix misspelled diagnostic.
27179 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27182 * gimplify.c (gimplify_omp_for): Replace index var in outer
27183 taskloop statement with an artificial variable and add
27184 OMP_CLAUSE_PRIVATE clause for it.
27186 2017-03-08 Richard Biener <rguenther@suse.de>
27188 PR tree-optimization/79955
27189 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
27190 for accesses that are completely outside of the variable.
27192 2017-03-08 Andrew Haley <aph@redhat.com>
27194 PR tree-optimization/79943
27195 * tree-ssa-loop-split.c (compute_new_first_bound): When
27196 calculating the new upper bound, (END-BEG) should be added, not
27199 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27201 * config/avr/avr.md (setmemhi): Make sure match_dup
27202 operand number comes before match_scratch.
27204 2017-03-08 Richard Biener <rguenther@suse.de>
27206 PR tree-optimization/79920
27207 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
27208 with ncopies == 1 to ...
27209 (vect_transform_slp_perm_load): ... here. Properly compute
27210 all element loads by iterating VF times over the group. Do
27211 not handle ncopies (computed in a broken way) in
27212 vect_create_mask_and_perm.
27214 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27217 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
27218 is a uniform vector, use uniform_vector_p return value instead of
27219 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
27221 2017-03-07 Marek Polacek <polacek@redhat.com>
27223 PR middle-end/79809
27224 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
27225 (alloca_call_type): Likewise.
27227 2017-03-07 Martin Liska <mliska@suse.cz>
27229 * gcov.c (process_args): Put comment to correct location.
27231 2017-03-07 Martin Liska <mliska@suse.cz>
27233 PR middle-end/68270
27234 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
27235 Use array_at_struct_end_p instead of DECL_CHAIN (field).
27236 (chkp_narrow_bounds_for_field): Likewise.
27237 (chkp_parse_array_and_component_ref): Pass one more argument to
27240 2017-03-07 Richard Biener <rguenther@suse.de>
27242 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
27245 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
27247 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
27248 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
27250 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27253 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
27254 to end of description.
27255 (PARAM_MAX_STORES_TO_MERGE): Likewise.
27257 2017-03-07 Jakub Jelinek <jakub@redhat.com>
27259 PR rtl-optimization/79901
27260 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
27262 (*avx512f_<code><mode>3<mask_name>): ... this.
27263 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
27264 iterator instead of VI8_AVX2_AVX512BW.
27266 PR rtl-optimization/79901
27267 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
27268 min/max expander, expand it using expand_vec_cond_expr.
27271 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
27274 2017-03-06 Jakub Jelinek <jakub@redhat.com>
27277 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
27278 to void * for PCH reasons.
27279 * dwarf2out.c (output_loc_operands, output_die): Cast
27280 v.val_vec.array to unsigned char *.
27282 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
27285 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
27288 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
27290 PR rtl-optimization/79571
27291 * lra-constraints.c (process_alt_operands): Calculate static
27292 reject and subtract it from overall when only addresses will be
27295 2017-03-06 Julia Koval <julia.koval@intel.com>
27298 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
27299 incoming stack boundary to 128 for 64-bit targets.
27301 2017-03-06 Richard Biener <rguenther@suse.de>
27303 PR tree-optimization/79894
27304 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
27305 to NULL after folding it.
27307 2017-03-06 Richard Biener <rguenther@suse.de>
27309 PR tree-optimization/79824
27310 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
27311 check disabling peeling for gaps.
27313 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
27315 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27316 attributes): Document gettimeofday.
27318 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
27320 * config/s390/s390.c (s390_option_override_internal): Set
27321 PARAM_MIN_VECT_LOOP_BOUND
27323 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
27325 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
27326 * config/s390/s390.md: Likewise.
27328 2017-03-06 Jakub Jelinek <jakub@redhat.com>
27331 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
27332 (<avx2_avx512>_perm<mode>): Rename to ...
27333 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
27335 (<avx512>_perm<mode>_mask): Rename to ...
27336 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
27338 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
27339 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
27340 instead of VI8F_256_512.
27341 (avx512f_perm<mode>): New define_expand.
27342 (avx512f_perm<mode>_mask): Likewise.
27343 (avx512f_perm<mode>_1<mask_name>): New define_insn.
27344 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
27346 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27348 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
27349 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
27351 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
27353 2017-03-06 Martin Liska <mliska@suse.cz>
27356 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
27357 when having a SSA NAME w/o VAR_DECL assigned to it.
27359 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27361 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
27362 msa_dpsub_<su>_d): Fix MODE for vec_select.
27364 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27366 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
27368 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
27370 2017-03-06 Richard Biener <rguenther@suse.de>
27372 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
27373 * plugin.c (register_plugin_info): Likewise.
27374 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
27376 2017-03-05 Jakub Jelinek <jakub@redhat.com>
27378 * config/i386/sse.md (sse_storehps, sse_storelps,
27379 avx_<castmode><avxsizesuffix>_<castmode>,
27380 avx512f_<castmode><avxsizesuffix>_<castmode>,
27381 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
27382 in condition that at least one operand is not a MEM.
27384 2017-03-03 Jakub Jelinek <jakub@redhat.com>
27386 PR middle-end/79805
27387 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
27388 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
27390 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
27391 gimple_call_nothrow_p flag based on whether original builtin can throw.
27392 If it can, emit following stmts on the fallthrough edge.
27393 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
27394 don't create new bb if inserting just debug stmts on the edge, try to
27395 insert them on the fallthru bb or just reset debug stmts.
27397 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
27400 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
27401 restore recog_data (including the operand rtxes inside it) around
27402 the call to get_insn_template.
27404 2017-03-03 Martin Sebor <msebor@redhat.com>
27406 PR tree-optimization/79699
27407 * context.c (context::~context): Free MPFR caches to avoid
27408 a memory leak on program exit.
27410 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27412 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
27413 Use wide_int::ulow () instead of .elt (0).
27415 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
27417 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
27418 (*pushxf): Limit oF constraint to 32bit targets and add oC
27419 constraint for 64bit targets.
27420 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
27421 (*pushdf): Change rmF constraint to rmC.
27423 2017-03-03 Martin Liska <mliska@suse.cz>
27425 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
27426 Remove unused variable.
27428 2017-03-03 Jakub Jelinek <jakub@redhat.com>
27431 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
27432 is a memory operand, increase num_memory.
27433 (ix86_expand_args_builtin): Likewise.
27435 2017-03-03 Jan Hubicka <jh@suse.cz>
27438 * ipa-devirt.c (maybe_record_node): Properly handle
27439 __cxa_pure_virtual visibility.
27441 2017-03-03 Martin Liska <mliska@suse.cz>
27443 PR tree-optimization/79803
27444 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
27446 (pass_loop_prefetch::execute): Disabled optimization if an
27447 assumption about L1 cache size is not met.
27449 2017-03-03 Martin Liska <mliska@suse.cz>
27451 PR rtl-optimization/79574
27452 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
27453 (hash_scan_set): Likewise.
27454 (dump_hash_table): Likewise.
27455 (hoist_code): Likewise.
27457 2017-03-03 Richard Biener <rguenther@suse.de>
27459 * fixed-value.c (fixed_from_string): Restore use of elt (1)
27460 in place of uhigh ().
27461 (fixed_convert_from_real): Likewise.
27463 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
27466 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
27468 2017-03-03 Richard Biener <rguenther@suse.de>
27470 PR middle-end/79818
27471 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
27472 TYPE_OVERFLOW_UNDEFINED check.
27474 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27476 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
27478 (vector_ae_<mode>_p): Likewise.
27479 (vector_nez_<mode>_p): Likewise.
27480 (vector_ne_v2di_p): Likewise.
27481 (vector_ae_v2di_p): Likewise.
27482 (vector_ne_<mode>_p): Likewise.
27483 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
27485 (vsx_tsqrt<mode>2_fe): Likewise.
27487 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
27490 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
27492 2017-03-02 Jakub Jelinek <jakub@redhat.com>
27494 PR rtl-optimization/79780
27495 * cprop.c (one_cprop_pass): When second and further conditional trap
27496 in a single basic block is turned into an unconditional trap, turn it
27497 into a deleted note to avoid RTL verification failures.
27499 2017-03-02 Richard Biener <rguenther@suse.de>
27501 * fold-const.c (const_binop): Use ulow () instead of elt (0).
27503 2017-03-02 Richard Biener <rguenther@suse.de>
27505 PR tree-optimization/79345
27507 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
27508 param and abort the walk, returning -1 if it is hit.
27509 (walk_aliased_vdefs): Take a limit param and pass it on.
27510 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
27511 defaulting to 0 and return a signed int.
27512 * tree-ssa-uninit.c (struct check_defs_data): New struct.
27513 (check_defs): New helper.
27514 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
27515 about uninitialized memory.
27516 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
27517 bogus uninitialized warning.
27518 (fixed_convert_from_real): Likewise.
27520 2017-03-02 Bin Cheng <bin.cheng@arm.com>
27522 PR tree-optimization/66768
27523 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
27524 iv_use if base object can't be determined.
27526 2017-03-02 Jakub Jelinek <jakub@redhat.com>
27528 PR tree-optimization/79345
27529 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
27530 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
27531 (get_pattern_stats): Initialize it.
27532 * genemit.c (gen_expand): Verify match_scratch numbers come after
27533 match_operand/match_dup numbers.
27534 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
27535 match_scratch numbers.
27536 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
27538 * config/s390/s390.md (trunctdsd2): Likewise.
27540 2017-03-02 Richard Biener <rguenther@suse.de>
27542 * wide-int.h (wide_int_storage::operator=): Implement in terms
27545 2017-03-02 Richard Biener <rguenther@suse.de>
27547 PR tree-optimization/79777
27548 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
27549 the to insert expression to sth existing.
27551 2017-03-01 Martin Sebor <msebor@redhat.com>
27553 PR middle-end/79692
27554 * gimple-ssa-sprintf.c
27555 (directive::known_width_and_precision): New function.
27556 (format_integer): Use it.
27557 (get_mpfr_format_length): Consider the full range of precision
27558 when computing %g output with the # flag. Set the likely byte
27559 count to 3 rather than 1 when precision is indeterminate.
27560 (format_floating): Correct the lower bound of precision.
27562 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27564 * doc/invoke.texi: Document default code model for 64-bit Linux.
27566 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27569 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
27570 udiv rather than div since input pattern is unsigned.
27572 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
27574 * config/i386/i386.c (print_reg): Warn for values of
27575 unsupported size in integer register.
27577 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
27580 * config/rs6000/predicates.md (current_file_function_operand): Do
27581 not allow self calls to be local if the function is replaceable.
27583 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
27586 * config/rs6000/altivec.h (vec_ctz and others): Change the
27587 preprocessor macro that controls conditional compilation from
27588 _ARCH_PWR9 to __POWER9_VECTOR__.
27589 (vec_all_ne): Change parameterization of __altivec_scalar_pred
27590 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
27591 control (instead of _ARCH_PWR9 control) so that template
27592 definition uses power9-specific function.
27593 (vec_any_eq): Likewise.
27594 (vec_all_ne): Change macro definition to use a power9-specific
27595 expansion under #ifdef __POWER9_VECTOR__ control (instead of
27596 _ARCH_PWR9 control).
27597 (vec_any_eq) Likewise.
27598 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
27599 expansion for CMPNEF to remove support for xvcmpnesp instruction.
27600 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
27601 support for xvcmpnedp instruction.
27602 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
27603 macro expansion so that Power9 implementation of vec_all_ne does
27604 not use the AltiVec predicate framework.
27605 (VCMPNEH_P): Likewise.
27606 (VCMPNEW_P): Likewise.
27607 (VCMPNED_P): Likewise.
27608 (VCMPNEFP_P): Likewise.
27609 (VCMPNEDP_P): Likewise.
27610 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
27611 implementation of vec_any_eq to not use AltiVec predicate
27613 (VCMPAEH_P): Likewise.
27614 (VCMPAEW_P): Likewise.
27615 (VCMPAED_P): Likewise.
27616 (VCMPAEFP_P): Likewise.
27617 (VCMPAEDP_P): Likewise.
27618 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
27619 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
27620 not use the AltiVec predicate framework.
27621 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
27622 of vec_any_eq to not use AltiVec predicate framework.
27623 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27624 support for predefined __POWER9_VECTOR__ macro to indicate that
27625 Power9 instruction selection is enabled.
27626 (altivec_overloaded_builtins): Remove extraneous
27627 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
27628 function argument types RS6000_BTI_bool_V16QI and
27629 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
27630 entry for overloaded function argument types RS6000_BTI_bool_V4SI
27631 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
27632 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
27633 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
27634 Power9 for implementations of vec_cmpne. Change the signature for
27635 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
27636 (representing vec_all_ne) to remove the previously described first
27637 argument of type RS6000_BTI_INTSI, as this was an artifact of
27638 reliance on the AltiVec predicate framework, which is no longer
27639 used in the implementation of these functions. Add
27640 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
27641 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
27642 since, unlike the AltiVec predicate framework implementation, we
27643 do not share function descriptors between vec_alle and vec_anyeq.
27644 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
27645 set of modes that receive special treatment even when
27646 TARGET_P9_VECTOR is true. The special treatment emits code that
27647 does not depend on Power9 instructions.
27648 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
27649 define_expand to not rely on AltiVec predicate framework.
27650 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
27652 (vector_ne_v2di_p): Change this define_expand to not rely on
27653 AltiVec predicate framework.
27654 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
27656 (vector_ne_<mode>_p): Change this define_expand to not rely on
27657 AltiVec predicate framework.
27658 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
27660 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
27661 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
27662 define_insn pattern.
27663 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
27664 define_insn pattern because the xvcmpne<VSs>. instruction is not
27666 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
27667 instruction is not supported.
27669 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27671 * config/nvptx/nvptx.c: Include intl.h.
27673 2017-03-01 Martin Jambor <mjambor@suse.cz>
27676 * ipa-prop.h (ipa_bits): Removed field known.
27677 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
27678 to pointers. Adjusted their comments to warn about their sharing.
27679 (ipcp_transformation_summary): Change bits to a vector of pointers.
27680 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
27681 (ipa_get_ipa_bits_for_value): Declare.
27682 * tree-vrp.h (value_range): Mark as GTY((for_user)).
27683 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
27684 (ipa_bits_hash_table): Likewise.
27685 (ipa_vr_ggc_hash_traits): Likewise.
27686 (ipa_vr_hash_table): Likewise.
27687 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
27688 being pointers and vr_known being removed.
27689 (ipa_set_jf_unknown): Likewise.
27690 (ipa_get_ipa_bits_for_value): New function.
27691 (ipa_set_jfunc_bits): Likewise.
27692 (ipa_get_value_range): New overloaded functions.
27693 (ipa_set_jfunc_vr): Likewise.
27694 (ipa_compute_jump_functions_for_edge): Use the above functions to
27695 construct bits and vr parts of jump functions.
27696 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
27697 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
27699 (ipcp_grow_transformations_if_necessary): Also allocate
27700 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
27702 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
27703 them. Fix too long lines.
27704 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
27705 vr_known being removed.
27706 (ipa_read_jump_function): Use new setter functions to construct bits
27707 and vr parts of jump functions or set them to NULL.
27708 (write_ipcp_transformation_info): Adjust for bits being pointers.
27709 (read_ipcp_transformation_info): Likewise.
27710 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
27712 Include gt-ipa-prop.h.
27713 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
27715 (ipcp_store_bits_results): Likewise.
27716 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
27717 Do not write to existing jump functions but use a temporary instead.
27719 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27722 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
27723 attempt to use its first operand as BIT_FIELD_REF base.
27725 2017-03-01 Richard Biener <rguenther@suse.de>
27727 PR middle-end/79721
27728 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
27729 interpolating formula in wrapping arithmetic.
27730 (chrec_apply): Convert chrec_evaluate return value to wanted type.
27732 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27734 PR tree-optimization/79734
27735 * tree-vect-generic.c (expand_vector_condition): Optimize
27736 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
27737 Handle VEC_COND_EXPR where comparison has different inner width from
27738 type's inner width.
27740 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
27742 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
27743 markup, and similar issues. Remove @opindex entries for things
27744 that aren't options. Add missing -mmpy-option entries.
27746 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27748 PR tree-optimization/79737
27749 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
27750 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
27751 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
27752 instead of byte_size. Formatting fix.
27753 (shift_bytes_in_array_right): Formatting fix.
27755 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
27758 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
27759 condition on optimize for the leaf function test.
27761 2017-02-28 Martin Liska <mliska@suse.cz>
27764 * read-rtl-function.c (function_reader::handle_unknown_directive):
27765 Bail out when one uses -flto.
27767 2017-02-28 Martin Liska <mliska@suse.cz>
27769 * common.opt: Replace space with tabular for options of <number>
27771 * config/i386/i386.opt: Show <number> value for
27772 -mlarge-data-threshold.
27773 * opts.c (print_filtered_help): Do not display number in hexadecimal
27776 2017-02-28 Martin Liska <mliska@suse.cz>
27778 * common.opt: Fix --help=option -Q for options which are of
27781 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
27783 * config/i386/i386.c (print_reg): Error out for values
27784 of 8-bit size in invalid integer register.
27786 2017-02-28 Martin Sebor <msebor@redhat.com>
27788 PR tree-optimization/79691
27789 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
27791 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27794 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
27795 gcc_unreachable with output_operand_lossage.
27797 2017-02-28 Richard Biener <rguenther@suse.de>
27799 PR tree-optimization/79740
27800 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
27802 (visit_nary_op): Insert the nary into the hashtable if we
27803 pattern-matched sth.
27804 * tree-ssa-pre.c (eliminate_insert): Robustify.
27806 2017-02-28 Richard Biener <rguenther@suse.de>
27808 PR middle-end/79731
27809 * fold-const.c (decode_field_reference): Reject out-of-bound
27812 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27814 * config/i386/i386.c: Include intl.h.
27815 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
27816 instead of just cond ? "..." : "...".
27817 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
27818 * coverage.c (read_counts_file): Likewise.
27819 * omp-offload.c: Include intl.h.
27820 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
27821 of just cond ? "..." : "...".
27822 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
27823 of just cond ? "..." : "...".
27825 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
27828 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
27830 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
27831 'tune for' CPU name.
27832 * config/arm/arm-cpu-data.h: Regenerated.
27834 2017-02-28 Richard Biener <rguenther@suse.de>
27836 PR tree-optimization/79732
27837 * tree-inline.c (expand_call_inline): Do not shadow var.
27839 2017-02-28 Richard Biener <rguenther@suse.de>
27841 PR tree-optimization/79723
27842 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
27843 address-space properly.
27845 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
27847 * doc/optinfo.texi (Optimization groups): Fix option used for
27849 * doc/invoke.texi (-fopt-info): Document "omp".
27850 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
27851 (OPTGROUP_ALL): Add OPTGROUP_OMP.
27852 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
27853 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
27854 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
27856 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
27858 * dumpfile.c (optgroup_options): Instead of "openmp", associate
27859 OPTGROUP_OMP with "omp".
27861 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
27864 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
27865 for arithmetic shift of unsigned V2DI.
27867 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
27869 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
27870 arc/linux.h headers.
27871 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
27872 (LINK_SPEC): Likewise.
27873 (ARC_TLS_EXTRA_START_SPEC): Likewise.
27874 (EXTRA_SPECS): Likewise.
27875 (STARTFILE_SPEC): Likewise.
27876 (ENDFILE_SPEC): Likewise.
27877 (LIB_SPEC): Likewise.
27878 (TARGET_SDATA_DEFAULT): Likewise.
27879 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
27880 (MULTILIB_DEFAULTS): Likewise.
27881 (DWARF2_UNWIND_INFO): Likewise.
27882 * config/arc/big.h: New file.
27883 * config/arc/elf.h: Likewise.
27884 * config/arc/linux.h: Likewise.
27885 * config/arc/t-uClibc: Remove.
27887 2017-02-27 Bin Cheng <bin.cheng@arm.com>
27889 PR tree-optimization/77536
27890 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
27891 (tree_transform_and_unroll_loop): Use above function to compute the
27892 estimated niter of unrolled loop and use it when scaling profile.
27893 Also use count info rather than frequency if it's non-zero.
27894 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
27895 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
27896 (vect_transform_loop): Call above function.
27898 2017-02-27 Richard Biener <rguenther@suse.de>
27900 PR tree-optimization/45397
27901 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
27902 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
27903 (visit_nary_op): Add pattern matching for CSEing sign-changed
27904 or truncated operations with wider ones.
27906 2017-02-27 Richard Biener <rguenther@suse.de>
27908 PR tree-optimization/79690
27909 * tree-vect-stmts.c (vectorizable_store): Use vector type
27910 built from the DR with address-space.
27912 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
27914 * doc/invoke.texi (Optimize Options): Refine the description
27915 of asan-use-after-return.
27917 2017-02-25 Alan Modra <amodra@gmail.com>
27919 PR rtl-optimization/79584
27920 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
27921 base, not ad->base_term, the reg within base. Remove assertion
27922 that ad->base == ad->base_term. Replace gen_int_mode using
27923 bogus mode with const0_rtx.
27925 2017-02-25 Jakub Jelinek <jakub@redhat.com>
27927 PR middle-end/79396
27928 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
27929 FMA_EXPR like tcc_binary or tcc_unary.
27931 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
27934 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
27936 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
27937 (output_loc_operands): Handle DW_OP_call_ref and
27938 DW_OP_GNU_variable_value.
27939 (struct variable_value_struct): New type.
27940 (struct variable_value_hasher): Likewise.
27941 (variable_value_hash): New variable.
27942 (string_types): Remove.
27943 (copy_loc_descr): New function.
27944 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
27945 (prepend_loc_descr_to_each): New function.
27946 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
27947 instead of add_loc_descr_to_each if the first argument is single
27948 location list and the second has multiple.
27949 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
27950 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
27951 when looking for variable value which doesn't have other location info.
27952 (loc_list_from_tree): Formatting fix.
27953 (gen_array_type_die): Simplify DW_AT_string_length handling.
27954 (adjust_string_types): Remove.
27955 (gen_subprogram_die): Don't call adjust_string_types nor test/set
27956 string_types. Call resolve_variable_values.
27957 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
27958 (resolve_addr_in_expr): Likewise. Add A argument.
27959 (copy_deref_exprloc): Remove deref argument. Adjust for the
27960 original expression being DW_OP_GNU_variable_value with optionally
27961 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
27962 optionally after it.
27963 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
27964 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
27965 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
27966 (variable_value_hasher::hash, variable_value_hasher::equal): New
27968 (resolve_variable_value_in_expr, resolve_variable_value,
27969 resolve_variable_values, note_variable_value_in_expr,
27970 note_variable_value): New functions.
27971 (dwarf2out_early_finish): Call note_variable_value on all toplevel
27974 2017-02-24 Jakub Jelinek <jakub@redhat.com>
27977 * opts.h (handle_generated_option): Add GENERATED_P argument.
27978 * opts-common.c (handle_option): Adjust function comment.
27979 (handle_generated_option): Add GENERATED_P argument, pass it to
27981 (control_warning_option): Pass false to handle_generated_option
27983 * opts.c (maybe_default_option): Pass true to handle_generated_option
27985 * optc-gen.awk: Likewise.
27987 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27989 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
27990 a REG, look at the REG it is a SUBREG of.
27991 (splitter for cmpeqsi_t): Ditto.
27993 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27995 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
27996 the special USEs with the pattern of the insn, not the insn itself.
27998 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
28001 * doc/invoke.texi: Document -mload-store-pairs.
28003 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28004 Sandra Loosemore <sandra@codesourcery.com>
28006 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
28007 argument isn't a CONST_INT.
28008 (nios2_alternate_compare_const): Assert op is a CONST_INT.
28009 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
28010 (nios2_validate_compare): Bypass alternate compare logic if *op2
28011 is not a CONST_INT.
28012 (ldstwm_operation_p): Return false if first_base is not a REG or
28013 if first_offset is not a CONST_INT.
28015 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28017 * config/cris/cris.md: Use correct operand in a define_peephole2.
28019 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28021 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
28023 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28025 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
28026 this_insn if it is an INSN or JUMP_INSN.
28027 (force_offsettable): Look at base, not at addr.
28028 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
28029 on things that aren't necessarily CONST_INTs.
28031 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
28033 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
28034 -mfpmath=sse is the default also for x86-32 targets with SSE2
28035 instruction set when @option{-ffast-math} is enabled
28037 2017-02-24 Jeff Law <law@redhat.com>
28039 PR rtl-optimizatoin/79286
28040 * ira.c (update_equiv_regs): Drop may_trap_p exception to
28043 2017-02-24 Richard Biener <rguenther@suse.de>
28045 PR tree-optimization/79389
28046 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
28049 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
28051 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
28052 function comment to reflect reality.
28053 (loop_exits_before_overflow): Fix typo in function description.
28055 2017-02-24 Richard Biener <rguenther@suse.de>
28057 PR tree-optimization/79389
28058 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
28059 properly that a threading opportunity exists. Detect conditional
28060 copy/constant propagation opportunities.
28062 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
28064 * config/visium/visium.md (type): Add trap.
28065 (b): New mode attribute.
28066 (*btst): Rename into...
28067 (*btst<mode>): ...this and adjust.
28068 (*cbranchsi4_btst_insn): Rename into...
28069 (*cbranch<mode>4_btst_insn): ...this and adjust.
28070 (trap): New define_insn.
28072 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28074 PR tree-optimization/79389
28075 * ifcvt.c (struct noce_if_info): Add rev_cond field.
28076 (noce_reversed_cond_code): New function.
28077 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
28078 reversed_comparison_code. Formatting fix.
28079 (noce_try_store_flag): Test rev_cond != NULL in addition to
28080 reversed_comparison_code.
28081 (noce_try_store_flag_constants): Likewise.
28082 (noce_try_store_flag_mask): Likewise.
28083 (noce_try_addcc): Use rev_cond if non-NULL instead of
28084 reversed_comparison_code.
28085 (noce_try_cmove_arith): Likewise. Formatting fixes.
28086 (noce_try_minmax, noce_try_abs): Clear rev_cond.
28087 (noce_find_if_block): Initialize rev_cond.
28088 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
28089 instead of false as last argument never attempt to reverse it
28092 2017-02-23 Bin Cheng <bin.cheng@arm.com>
28094 PR tree-optimization/79663
28095 * tree-predcom.c (combine_chains): Process refs in reverse order
28096 only for ZERO length chains, and add explaining comment.
28098 2017-02-23 Jeff Law <law@redhat.com>
28100 PR tree-optimization/79578
28101 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
28102 in call to operand_equal_p.
28104 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
28107 * config/i386/cpuid.h: Fix another undefined behavior.
28109 2017-02-23 Richard Biener <rguenther@suse.de>
28111 PR tree-optimization/79683
28112 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
28113 vector types for data-refs.
28115 2017-02-23 Martin Liska <mliska@suse.cz>
28117 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
28119 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28121 PR middle-end/79665
28122 * internal-fn.c (get_range_pos_neg): Moved to ...
28123 * tree.c (get_range_pos_neg): ... here. No longer static.
28124 * tree.h (get_range_pos_neg): New prototype.
28125 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
28126 are known to be in between 0 and signed maximum inclusive, try to
28127 expand both unsigned and signed divmod and use the cheaper one from
28130 2017-02-22 Jeff Law <law@redhat.com>
28132 PR tree-optimization/79578
28133 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
28134 to compare base operands.
28136 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
28139 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
28140 gpc_reg_operand instead of fpr_reg_operand.
28142 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
28144 * config/mips/mips.c (mips_return_in_memory): Force FP
28145 vector types to be returned in memory for o32 ABI.
28147 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28149 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
28150 instead of DW_TAG_member for static data member declarations and don't
28151 set no_linkage_name for static inline data members.
28152 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
28155 2017-02-22 Martin Liska <mliska@suse.cz>
28157 * doc/invoke.texi: Replace inequality signs with square brackets
28160 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28162 PR tree-optimization/68644
28163 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
28165 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
28168 * lra-constraints.c (simplify_operand_subreg): Handle
28169 WORD_REGISTER_OPERATIONS targets.
28171 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28174 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
28175 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
28176 elimination by swapping fld*.
28178 2017-02-22 Richard Biener <rguenther@suse.de>
28180 PR tree-optimization/79673
28181 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
28182 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
28183 irrelevant address-space qualifiers and avoiding a
28184 ADDR_SPACE_CONVERT_EXPR from fold_convert.
28186 2017-02-22 Richard Biener <rguenther@suse.de>
28188 PR tree-optimization/79666
28189 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
28190 to not symbolically negate if that may introduce undefined
28193 2017-02-22 Martin Liska <mliska@suse.cz>
28196 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
28197 * data-streamer-out.c (streamer_write_gcov_count_stream):
28199 * value-prof.c (stream_out_histogram_value): Make assert more
28200 precise based on type of counter.
28202 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
28205 * config/i386/i386.md (standard_x87sse_constant_load splitter):
28206 Use nonimmediate_operand instead of memory_operand for operand 1.
28207 (float-extend standard_x87sse_constant_load splitter): Ditto.
28209 2017-02-21 Jeff Law <law@redhat.com>
28211 PR tree-optimization/79621
28212 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
28213 blocks with edges to themselves.
28215 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28218 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
28219 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
28220 Use gimple_call_builtin_p.
28223 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
28224 on temporarily removed DEBUG_INSNs.
28226 PR tree-optimization/79649
28227 * tree-loop-distribution.c (classify_partition): Give up on
28228 non-generic address space loads/stores.
28230 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
28232 * doc/loop.texi (Loop manipulation): Remove nonexistent
28233 tree_ssa_loop_version from the documentation.
28234 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
28236 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28239 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
28240 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
28241 * config/rs6000/rs6000.c: Include except.h.
28242 (rs6000_expand_split_stack_prologue): Call
28243 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
28245 2017-02-21 Martin Jambor <mjambor@suse.cz>
28248 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
28249 have been analyzed.
28251 2017-02-21 Martin Jambor <mjambor@suse.cz>
28253 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
28254 for backward compatibility only.
28255 * doc/invoke.texi (Option Summary): Remove all references to
28256 -fipa-cp-alignment.
28258 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
28262 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28264 * lra-constraints.c (curr_insn_transform): Handle
28265 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
28267 2017-02-21 Martin Liska <mliska@suse.cz>
28269 * config/i386/i386.opt: Replace -masm-dialect with -masm.
28271 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
28273 PR translation/79638
28274 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
28276 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
28279 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
28280 (arm_function_ok_for_sibcall): Return false for an indirect call by
28281 descriptor if all the argument registers are used.
28282 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
28283 alignment of the function.
28285 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28287 PR tree-optimization/61441
28288 * simplify-rtx.c (simplify_const_unary_operation): For
28289 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
28290 the sNaN unmodified.
28292 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28294 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28295 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28296 instead of SYSTEM_HEADER_DIR.
28298 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
28299 Martin Liška <mliska@suse.cz>
28301 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
28302 Fix typos and grammar, use active voice, and clarify.
28304 2017-02-20 Marek Polacek <polacek@redhat.com>
28306 PR middle-end/79537
28307 * gimplify.c (gimplify_expr): Handle unused *&&L;.
28310 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
28312 2017-02-20 Jakub Jelinek <jakub@redhat.com>
28315 * config/i386/i386.c (ix86_expand_builtin): Handle
28316 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
28317 ix86_builtins_isa[fcode].isa as a requirement of those
28318 flags and any other flag in the bitmask.
28319 (ix86_init_mmx_sse_builtins): Use 0 instead of
28320 ~OPTION_MASK_ISA_64BIT as mask.
28321 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
28322 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
28323 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
28324 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
28326 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28329 * lra-constraints.c (split_reg): Check requested split mode
28330 is supported by the register.
28332 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28334 * lra-constraints.c (simplify_operand_subreg): Remove early
28337 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28340 * lra-constraints.c (curr_insn_transform): Tighten condition
28341 for converting SUBREG reloads from OP_OUT to OP_INOUT.
28343 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28346 * lra-constraints.c (curr_insn_transform): Handle
28347 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
28349 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
28352 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
28354 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
28356 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
28359 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
28362 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28364 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
28367 2017-02-18 Jakub Jelinek <jakub@redhat.com>
28369 * final.c (last_columnnum, override_columnnum): New variables.
28370 (final_start_function): Set last_columnnum, pass it to begin_prologue
28371 hook and pass 0 to dwarf2out_begin_prologue.
28372 (final_scan_insn): Update override_columnnum. Pass last_columnnum
28373 to source_line debug hook.
28374 (notice_source_line): Compute last_columnnum and for debug_column_info
28375 return true on column changes.
28376 * debug.h (struct gcc_debug_hooks): Add column argument to
28377 source_line and begin_prologue hooks.
28378 (debug_nothing_int_charstar_int_bool): Remove prototype.
28379 (debug_nothing_int_int_charstar,
28380 debug_nothing_int_int_charstar_int_bool): New prototypes.
28381 (dwarf2out_begin_prologue): Add column argument.
28382 * debug.c (do_nothing_debug_hooks): Adjust source_line and
28383 begin_prologue hooks.
28384 (debug_nothing_int_charstar_int_bool): Remove.
28385 (debug_nothing_int_int_charstar,
28386 debug_nothing_int_int_charstar_int_bool): New functions.
28387 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
28388 through to dwarf2out_source_line.
28389 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
28390 (dwarf2out_source_line): Add column argument, emit it if requested.
28391 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
28393 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
28394 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
28395 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
28396 through to dwarf2out_begin_prologue.
28397 (vmsdbgout_source_line): Add column argument, pass it through to
28398 dwarf2out_source_line.
28399 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
28400 dbxout_source_line caller.
28401 (dbxout_source_line): Add column argument.
28403 * common.opt (gno-column-info, gcolumn-info): New options.
28404 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
28405 (check_die): Also test for multiple DW_AT_decl_column attributes.
28406 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
28407 DW_AT_decl_column if requested.
28408 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
28410 (gen_variable_die): Likewise.
28411 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
28412 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
28415 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
28416 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
28417 (ix86_handle_option): Handle OPT_m3dnowa.
28418 * doc/invoke.texi (-m3dnowa): Document.
28419 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
28420 -m3dnowa instead of -m3dnow -march=athlon.
28423 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
28424 instead of gcc_assert for K, r and R code checks. Formatting fixes.
28426 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28429 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
28430 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
28431 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
28432 generator for vsx_xxpermdi_<mode>_be.
28433 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
28434 force big-endian semantics.
28435 (vsx_xxpermdi_<mode>_be): New define_expand with same
28436 implementation as previous version of vsx_xxpermdi_<mode>.
28438 2017-02-17 Jakub Jelinek <jakub@redhat.com>
28440 PR tree-optimization/79327
28441 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
28442 variable, its initialization and use.
28444 2017-02-17 Julia Koval <julia.koval@intel.com>
28446 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
28447 (OPTION_MASK_ISA_PKU_UNSET): New.
28448 (ix86_handle_option): Handle -mrdpid.
28449 * config/i386/cpuid.h (bit_RDPID): New.
28450 * config/i386/driver-i386.c (host_detect_local_cpu):
28451 Detect RDPID feature.
28452 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
28453 * config/i386/i386-c.c (ix86_target_macros_internal):
28455 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
28456 (ix86_valid_target_attribute_inner_p): Add "rdpid".
28457 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
28458 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
28459 * config/i386/i386.md (define_insn "rdpid"): New.
28460 * config/i386/i386.opt Add -mrdpid.
28461 * config/i386/immintrin.h (_rdpid_u32): New.
28463 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
28465 PR rtl-optimization/79541
28466 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
28467 instead of transforming it into USE.
28469 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
28471 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
28472 If HONOR_SNANS (SFmode) force the input to a register.
28473 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
28474 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
28475 an frsp or similar insn.
28477 2017-02-17 Martin Liska <mliska@suse.cz>
28479 PR rtl-optimization/79577
28480 * params.def (selsched-max-sched-times): Increase minimum to 1.
28482 2017-02-17 Martin Liska <mliska@suse.cz>
28484 PR rtl-optimization/79574
28485 * gcse.c (want_to_gcse_p): Prevent integer overflow.
28487 2017-02-17 Martin Liska <mliska@suse.cz>
28489 PR tree-optimization/79529
28490 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
28491 ssa_defined_default_def_p to handle cases which are implicitly
28493 * tree-ssa.c (ssa_defined_default_def_p): New function.
28494 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
28495 which are implicitly defined.
28496 * tree-ssa.h (ssa_defined_default_def_p): Declare.
28498 2017-02-17 Richard Biener <rguenther@suse.de>
28500 PR middle-end/79576
28501 * params.def (max-ssa-name-query-depth): Limit to 10.
28503 2017-02-17 Richard Biener <rguenther@suse.de>
28505 PR tree-optimization/79552
28506 * tree-ssa-structalias.c (visit_loadstore): Properly verify
28509 2017-02-17 Richard Biener <rguenther@suse.de>
28512 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
28514 2017-02-17 Marek Polacek <polacek@redhat.com>
28516 PR middle-end/79536
28517 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
28518 (fold_negate_expr): New wrapper.
28520 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
28522 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
28523 Correct terminology and de-emphasize pre-standard behavior.
28525 2017-02-16 Alan Modra <amodra@gmail.com>
28527 PR rtl-optimization/79286
28528 * ira.c (def_dominates_uses): New function.
28529 (update_equiv_regs): Don't create an equivalence for insns that
28530 may trap where the register def does not dominate the use.
28532 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
28534 PR rtl-optimization/78127
28535 * lra.c (lra): Call lra_eliminate before finish the loop after
28538 2017-02-16 Richard Biener <rguenther@suse.de>
28540 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
28542 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
28543 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
28544 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
28545 (isl_val_int_from_wi): New function.
28546 (extract_affine_gmp): Rename to ...
28547 (extract_affine_wi): ... this, take a widest_int.
28548 (extract_affine_int): Just wrap extract_affine_wi.
28549 (add_param_constraints): Use isl_val_int_from_wi.
28550 (add_loop_constraints): Likewise, and extract_affine_wi.
28552 2017-02-15 Jeff Law <law@redhat.com>
28554 PR middle-end/79521
28555 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
28556 ira_init_register_move_cost_if_necessary.
28558 2017-02-15 Martin Sebor <msebor@redhat.com>
28560 PR middle-end/32003
28561 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
28562 removed in a prior commit.
28564 2017-02-15 Bin Cheng <bin.cheng@arm.com>
28566 PR tree-optimization/79347
28567 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
28568 counters during peeling.
28570 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
28572 * Makefile.in (site.exp): Remove "set ISLVER".
28574 2017-02-15 Jakub Jelinek <jakub@redhat.com>
28577 * real.c (real_from_integer): Call real_convert even for decimal.
28579 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
28582 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
28584 2017-02-14 Andrew Pinski <apinski@cavium.com>
28586 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
28587 cores and change the partno/implementer to be correct.
28588 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
28589 the 'B" as the implementer.
28590 * config/aarch64/aarch64-tune.md: Regenerate.
28592 2017-02-14 Carl Love <cel@us.ibm.com>
28594 * config/rs6000/rs6000.c: Add case statement entry to make the
28595 xvcvuxdsp built-in argument unsigned.
28596 * config/rs6000/vsx.md: Fix the source and return operand types so they
28597 match the instruction definitions from the ISA document. Fix typo
28598 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
28601 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
28604 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
28605 member early_clobber_alts.
28606 * lra-lives.c (reg_early_clobber_p): New.
28607 (process_bb_lives): Use it.
28608 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
28609 (debug_operand_data): Initialize early_clobber_alts.
28610 (setup_operand_alternative): Set up early_clobber_alts.
28611 (collect_non_operand_hard_regs): Ditto. Pass early clobber
28612 alternatives to new_insn_reg.
28613 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
28615 (lra_update_insn_regno_info): Pass the new arg.
28617 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28619 PR middle-end/79505
28620 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
28621 (new_oacc_loop_raw): Don't clear already cleared fields.
28624 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
28625 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
28626 _mm512_prefetch_i64gather_ps): New inline functions and macros.
28628 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
28631 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
28633 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
28636 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
28637 the extra instruction to the right place to store 128-bit constant
28640 2017-02-14 Martin Sebor <msebor@redhat.com>
28642 PR middle-end/79448
28643 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
28644 warning for strings of unknown length.
28646 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
28648 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
28650 2017-02-14 Jeff Law <law@redhat.com>
28653 * ira-costs.c (scan_one_insn): Initialize register move costs
28654 for pseudos seen in USE/CLOBBER insns.
28656 PR tree-optimization/79095
28657 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
28658 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
28659 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
28660 if the operands are known to be not equal, then the resulting range
28662 (intersect_ranges): If the new range is ~[0,0] and the old range is
28663 wide, then prefer ~[0,0].
28664 * tree-vrp.c (overflow_comparison_p_1): New function.
28665 (overflow_comparison_p): New function.
28666 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
28667 if NAME is used in an overflow test.
28668 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
28669 overflow check that can be expressed as an equality test, then adjust
28670 ops to be that equality test.
28672 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28674 * config/s390/s390-builtin-types.def: Remove flags argument.
28675 * config/s390/s390.c (s390_init_builtins): Likewise.
28677 2017-02-14 Martin Liska <mliska@suse.cz>
28679 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
28680 vector. Fix trailing white spaces.
28682 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
28684 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
28687 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28689 PR rtl-optimization/68664
28690 * config/arm/arm.c (arm_sched_can_speculate_insn):
28691 New function. Declare prototype.
28692 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
28694 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28696 PR rtl-optimization/68664
28697 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
28699 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
28701 2017-02-14 Amit Pawar <amit.pawar@amd.com>
28703 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
28704 max skip bytes for function, loop and jump.
28706 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28708 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
28709 ABS_EXPR for gimple dump.
28711 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28714 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
28716 PR tree-optimization/79408
28717 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
28718 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
28719 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
28720 also if rhs1 is INTEGER_CST.
28722 2017-02-14 Richard Biener <rguenther@suse.de>
28724 PR middle-end/79432
28725 * tree-into-ssa.c (insert_phi_nodes): When the function can
28726 have abnormal edges rewrite SSA names with broken use-def
28727 dominance out of SSA and register them for PHI insertion.
28729 2017-02-13 Martin Sebor <msebor@redhat.com>
28731 PR middle-end/79496
28732 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
28733 clearing info.nowrite flag when snprintf size argument is a range.
28735 2017-02-13 Jakub Jelinek <jakub@redhat.com>
28737 * cprop.c (cprop_jump): Add missing space in string literal.
28738 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
28739 (get_constraint_for_component_ref): Likewise.
28740 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
28741 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
28742 * lra-constraints.c (process_alt_operands): Likewise.
28743 * ipa-inline.c (inline_small_functions): Likewise.
28744 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
28745 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28746 * trans-mem.c (diagnose_tm_1_op): Likewise.
28747 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
28748 (grid_parallel_clauses_gridifiable): Likewise.
28750 * config/nvptx/mkoffload.c (process): Add space in between
28753 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
28754 "MOD4_SSE_REGS" and "ALL_REGS".
28756 * spellcheck.c (test_data): Add , in between "foo" and "food".
28758 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28761 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
28762 boundary crossing check and subsequent code generation agree.
28764 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28766 * config/aarch64/aarch64.c (has_memory_op): Delete.
28767 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
28770 2017-02-13 Jakub Jelinek <jakub@redhat.com>
28772 PR rtl-optimization/79388
28773 PR rtl-optimization/79450
28774 * combine.c (distribute_notes): When removing TEM_INSN for which
28775 corresponding dest has last value recorded, invalidate that last
28778 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28780 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
28781 of explicit '@'. Add missing assembly comment marker on branch costs
28784 2017-02-13 Nathan Sidwell <nathan@acm.org>
28786 * gengtype-lex.l (<in_struct>): Add '/'.
28788 2017-02-13 Martin Liska <mliska@suse.cz>
28791 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
28793 2017-02-13 Richard Biener <rguenther@suse.de>
28795 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
28797 * configure: Re-generate.
28798 * config.in: Likewise.
28799 * graphite-dependences.c: Simplify as if
28800 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
28801 * graphite-isl-ast-to-gimple.c: Likewise.
28802 * graphite-optimize-isl.c: Likewise.
28803 * graphite-poly.c: Likewise.
28804 * graphite-sese-to-poly.c: Likewise.
28805 * graphite.h: Likewise.
28806 * toplev.c: Include isl/version.h and use isl_version () for
28807 printing the ISL version.
28808 * doc/install.texi: Update ISL requirement.
28810 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
28812 * doc/standards.texi (Standards): Update reference to
28815 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
28817 * doc/extend.texi (Named Address Spaces): sourceware.org now
28819 * doc/install.texi (Binaries): Ditto.
28822 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
28824 * doc/cpp.texi: Replace "stringify"/"stringification" with C
28825 standard terminology "stringize"/"stringizing" throughout.
28826 * doc/cppinternals.texi: Likewise.
28828 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
28830 * doc/extend.texi: Fix some spelling mistakes and typos.
28831 * doc/invoke.texi: Likewise.
28833 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
28836 * params.def (inline-min-speedup) Change from 10 to 8.
28838 2017-02-11 Jakub Jelinek <jakub@redhat.com>
28840 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
28843 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
28846 * ipa-inline-analysis.c (get_minimal_bb): New function.
28847 (record_modified): Use it.
28848 (remap_edge_change_prob): Handle also ancestor functions.
28850 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
28852 * doc/contrib.texi (Contributors): Remove broken link into
28853 the Mauve CVS repository.
28855 2017-02-11 Jakub Jelinek <jakub@redhat.com>
28857 PR middle-end/79454
28858 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
28859 result computation whenever lhs doesn't have vector mode, not
28860 just when it has BLKmode.
28862 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
28864 * doc/makefile.texi (profiledbootstrap): Refer to the
28865 installation instructions only in textual form.
28867 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28870 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
28872 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
28874 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
28875 (Specific): Update mingw-w64 reference.
28877 (Specific): Remove broken link to Renesas RX processor.
28879 2017-02-10 Richard Biener <rguenther@suse.de>
28881 * toplev.c (process_options): Do not mention obsolete graphite
28882 options when printing sorry message about missing graphite support.
28883 Mention -floop-nest-optimize.
28885 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
28887 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
28888 (vtst_p16): Likewise.
28889 (vtstq_p8): Likewise.
28890 (vtstq_p16): Likewise.
28892 (vtstq_p64): Likewise.
28893 * config/arm/arm_neon.h (vgetq_lane_p64): New.
28894 (vset_lane_p64): New.
28895 (vsetq_lane_p64): New.
28897 2017-02-10 Jakub Jelinek <jakub@redhat.com>
28899 PR tree-optimization/79411
28900 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
28901 stmt operands are SSA_NAMEs used in abnormal phis.
28902 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
28905 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
28908 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
28911 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
28913 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
28915 2017-02-09 Jakub Jelinek <jakub@redhat.com>
28917 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
28921 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
28922 not arbitrary TREE_CONSTANT.
28925 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
28926 "omp declare target link" attribute unless is_global_var.
28927 * omp-offload.c (find_link_var_op): Likewise.
28929 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
28930 Chung-Lin Tang <cltang@codesourcery.com>
28932 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
28934 (gimplify_adjust_omp_clauses): Don't delete TILE.
28935 (gimplify_omp_for): Deal with TILE.
28936 * internal-fn.c (expand_GOACC_TILE): New function.
28937 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
28939 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
28940 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
28942 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
28943 avoid DIV for outermost collapse var.
28944 (expand_oacc_for): Insert tile element loop as needed. Adjust.
28945 Remove out of date comments, fix whitespace.
28946 * omp-general.c (omp_extract_for_data): Deal with tiling.
28947 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
28948 adjust OLF_DIM_BASE value.
28949 (struct omp_for_data): Add tiling field.
28950 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
28951 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
28952 for auto loops. Remove default auto determining, moved to
28953 oacc_loop_fixed_partitions.
28954 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
28955 stmts, add e_mask field.
28956 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
28957 (oacc_thread_numbers): Use oacc_dim_call.
28958 (oacc_xform_tile): New.
28959 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
28960 (finish_oacc_loop): Adjust for ifns vector.
28961 (oacc_loop_discover_walk): Append loop abstraction sites to list,
28962 add case for GOACC_TILE fns.
28963 (oacc_loop_xform_loop): Delete.
28964 (oacc_loop_process): Iterate over call list directly, and add
28965 handling for GOACC_TILE fns.
28966 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
28968 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
28969 vector partitioning to outer loops. Assign 2 partitions to loops
28970 when available. Add TILE handling.
28971 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
28972 (execite_oacc_device_lower): Process GOACC_TILE fns,
28973 ignore unknown specs.
28974 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
28975 * tree.c (omp_clause_num_ops): Adjust TILE ops.
28976 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
28978 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
28980 * configure.ac (ACX_BUGURL): Update.
28981 * configure: Regenerate.
28983 2017-02-09 Richard Biener <rguenther@suse.de>
28985 PR tree-optimization/69823
28986 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
28987 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
28989 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28991 * config/arc/arc-c.def: Add __NPS400__ definition.
28992 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
28993 (TARGET_NPS400): Define.
28995 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28997 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
28999 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
29000 pointer, arch_info.
29001 (arc_cpu_types): Fill the arch_info field with a pointer into the
29002 arc_arch_types table.
29003 (arc_selected_cpu): Declare.
29004 * config/arc/arc.c (arc_selected_cpu): Make global.
29005 (arc_selected_arch): Delete.
29006 (arc_base_cpu): Delete.
29007 (arc_override_options): Remove references to deleted variables,
29008 update access to arch information.
29009 (ARC_OPT): Update access to arch information.
29010 (ARC_OPTX): Likewise.
29011 * config/arc/arc.h (arc_base_cpu): Remove declaration.
29012 (TARGET_ARC600): Update access to arch information.
29013 (TARGET_ARC601): Likewise.
29014 (TARGET_ARC700): Likewise.
29015 (TARGET_EM): Likewise.
29016 (TARGET_HS): Likewise.
29017 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
29020 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
29023 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
29024 condition/operands for integer GE/LE/GEU/LEU operations.
29026 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
29028 PR translation/79397
29029 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
29032 2017-02-08 Martin Jambor <mjambor@suse.cz>
29035 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
29036 whether allocation happened.
29037 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
29038 nothing was allocated.
29040 2017-02-08 Jakub Jelinek <jakub@redhat.com>
29042 PR tree-optimization/79408
29043 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
29044 constant, but SSA_NAME with a known integer range, use the minimum
29045 of that range instead of op1 to determine if modulo can be replaced
29046 with its first operand.
29048 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29050 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
29052 2017-02-08 Richard Biener <rguenther@suse.de>
29054 PR tree-optimization/71824
29055 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29056 Check all loops contained in the merged region.
29058 2017-02-07 Andrew Pinski <apinski@cavium.com>
29060 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
29062 2017-02-07 Andrew Pinski <apinski@cavium.com>
29064 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
29065 (thunderxt88): Likewise.
29066 (thunderxt81): Disable LSE and change v8.1 to v8.
29067 (thunderxt83): Likewise.
29069 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29070 Richard Biener <rguenther@suse.de>
29072 PR middle-end/79399
29073 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
29074 type from int to size_t.
29075 * ira-costs.c (struct_costs_size): Change type from int to size_t.
29077 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29079 PR rtl-optimization/79386
29080 * cprop.c (bypass_conditional_jumps): Initialize
29081 bypass_last_basic_block already before splitting bbs after
29082 unconditional traps...
29083 (bypass_conditional_jumps): ... rather than here.
29086 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
29087 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
29088 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
29089 fix -masm=intel patterns.
29091 2017-02-07 Richard Biener <rguenther@suse.de>
29093 PR tree-optimization/79256
29094 PR middle-end/79278
29095 * builtins.c (get_object_alignment_2): Use min_align_of_type
29096 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
29097 and ADJUST_FIELD_ALIGN.
29099 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
29101 * doc/tm.texi: Regenerate.
29102 * stor-layout.c (layout_decl): Adjust.
29103 (update_alignment_for_field): Likewise.
29104 (place_field): Likewise.
29105 (min_align_of_type): Likewise.
29106 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
29107 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
29108 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
29109 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
29110 * config/frv/frv.c (frv_adjust_field_align): Likewise.
29111 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
29112 * config/i386/i386.c (x86_field_alignment): Likewise.
29113 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
29114 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
29115 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
29116 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
29117 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
29118 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
29122 2017-01-30 Richard Biener <rguenther@suse.de>
29124 PR tree-optimization/79256
29125 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
29126 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
29129 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
29131 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
29132 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
29133 builtins to SImode and emit a zero-extend, if necessary.
29135 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29137 * docs/invoke.texi (RISC-V Options): Alphabetize.
29139 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29141 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
29144 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29146 * config/riscv/riscv.c: New file.
29147 * common/config/riscv/riscv-common.c: Likewise.
29148 * config.gcc: Likewise.
29149 * config/riscv/constraints.md: Likewise.
29150 * config/riscv/elf.h: Likewise.
29151 * config/riscv/generic.md: Likewise.
29152 * config/riscv/linux.h: Likewise.
29153 * config/riscv/multilib-generator: Likewise.
29154 * config/riscv/peephole.md: Likewise.
29155 * config/riscv/pic.md: Likewise.
29156 * config/riscv/predicates.md: Likewise.
29157 * config/riscv/riscv-builtins.c: Likewise.
29158 * config/riscv/riscv-c.c: Likewise.
29159 * config/riscv/riscv-ftypes.def: Likewise.
29160 * config/riscv/riscv-modes.def: Likewise.
29161 * config/riscv/riscv-opts.h: Likewise.
29162 * config/riscv/riscv-protos.h: Likewise.
29163 * config/riscv/riscv.h: Likewise.
29164 * config/riscv/riscv.md: Likewise.
29165 * config/riscv/riscv.opt: Likewise.
29166 * config/riscv/sync.md: Likewise.
29167 * config/riscv/t-elf-multilib: Likewise.
29168 * config/riscv/t-linux: Likewise.
29169 * config/riscv/t-linux-multilib: Likewise.
29170 * config/riscv/t-riscv: Likewise.
29171 * configure.ac: Likewise.
29172 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
29173 Waterman as RISC-V maintainers.
29174 * doc/install.texi: Add RISC-V entries.
29175 * doc/invoke.texi: Add RISC-V options section.
29176 * doc/md.texi: Add RISC-V constraints section.
29177 * configure: Regenerated.
29179 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
29182 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
29183 false values to be constant vectors with all 0 or all 1 bits set.
29184 (vcondu<mode><mode>): Likewise.
29185 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
29187 (fpmask_comparison_operator): Update comment.
29188 (vecint_comparison_operator): New predicate.
29189 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
29190 vector conditionals when the true and false values are constant
29191 vectors with all 0 bits or all 1 bits set.
29193 2017-02-06 Martin Sebor <msebor@redhat.com>
29195 PR tree-optimization/79376
29196 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
29198 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
29200 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
29201 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
29202 to simplify split condition.
29204 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29206 * omp-expand.c (oxpand_omp_atomic_fetch_op,
29207 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
29210 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
29212 PR rtl-optimization/68664
29213 * target.def (can_speculate_insn): New hook.
29214 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
29215 * doc/tm.texi: Regenerate.
29216 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
29217 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
29218 (rs6000_sched_can_speculate_insn): New function.
29220 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29222 PR tree-optimization/79284
29223 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
29224 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
29225 vectorizable_mask_load_store, vectorizable_operation,
29226 vect_is_simple_cond, get_same_sized_vectype): Use it instead
29227 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
29228 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
29229 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
29230 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
29231 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
29232 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
29233 is_gimple_assign (stmt). Replace another such test with
29234 is_gimple_assign (stmt).
29236 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
29239 * config/avr/avr.c (rtl-iter.h): Include it.
29240 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
29241 (avr_legitimate_combined_insn): ...and implementation.
29243 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
29245 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
29246 * config/s390/s390.c (s390_const_operand_ok)
29247 (s390_canonicalize_comparison, s390_extract_part)
29248 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
29249 (s390_contiguous_bitmask_p, s390_rtx_costs)
29250 (legitimize_pic_address): Likewise.
29251 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
29252 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
29253 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
29254 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
29255 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
29257 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
29259 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
29260 REGNO($0) == REGNO($1).
29262 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29264 * config/s390/linux.h(SIZE_TYPE): Add comment.
29266 2017-02-06 Julian Brown <julian@codesourcery.com>
29267 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29268 Virendra Pathak <virendra.pathak@broadcom.com>
29270 * config/aarch64/aarch64-cores.def: Change the scheduler
29272 * config/aarch64/aarch64.md: Include thunderx2t99.md.
29273 * config/aarch64/thunderx2t99.md: New file.
29275 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29277 * doc/standards.texi (Go Language): Update link to language
29280 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
29282 * tree-eh.c (lower_resx): Sanitize profile.
29283 (cleanup_empty_eh_move_lp): Likewise.
29285 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
29288 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
29290 * cfgloopmanip.h (loop_version): Update prototype.
29291 * modulo-sched.c (sms_schedule): Update call of loop_version.
29292 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
29293 * tree-parloops.c (gen_parallel_loop): Likewise.
29294 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
29295 * tree-ssa-loop-split.c (split_loop): Likewise.
29296 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
29297 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
29299 2017-02-05 Martin Liska <mliska@suse.cz>
29302 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
29304 (print_operand_address): Initialize a struct to zero.
29306 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29308 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
29309 garbage collector only in textual form.
29311 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29313 * doc/extend.texi (x86 specific memory model extensions for
29314 transactional memory): Simplify a phrase.
29316 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
29319 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
29320 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
29321 (atomic_storedi_1): Likewise.
29323 2017-02-04 Jakub Jelinek <jakub@redhat.com>
29325 PR tree-optimization/79338
29326 * tree-parloops.c (gather_scalar_reductions): Don't call
29327 vect_analyze_loop_form for loop->inner before destroying loop's
29330 2017-02-03 Martin Sebor <msebor@redhat.com>
29332 PR tree-optimization/79327
29333 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
29334 when precision has resulted in leading zeros.
29335 (format_integer): Adjust the likely counter to assume an unknown
29336 argument that may be zero is non-zero.
29338 2017-02-03 Jason Merrill <jason@redhat.com>
29341 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
29342 avoid copying non-taken branch.
29344 2017-02-03 Jakub Jelinek <jakub@redhat.com>
29346 PR tree-optimization/79340
29347 * tree-vect-loop.c (vectorizable_reduction): Release
29348 vec_defs elements after safe_splicing them into other vectors.
29351 PR tree-optimization/79327
29352 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
29353 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
29355 (format_integer): Use wide_int_to_tree instead of build_int_cst
29356 + to_?hwi. If argmin is NULL, just set argmin and argmax to
29357 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
29358 of shortest and longest sequence.
29360 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
29362 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
29363 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
29365 2017-02-03 Walter Lee <walt@tilera.com>
29368 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
29369 after initial stackframe link reg save.
29370 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
29372 2017-02-03 Jakub Jelinek <jakub@redhat.com>
29375 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
29376 wu for stxssp alternative.
29378 2017-02-03 Martin Sebor <msebor@redhat.com>
29380 PR tree-optimization/79352
29381 * gimple-fold.c (get_range_strlen): Add argument.
29382 (get_range_strlen): Change return type to bool.
29383 (get_maxval_strlen): Pass in a dummy argument.
29384 * gimple-fold.h (get_range_strlen): Change return type to bool.
29385 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
29386 * tree.h (array_at_struct_end_p): Add argument.
29387 * tree.c (array_at_struct_end_p): Handle it.
29389 2017-02-03 Martin Liska <mliska@suse.cz>
29392 * multiple_target.c (create_dispatcher_calls): Redirect edge
29393 from a caller of a dispatcher.
29394 (expand_target_clones): Make the clones local.
29395 (ipa_target_clone): Do both target clones and resolvers.
29396 (ipa_dispatcher_calls): Remove the pass.
29397 (pass_dispatcher_calls::gate): Likewise.
29398 (make_pass_dispatcher_calls): Likewise.
29399 * passes.def (pass_target_clone): Put as very first IPA early
29402 2017-02-03 Martin Liska <mliska@suse.cz>
29404 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
29405 in case of a function with ifunc attribute.
29407 2017-02-03 Martin Liska <mliska@suse.cz>
29409 * cgraph.c (cgraph_node::dump): Dump function version info.
29410 * symtab.c (symtab_node::dump_base): Add missing new line.
29412 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
29414 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
29415 (ifcombine_ifandif): Use it.
29417 2017-02-03 Martin Liska <mliska@suse.cz>
29419 * doc/invoke.texi: Document default value for
29420 use-after-scope-direct-emission-threshold.
29422 2017-02-03 Martin Liska <mliska@suse.cz>
29424 PR tree-optimization/79339
29425 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
29426 (format_floating): Likewise.
29428 2017-02-03 Martin Liska <mliska@suse.cz>
29431 * ipa-prop.c (ipa_node_params_t::insert): Remove current
29433 (ipa_node_params_t::remove): Likewise.
29434 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
29435 initialization from removed ipa_node_params_t::insert.
29436 (ipa_node_params::~ipa_node_params): Move from removed
29437 ipa_node_params_t::release.
29438 * symbol-summary.h (symbol_summary::m_released): New member.
29439 Do not release a summary twice. Do not allow to call finalizer
29440 for types of a summary that live in GGC memory.
29442 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29444 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
29447 2017-02-02 Martin Sebor <msebor@redhat.com>
29449 PR middle-end/79275
29450 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
29451 (format_string): Tighten up the range of output for non-constant
29452 strings and correct the expected range for wide non-constant strings.
29454 2017-02-02 Martin Sebor <msebor@redhat.com>
29456 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
29458 PR middle-end/32003
29459 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
29461 (-fdump-tree-@var): Add to index and document how to come up
29462 with pass-specific option and dump file names.
29463 (-fdump-passes): Clarify where to look for output.
29465 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
29467 PR middle-end/77445
29468 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
29469 statistics of the analyzed path; allow threading for speed when
29470 any of BBs along the path are optimized for speed.
29472 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
29474 PR middle-end/78468
29475 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
29476 settings of the virtual registers.
29479 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
29481 * explow.c (get_dynamic_stack_size): Take known alignment of stack
29482 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
29485 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29487 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
29488 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
29490 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29492 * config/s390/s390.md: Add missing comments with the expanded
29494 * config/s390/vector.md: Likewise.
29495 * config/s390/vx-builtins.md: Likewise.
29497 2017-02-02 Jakub Jelinek <jakub@redhat.com>
29500 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
29501 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
29502 conditions on a single line.
29504 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29506 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
29507 __S390_VX__ to __VX__.
29509 2017-02-01 Andrew Pinski <apinski@cavium.com>
29511 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
29512 stmt_info to record_stmt_cost.
29513 (vect_get_known_peeling_cost): Pass stmt_info if known to
29515 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
29516 cpu_vector_cost field into
29517 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
29518 field into vec_int_stmt_cost and vec_fp_stmt_cost.
29519 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
29520 splitting of scalar_stmt_cost and vec_stmt_cost.
29521 (thunderx_vector_cost): Likewise.
29522 (cortexa57_vector_cost): LIkewise.
29523 (exynosm1_vector_cost): Likewise.
29524 (xgene1_vector_cost): Likewise.
29525 (thunderx2t99_vector_cost): Improve after the splitting of the two
29527 (aarch64_builtin_vectorization_cost): Update for the splitting of
29528 scalar_stmt_cost and vec_stmt_cost.
29530 2017-02-01 Torvald Riegel <triegel@redhat.com>
29531 Richard Henderson <rth@redhat.com>
29533 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
29534 conditional on existance of a fast atomic load.
29535 * optabs-query.c (can_atomic_load_p): New function.
29536 * optabs-query.h (can_atomic_load_p): Declare it.
29537 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
29538 no fast atomic load is available for the particular size of access.
29539 (expand_atomic_compare_and_swap): Likewise.
29540 (expand_atomic_load): Likewise.
29541 (expand_atomic_store): Likewise.
29542 (expand_atomic_fetch_op): Likewise.
29543 * testsuite/lib/target-supports.exp
29544 (check_effective_target_sync_int_128): Remove x86 because it provides
29545 no fast atomic load.
29546 (check_effective_target_sync_int_128_runtime): Likewise.
29548 2017-02-01 Richard Biener <rguenther@suse.de>
29550 * graphite.c: Include tree-vectorizer.h for find_loop_location.
29551 (graphite_transform_loops): Provide opt-info for optimized nests.
29552 * tree-parloop.c (parallelize_loops): Provide opt-info for
29553 parallelized loops.
29555 2017-02-01 Richard Biener <rguenther@suse.de>
29557 PR middle-end/79315
29558 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
29559 was not set before.
29561 2017-02-01 Richard Biener <rguenther@suse.de>
29563 PR tree-optimization/71824
29564 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29565 Verify the loops are valid in the merged SESE region.
29566 (scop_detection::can_represent_loop_1): Check analyzing the
29567 evolution of the number of iterations in the region succeeds.
29569 2017-01-31 Ian Lance Taylor <iant@golang.org>
29571 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
29572 REG_ARGS_SIZE note to 32-bit push insns and call insn.
29574 2017-01-31 David Malcolm <dmalcolm@redhat.com>
29576 PR preprocessor/79210
29577 * input.c (get_substring_ranges_for_loc): Replace line_width
29578 assertion with error-handling.
29580 2017-01-31 Richard Biener <rguenther@suse.de>
29582 PR tree-optimization/77318
29583 * graphite-sese-to-poly.c (extract_affine): Fix assert.
29584 (create_pw_aff_from_tree): Take loop parameter.
29585 (add_condition_to_pbb): Pass loop of the condition to
29586 create_pw_aff_from_tree.
29588 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29590 * config/s390/s390.c (s390_asan_shadow_offset): New function.
29591 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
29593 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
29597 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
29599 (convert_int_to_float128): Likewise.
29600 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
29601 (convert_int_to_float128): Likewise.
29602 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
29603 (UNSPEC_IEEE128_CONVERT): Likewise.
29604 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
29605 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
29606 Use local variables for IBM extended format.
29607 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
29608 (fix_trunc<mode>si2_fprs): Likewise.
29609 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
29610 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
29611 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
29612 to know that we can now have integers of all sizes in vector
29614 (fix<uns>_<mode>di2_hw): Likewise.
29615 (float<uns>_<mode>si2_hw): Likewise.
29616 (fix_<mode>si2_hw): Likewise.
29617 (fixuns_<mode>si2_hw): Likewise.
29618 (float<uns>_<mode>di2_hw): Likewise.
29619 (float_<mode>di2_hw): Likewise.
29620 (float_<mode>si2_hw): Likewise.
29621 (floatuns_<mode>di2_hw): Likewise.
29622 (floatuns_<mode>si2_hw): Likewise.
29623 (xscvqp<su>wz_<mode>): Delete, no longer used.
29624 (xscvqp<su>dz_<mode>): Likewise.
29625 (xscv<su>dqp_<mode>): Likewise.
29626 (ieee128_mfvsrd_64bit): Likewise.
29627 (ieee128_mfvsrd_32bit): Likewise.
29628 (ieee128_mfvsrwz): Likewise.
29629 (ieee128_mtvsrw): Likewise.
29630 (ieee128_mtvsrd_64bit): Likewise.
29631 (ieee128_mtvsrd_32bit): Likewise.
29633 2017-01-31 Martin Liska <mliska@suse.cz>
29636 * ipa-prop.c (ipa_free_all_node_params): Call release method
29637 instead of ~sumbol_summary to not to trigger double times
29640 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
29642 PR tree-optimization/71691
29643 * bitmap.h (class auto_bitmap): New.
29644 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
29645 is_maybe_undefined instead of ssa_undefined_value_p.
29647 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29649 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
29650 __S390_ARCH_LEVEL__ to __ARCH__.
29652 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29654 PR tree-optimization/79267
29655 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
29656 if should_remove_lhs_p is true.
29658 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
29661 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
29662 (add_alignment_attribute): New.
29663 (base_type_die): Add alignment attribute.
29664 (subrange_type_die): Likewise.
29665 (modified_type_die): Likewise.
29666 (gen_array_type_die): Likewise.
29667 (gen_descr_array_type_die: Likewise.
29668 (gen_enumeration_type_die): Likewise.
29669 (gen_subprogram_die): Likewise.
29670 (gen_variable_die): Likewise.
29671 (gen_field_die): Likewise.
29672 (gen_ptr_to_mbr_type_die): Likewise.
29673 (gen_struct_or_union_type_die): Likewise.
29674 (gen_subroutine_type_die): Likewise.
29675 (gen_typedef_die): Likewise.
29676 (base_type_cmp): Compare alignment attribute.
29678 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29681 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
29682 (setb_unsigned) New pattern for setb with CCUNS.
29683 * config/rs6000/rs6000.c (expand_block_compare): Use a different
29684 subfc./subfe sequence to avoid overflow problems. Generate a
29685 shorter sequence with cmpld/setb for power9.
29686 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
29687 for generating subfc. instruction.
29688 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
29689 now uses this instruction.
29691 2017-01-30 Ian Lance Taylor <iant@google.com>
29694 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
29695 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
29697 2017-01-30 Martin Sebor <msebor@redhat.com>
29699 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
29700 Move constant to the right of a relational operator.
29701 (get_mpfr_format_length, format_character, format_string): Ditto.
29702 (should_warn_p, maybe_warn): Same.
29704 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
29706 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
29709 * asan.c (get_translation_unit_decl): Remove function.
29710 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
29712 2017-01-30 Martin Liska <mliska@suse.cz>
29714 PR gcov-profile/79259
29715 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
29716 -fprofile-generate.
29718 2017-01-30 Martin Liska <mliska@suse.cz>
29721 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
29722 Initialize variables with NULL value.
29724 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
29727 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
29729 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
29731 2017-01-30 Richard Biener <rguenther@suse.de>
29733 PR tree-optimization/79276
29734 * tree-vrp.c (process_assert_insertions): Properly adjust common
29735 when removing a duplicate.
29737 * gcc.dg/torture/pr79276.c: New testcase.
29739 2017-01-30 Richard Biener <rguenther@suse.de>
29741 PR tree-optimization/79256
29742 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
29743 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
29745 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
29747 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
29750 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
29751 ("*r<noxa>sbg_<mode>_sll_bitmask")
29752 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
29753 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
29754 Use contiguous_bitmask_nowrap_operand.
29756 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29759 * config/rs6000/altivec.h (vec_xl): Revise #define.
29760 (vec_xst): Likewise.
29762 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
29764 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
29766 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
29768 PR rtl-optimization/79194
29769 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
29770 traps before call to bypass_conditional_jumps.
29772 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
29774 PR tree-optimization/71374
29775 * lra-constraints.c (check_conflict_input_operands): New.
29776 (match_reload): Use it.
29778 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
29781 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
29782 account to calculate conflict_set.
29784 2017-01-27 Bin Cheng <bin.cheng@arm.com>
29786 PR rtl-optimization/78559
29787 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
29788 other_insn in combine.
29790 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29792 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
29793 uint16_type_node for BT_UINT16.
29795 2017-01-27 David Malcolm <dmalcolm@redhat.com>
29797 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
29798 "RTL Tests" to menu.
29799 (GIMPLE Tests): New node.
29800 (RTL Tests): New node.
29802 2017-01-27 Richard Biener <rguenther@suse.de>
29804 PR tree-optimization/79245
29805 * tree-loop-distribution.c (distribute_loop): Apply cost
29806 modeling also to detected patterns.
29808 2017-01-27 Richard Biener <rguenther@suse.de>
29810 PR tree-optimization/71433
29811 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
29812 (compare_assert_loc): New function.
29813 (process_assert_insertions): Sort and optimize assert locations
29814 to remove duplicates and push down identical assertions on
29815 edges to their destination block.
29817 2017-01-27 Richard Biener <rguenther@suse.de>
29819 PR tree-optimization/79244
29820 * tree-vrp.c (remove_range_assertions): Forcefully propagate
29821 out SSA names even if abnormal.
29823 2017-01-27 Jakub Jelinek <jakub@redhat.com>
29825 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
29826 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
29827 instead of MPFR_RNDN.
29829 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
29832 * arm.c (arm_option_override): Don't call build_target_option_node
29833 until after doing all option overrides.
29834 (arm_valid_target_attribute_tree): Likewise.
29836 2017-01-27 Martin Liska <mliska@suse.cz>
29838 * doc/invoke.texi (-fprofile-arcs): Document profiling support
29839 for {cd}tors and C++ {cd}tors.
29841 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
29843 * config/s390/s390.md ("*setmem_long_and")
29844 ("*setmem_long_and_31z"): Use zero_extend instead of and.
29846 2017-01-26 Martin Sebor <msebor@redhat.com>
29848 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
29851 2017-01-26 Martin Sebor <msebor@redhat.com>
29853 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
29854 HAVE_DFmode before using XFmode or DFmode.
29855 (parse_directive): Avoid using the z length modifier to avoid
29856 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
29858 PR middle-end/78703
29859 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
29860 to accept adjustment as an array.
29861 (get_int_range): New function.
29862 (struct directive): Make width and prec arrays.
29863 (directive::set_width, directive::set_precision): Call get_int_range.
29864 (format_integer, format_floating): Handle width and precision ranges.
29865 (format_string, parse_directive): Same.
29867 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29870 * dwarf2out.c (generate_skeleton_bottom_up): For children with
29871 comdat_type_p set, just clone them, but keep the children in the
29875 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
29876 which have direct callers with -fvar-tracking-assignments enabled
29878 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
29879 inside of type units.
29881 2017-01-26 Martin Sebor <msebor@redhat.com>
29883 PR middle-end/78703
29884 * gimple-ssa-sprintf.c (struct result_range): Add likely and
29886 (struct format_result): Replace number_chars, number_chars_min,
29887 and number_chars_max with a single member of struct result_range.
29889 (format_result::operator+=): Adjust.
29890 (struct fmtresult): Remove bounded. Handle likely and unlikely
29892 (fmtresult::adjust_for_width_or_precision): New function.
29893 (fmtresult:type_max_digits): New function.
29894 (bytes_remaining): Handle likely and unlikely counters.
29895 (min_bytes_remaining): Remove.
29896 (format_percent): Simplify.
29897 (format_integer, format_floating): Set likely and unlikely counters.
29898 (get_string_length, format_character, format_string): Same.
29899 (format_plain, should_warn_p): New function.
29900 (maybe_warn): Call should_warn_p. Update diagnostic messages
29901 and handle those for all directives, including plain strings.
29902 (format_directive): Handle likely and unlikely counters.
29903 Remove unnecessary quoting from diagnostics. Add an informational
29905 (add_bytes): Remove.
29906 (pass_sprintf_length::compute_format_length): Simplify.
29907 (try_substitute_return_value): Handle likely and unlikely counters.
29909 2017-01-26 Carl Love <cel@us.ibm.com>
29911 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
29912 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
29914 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
29917 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
29918 endianess for subregs into account.
29919 * lra-constraints.c (lra_constraints): Do risky transformations
29920 always on the first iteration.
29921 * lra-lives.c (check_pseudos_live_through_calls): Add arg
29922 last_call_used_reg_set.
29923 (process_bb_lives): Define and use last_call_used_reg_set.
29924 * lra.c (lra): Always continue after lra_constraints on the first
29927 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
29929 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
29931 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
29933 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29935 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
29936 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
29937 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
29938 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
29939 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
29940 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
29941 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
29942 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
29943 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
29945 2017-01-26 Marek Polacek <polacek@redhat.com>
29948 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
29949 for the third operand.
29951 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29953 PR middle-end/79236
29954 * omp-low.c (struct omp_context): Add simt_stmt field.
29955 (scan_omp_for): Return omp_context *.
29956 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
29957 context to the _simt_ SIMD stmt.
29958 (lower_omp_for): For combined SIMD with sibling _simt_
29959 SIMD, make sure to use the same decls in _looptemp_
29960 clauses as in the sibling.
29962 2017-01-26 David Sherwood <david.sherwood@arm.com>
29964 PR middle-end/79212
29965 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
29968 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29971 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
29972 emit fld b; fld a; if possible.
29974 * brig-builtins.def: Update copyright years.
29975 * config/arm/arm_acle_builtins.def: Update copyright years.
29977 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
29980 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
29981 constraint instead of o for the stxsd instruction.
29983 2017-01-25 Carl Love <cel@us.ibm.com>
29985 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
29986 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
29988 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
29990 * doc/invoke.texi (C++ Dialect Options): Fix typo.
29992 2017-01-25 Richard Biener <rguenther@suse.de>
29994 PR tree-optimization/69264
29995 * target.def (vector_alignment_reachable): Improve documentation.
29996 * doc/tm.texi: Regenerate.
29997 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
29999 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
30000 earlier changes with respect to TYPE_USER_ALIGN.
30001 (vector_alignment_reachable_p): Likewise. Improve dumping.
30003 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30006 * config/arm/arm.md (xordi3): Force constant operand into a register
30009 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30011 * doc/invoke.texi (-fstore-merging): Correct default optimization
30012 levels at which it is enabled.
30013 (-O): Move -fstore-merging from list to...
30016 2017-01-25 Richard Biener <rguenther@suse.de>
30019 * omp-expand.c: Include debug.h.
30020 (expand_omp_taskreg): Make sure to generate early debug before
30021 outlining anything from a function.
30022 (expand_omp_target): Likewise.
30023 (grid_expand_target_grid_body): Likewise.
30025 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
30028 * asan.c (get_translation_unit_decl): New function.
30029 (asan_add_global): Extract modules file name from globals
30030 TRANSLATION_UNIT_DECL name.
30032 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
30035 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
30036 for long calls with APCS frame and VFP.
30038 2017-01-24 David Malcolm <dmalcolm@redhat.com>
30040 * cfg.c (original_copy_tables_initialized_p): New function.
30041 * cfg.h (original_copy_tables_initialized_p): New decl.
30042 * cfgrtl.c (relink_block_chain): Guard the call to
30043 free_original_copy_tables with a call to
30044 original_copy_tables_initialized_p.
30045 * cgraph.h (symtab_node::native_rtl_p): New decl.
30046 * cgraphunit.c (symtab_node::native_rtl_p): New function.
30047 (symtab_node::needed_p): Don't assert for early assembly output
30048 for __RTL functions.
30049 (cgraph_node::finalize_function): Set "force_output" for __RTL
30051 (cgraph_node::analyze): Bail out early for __RTL functions.
30052 (analyze_functions): Update assertion to support __RTL functions.
30053 (cgraph_node::expand): Bail out early for __RTL functions.
30054 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
30056 * function.h (struct function): Update comment for field
30058 * gimple-expr.c: Include "tree-pass.h".
30059 (gimple_has_body_p): Return false for __RTL functions.
30060 * Makefile.in (OBJS): Add run-rtl-passes.o.
30061 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
30063 (gcc::pass_manager::get_clean_slate): New accessor.
30064 * passes.c: Include "insn-addr.h".
30065 (should_skip_pass_p): Add logging. Update logic for running
30066 "expand" to be compatible with both __GIMPLE and __RTL. Guard
30067 property-provider override so it is only done for gimple passes.
30069 (skip_pass): New function.
30070 (execute_one_pass): Call skip_pass when skipping passes.
30071 * read-md.c (md_reader::read_char): Support filtering
30072 the input to a subset of line numbers.
30073 (md_reader::md_reader): Initialize fields
30074 m_first_line and m_last_line.
30075 (md_reader::read_file_fragment): New function.
30076 * read-md.h (md_reader::read_file_fragment): New decl.
30077 (md_reader::m_first_line): New field.
30078 (md_reader::m_last_line): New field.
30079 * read-rtl-function.c (function_reader::create_function): Only
30080 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
30081 curr_properties. Set DECL_INITIAL to a dummy block.
30082 (read_rtl_function_body_from_file_range): New function.
30083 * read-rtl-function.h (read_rtl_function_body_from_file_range):
30085 * run-rtl-passes.c: New file.
30086 * run-rtl-passes.h: New file.
30088 2017-01-24 Jeff Law <law@redhat.com>
30090 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
30093 2017-01-24 Bin Cheng <bin.cheng@arm.com>
30095 PR tree-optimization/79159
30096 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
30097 (record_nonwrapping_iv): Improve boundary using above function if no
30098 value range information.
30100 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
30101 Martin Jambor <mjambor@suse.cz>
30103 * brig-builtins.def: New file.
30104 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
30105 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
30106 (DEF_HSAIL_SAT_BUILTIN): Likewise.
30107 (DEF_HSAIL_INTR_BUILTIN): Likewise.
30108 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
30109 * builtin-types.def (BT_INT8): New.
30110 (BT_INT16): Likewise.
30111 (BT_UINT8): Likewise.
30112 (BT_UINT16): Likewise.
30113 (BT_FN_ULONG): Likewise.
30114 (BT_FN_UINT_INT): Likewise.
30115 (BT_FN_UINT_ULONG): Likewise.
30116 (BT_FN_UINT_LONG): Likewise.
30117 (BT_FN_UINT_PTR): Likewise.
30118 (BT_FN_ULONG_PTR): Likewise.
30119 (BT_FN_INT8_FLOAT): Likewise.
30120 (BT_FN_INT16_FLOAT): Likewise.
30121 (BT_FN_UINT32_FLOAT): Likewise.
30122 (BT_FN_UINT16_FLOAT): Likewise.
30123 (BT_FN_UINT8_FLOAT): Likewise.
30124 (BT_FN_UINT64_FLOAT): Likewise.
30125 (BT_FN_UINT16_UINT32): Likewise.
30126 (BT_FN_UINT32_UINT16): Likewise.
30127 (BT_FN_UINT16_UINT16_UINT16): Likewise.
30128 (BT_FN_INT_PTR_INT): Likewise.
30129 (BT_FN_UINT_PTR_UINT): Likewise.
30130 (BT_FN_LONG_PTR_LONG): Likewise.
30131 (BT_FN_ULONG_PTR_ULONG): Likewise.
30132 (BT_FN_VOID_UINT64_UINT64): Likewise.
30133 (BT_FN_UINT8_UINT8_UINT8): Likewise.
30134 (BT_FN_INT8_INT8_INT8): Likewise.
30135 (BT_FN_INT16_INT16_INT16): Likewise.
30136 (BT_FN_INT_INT_INT): Likewise.
30137 (BT_FN_UINT_FLOAT_UINT): Likewise.
30138 (BT_FN_FLOAT_UINT_UINT): Likewise.
30139 (BT_FN_ULONG_UINT_UINT): Likewise.
30140 (BT_FN_ULONG_UINT_PTR): Likewise.
30141 (BT_FN_ULONG_ULONG_ULONG): Likewise.
30142 (BT_FN_UINT_UINT_UINT): Likewise.
30143 (BT_FN_VOID_UINT_PTR): Likewise.
30144 (BT_FN_UINT_UINT_PTR: Likewise.
30145 (BT_FN_UINT32_UINT64_PTR): Likewise.
30146 (BT_FN_INT_INT_UINT_UINT): Likewise.
30147 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
30148 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
30149 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
30150 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
30151 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
30152 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
30153 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
30154 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
30155 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
30156 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
30157 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
30158 * doc/frontends.texi: List BRIG FE.
30159 * doc/install.texi (Testing): Add BRIG tesring requirements.
30160 * doc/invoke.texi (Overall Options): Mention BRIG.
30161 * doc/standards.texi (Standards): Doucment BRIG HSA version.
30163 2017-01-24 Richard Biener <rguenther@suse.de>
30165 PR translation/79208
30166 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
30168 2017-01-24 Martin Jambor <mjambor@suse.cz>
30171 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
30172 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
30173 and known_contexts.
30175 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
30177 PR middle-end/79123
30178 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
30179 casts from signed to unsigned really don't have a range.
30181 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
30183 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
30184 GMP_RNDx for compatiblity.
30186 2017-01-24 Martin Liska <mliska@suse.cz>
30189 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
30190 that would prevent us to call alloca with -1 as argument.
30192 2017-01-24 Jakub Jelinek <jakub@redhat.com>
30194 * dwarf2out.c (output_compilation_unit_header, output_file_names):
30195 Avoid -Wformat-security warning.
30197 2017-01-23 Andrew Pinski <apinski@cavium.com>
30199 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
30202 2017-01-23 Martin Sebor <msebor@redhat.com>
30204 PR middle-end/78703
30205 * gimple-ssa-sprintf.c (warn_level): New global.
30206 (format_integer): Use it here and throughout the rest of the file.
30207 Use the same switch to compute sign as base.
30208 (maybe_warn): New function.
30209 (format_directive): Factor out warnings into maybe_warn.
30210 Add debugging output. Use warn_level.
30211 (add_bytes): Use warn_level.
30212 (pass_sprintf_length::compute_format_length): Add debugging output.
30213 (try_substitute_return_value): Same.
30214 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
30216 PR middle-end/78703
30217 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
30218 (struct fmtresult, format_integer, format_floating): Adjust.
30219 (fmtresult::fmtresult): Set max correctly in two argument ctor.
30220 (get_string_length, format_string,format_directive): Same.
30221 (pass_sprintf_length::compute_format_length): Same.
30222 (try_substitute_return_value): Simplify slightly.
30224 PR middle-end/78703
30225 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
30226 (fmtresult::operator+=): Outlined.
30227 (struct fmtresult): Add ctors.
30228 (struct conversion_spec): Rename...
30229 (struct directive): ...to this. Add and remove data members.
30230 (directive::set_width, directive::set_precision): New functions.
30231 (format_percent): Use fmtresult ctor.
30232 (get_width_and_precision): Remove.
30233 (format_integer): Make naming changes. Avoid computing width and
30235 (format_floating): Same. Adjust indentation.
30236 (format_character, format_none): New functions.
30237 (format_string): Moved character handling to format_character.
30238 (format_directive): Remove arguments, change return type.
30239 (parse_directive): New function.
30240 (pass_sprintf_length::compute_format_length): Move directive
30241 parsing to parse_directive.
30243 2017-01-23 Jakub Jelinek <jakub@redhat.com>
30245 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
30246 (assign_assembler_name_if_needed): ... this.
30247 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
30248 (assign_assembler_name_if_needed): ... this.
30249 (free_lang_data_in_cgraph): Adjust callers.
30250 * cgraphunit.c (cgraph_node::analyze): Likewise.
30251 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
30254 2017-01-23 Richard Biener <rguenther@suse.de>
30256 PR tree-optimization/79088
30257 PR tree-optimization/79188
30258 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
30259 resetting loop bounds after last path deletion. Reset loop
30260 bounds of the target loop, make code match the comments.
30261 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
30262 Make sure loops need no fixups.
30264 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
30266 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
30267 exponent support with double type for first argument.
30268 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
30269 type returned by __builtin_vec_extract_sig,
30270 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
30271 functions from "vector int" to "vector unsigned int" or from
30272 "vector long long int" to "vector unsigned long long int".
30273 Changed type returned by __builtin_vec_extract_exp,
30274 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
30275 functions from "vector int" to "vector unsigned int" or from
30276 "vector long long int" to "vector unsigned long long int".
30277 Changed return type of __builtin_vec_test_data_class,
30278 __builtin_vec_test_data_class_sp, and
30279 __builtin_vec_test_data_class_dp from "vector int" to
30280 "vector bool int" or from "vector long long int" to "vector bool
30281 long long int" and changed second argument type from "unsigned
30282 int" to "int". Added new overloaded function forms "vector float
30283 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
30284 "vector float __builtin_vec_insert_exp_sp (vector float, vector
30285 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
30286 double, vector unsigned long long int)" and "vector double
30287 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
30288 long int)". Changed return type of
30289 __builtin_scalar_test_data_class and
30290 __builtin_scalar_test_data_class_sp and
30291 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
30292 int" and changed second argument from "unsigned int" to "int".
30293 Changed type returned by __builtin_scalar_test_neg,
30294 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
30295 from "int" to "bool int". Added new overloaded function form
30296 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
30297 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
30298 exponent double-precision with floating point first argument.
30299 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
30300 documentation of scalar_test_data_class, scalar_test_neg,
30301 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
30302 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
30303 vec_test_data_class built-in functions to reflect refinements in
30304 their type signatures.
30306 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
30308 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
30310 (aarch64_elf_asm_destructor): Likewise.
30312 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
30314 PR rtl-optimization/78634
30315 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
30316 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
30317 * ifcvt.c (noce_try_cmove): Add missing cost check.
30319 PR rtl-optimization/71724
30320 * combine.c (if_then_else_cond): Look for situations where it is
30321 beneficial to undo the work of one of the recursive calls.
30323 2017-01-23 Bin Cheng <bin.cheng@arm.com>
30325 PR tree-optimization/70754
30326 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
30327 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
30328 combined stmt before it if not NULL.
30329 (combine_chains): Process refs reversely and compute dominance point
30332 2017-01-23 Martin Liska <mliska@suse.cz>
30334 PR tree-optimization/79196
30335 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
30336 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
30338 (strlen_optimize_stmt): Call the renamed function.
30340 2017-01-23 Michael Matz <matz@suse.de>
30342 PR tree-optimization/78384
30343 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
30345 2017-01-23 Richard Biener <rguenther@suse.de>
30347 PR tree-optimization/79186
30348 * tree-vrp.c (register_new_assert_for): Make sure we've seen
30349 both incoming edges before moving an assert.
30351 2017-01-23 Martin Jambor <mjambor@suse.cz>
30353 * ipa-prop.c (load_from_param_1): Removed.
30354 (load_from_unmodified_param): Bits from load_from_param_1 put back
30356 (load_from_param): Removed.
30357 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
30358 with stmt. Reverted back to use of load_from_unmodified_param.
30360 2017-01-23 Martin Jambor <mjambor@suse.cz>
30363 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
30364 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
30365 field a pointer to garbage collected vector, mark lattices and
30366 ipcp_orig_node with GTY((skip)).
30367 (ipa_get_param_count): Adjust to descriptors being a pointer.
30368 (ipa_get_param): Likewise.
30369 (ipa_get_type): Likewise.
30370 (ipa_get_param_move_cost): Likewise.
30371 (ipa_set_param_used): Likewise.
30372 (ipa_get_controlled_uses): Likewise.
30373 (ipa_set_controlled_uses): Likewise.
30374 (ipa_is_param_used): Likewise.
30375 (ipa_node_params_t): Move into garbage collector. New methods insert
30377 (ipa_node_params_sum): Annotate wth GTY(()).
30378 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
30380 (ipa_load_from_parm_agg): Adjust declaration.
30381 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
30382 * ipa-profile.c (ipa_profile): Likewise.
30383 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
30384 (ipa_populate_param_decls): Make descriptors parameter garbage
30386 (ipa_dump_param): Adjust to descriptors being a pointer.
30387 (ipa_alloc_node_params): Likewise.
30388 (ipa_initialize_node_params): Likewise.
30389 (load_from_param_1): Make descriptors parameter garbage collected.
30390 (load_from_unmodified_param): Likewise.
30391 (load_from_param): Likewise.
30392 (ipa_load_from_parm_agg): Likewise.
30393 (ipa_node_params::~ipa_node_params): Removed.
30394 (ipa_free_all_node_params): Remove call to delete operator.
30395 (ipa_node_params_t::insert): New.
30396 (ipa_node_params_t::remove): Likewise.
30397 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
30398 copy known_csts and known_contexts vectors.
30399 (ipa_read_node_info): Adjust to descriptors being a pointer.
30400 (ipcp_modif_dom_walker): Make m_descriptors field garbage
30402 (ipcp_transform_function): Make descriptors variable garbage
30405 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
30407 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
30408 * config/i386/avx512dqintrin.h: Ditto.
30409 * config/i386/avx512fintrin.h: Ditto.
30410 * config/i386/i386.c: Handle new builtins.
30411 * config/i386/i386-builtin.def: Add new builtins.
30412 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
30413 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
30415 2017-01-23 Jakub Jelinek <jakub@redhat.com>
30416 Martin Liska <mliska@suse.cz>
30418 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
30419 * asan.c (asan_expand_poison_ifn): Support stores and use
30420 appropriate ASAN report function.
30421 * internal-fn.c (expand_ASAN_POISON_USE): New function.
30422 * internal-fn.def (ASAN_POISON_USE): Declare.
30423 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
30424 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
30425 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
30426 ASAN_POISON calls w/o LHS.
30427 * tree-ssa.c (execute_update_addresses_taken): Create clobber
30428 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
30429 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
30430 * gimplify.c (asan_poison_variables): Add attribute
30431 use_after_scope_memory to variables that really needs to live
30433 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
30434 having the attribute.
30436 2017-01-23 Martin Liska <mliska@suse.cz>
30438 * asan.c (create_asan_shadow_var): New function.
30439 (asan_expand_poison_ifn): Likewise.
30440 * asan.h (asan_expand_poison_ifn): New declaration.
30441 * internal-fn.c (expand_ASAN_POISON): Likewise.
30442 * internal-fn.def (ASAN_POISON): New builtin.
30443 * sanopt.c (pass_sanopt::execute): Expand
30444 asan_expand_poison_ifn.
30445 * tree-inline.c (copy_decl_for_dup_finish): Make function
30447 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
30448 * tree-ssa.c (is_asan_mark_p): New function.
30449 (execute_update_addresses_taken): Rewrite local variables
30450 (identified just by use-after-scope as addressable) into SSA.
30452 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
30454 * doc/install.texi (Specific): opensource.apple.com uses https
30455 now. Remove trailing slash.
30457 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
30459 * README.Portability: Remove note on an Irix compatibility issue.
30461 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
30463 * gcov.c (INCLUDE_ALGORITHM): Define.
30464 (INCLUDE_VECTOR): Define.
30465 No longer include <vector> and <algorithm> directly.
30467 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
30469 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
30471 * doc/invoke.texi (Code Gen Options): Ditto.
30473 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
30476 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
30478 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
30480 rtl-optimization/79125
30481 * cprop.c (local_cprop_pass): Handle cases where we make an
30482 unconditional trap.
30484 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
30488 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
30489 read from, for big endian.
30491 2017-01-20 Jiong Wang <jiong.wang@arm.com>
30493 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
30494 register pauth builtins for LP64 only.
30496 2017-01-20 Marek Polacek <polacek@redhat.com>
30499 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
30502 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
30504 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
30506 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
30507 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
30508 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
30510 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30513 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
30514 in big-endian mode when they are not a single duplicated value.
30516 2017-01-20 Richard Biener <rguenther@suse.de>
30518 * BASE-VER: Bump to 7.0.1.
30520 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
30522 * omp-low.c (omplow_simd_context): New struct. Use it...
30523 (lower_rec_simd_input_clauses): ...here and...
30524 (lower_rec_input_clauses): ...here to hold common data. Adjust all
30525 references to idx, lane, max_vf, is_simt.
30527 2017-01-20 Graham Markall <graham.markall@embecosm.com>
30529 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
30532 2017-01-20 Martin Jambor <mjambor@suse.cz>
30534 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
30535 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
30537 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
30538 (GTFILES): Rename hsa.c to hsa-common.c.
30539 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
30540 * hsa-dump.c: Likewise.
30541 * hsa-gen.c: Likewise.
30542 * hsa-regalloc.c: Likewise.
30543 * ipa-hsa.c: Likewise.
30544 * omp-expand.c: Likewise.
30545 * omp-low.c: Likewise.
30546 * toplev.c: Likewise.
30548 2017-01-20 Marek Polacek <polacek@redhat.com>
30551 * doc/invoke.texi: Document -Wduplicated-branches.
30552 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
30553 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
30554 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
30555 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
30556 return 0 only when not OEP_LEXICOGRAPHIC.
30557 (fold_build_cleanup_point_expr): Use the expression
30558 location when building CLEANUP_POINT_EXPR.
30559 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
30560 * tree.c (add_expr): Handle error_mark_node.
30562 2017-01-20 Martin Liska <mliska@suse.cz>
30565 * tree-profile.c (init_ic_make_global_vars): Do not call
30567 (gimple_init_gcov_profiler): Likewise.
30569 2017-01-20 Martin Liska <mliska@suse.cz>
30572 * cgraph.h (maybe_create_reference): Remove argument and
30574 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
30576 * ipa-cp.c (create_specialized_node): Likewise.
30577 * symtab.c (symtab_node::maybe_create_reference): Handle
30578 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
30580 2017-01-20 Martin Liska <mliska@suse.cz>
30582 * read-rtl-function.c (function_reader::create_function): Use
30583 build_decl instread of build_decl_stat.
30585 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
30587 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
30588 * config/i386/avx512dqintrin.h: Ditto.
30589 * config/i386/avx512fintrin.h: Ditto.
30590 * config/i386/i386-builtin-types.def: Add new types.
30591 * config/i386/i386.c: Handle new types.
30592 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
30593 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
30594 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
30595 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
30596 (__builtin_ia32_kshiftridi): New.
30597 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
30599 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
30603 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
30604 define to rs6000_init_stack_protect_guard.
30605 (rs6000_init_stack_protect_guard): New function.
30607 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30608 Yunqiang Su <yunqiang.su@imgtec.com>
30610 * config.gcc (supported_defaults): Add madd4.
30611 (with_madd4): Add validation.
30612 (all_defaults): Add madd4.
30613 * config/mips/mips.opt (mmadd4): New option.
30614 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
30616 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
30618 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
30619 (ISA_HAS_FUSED_MADD4): Likewise.
30620 * doc/invoke.texi (-mmadd4): Document the new option.
30621 * doc/install.texi (--with-madd4): Document the new option.
30623 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30625 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
30626 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
30627 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
30628 (aarch64_init_pauth_hint_builtins): New.
30629 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
30630 (aarch64_expand_builtin): Expand new builtins.
30632 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30634 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
30635 * combine-stack-adj.c (no_unhandled_cfa): Handle
30636 REG_CFA_TOGGLE_RA_MANGLE.
30637 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
30638 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
30639 info for return address signing.
30640 (aarch64_expand_epilogue): Likewise.
30642 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30644 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
30645 * config/aarch64/aarch64-protos.h
30646 (aarch64_return_address_signing_enabled): New declaration.
30647 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
30649 (aarch64_expand_prologue): Sign return address before it's pushed onto
30651 (aarch64_expand_epilogue): Authenticate return address fetched from
30653 (aarch64_override_options): Sanity check for ILP32 and ISA level.
30654 (aarch64_attributes): New function attributes for "sign-return-address".
30655 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
30656 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
30657 ("*do_return"): Generate combined instructions according to key index.
30658 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
30659 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
30661 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
30662 * config/aarch64/aarch64.opt (msign-return-address=): New.
30663 * doc/extend.texi (AArch64 Function Attributes): Documents
30664 "sign-return-address=".
30665 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
30667 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30669 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
30670 overall option summary.
30672 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30674 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
30675 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
30676 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
30677 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
30679 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
30681 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
30682 -mpower9-minmax by default for -mcpu=power9.
30683 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
30684 128-bit floating point.
30686 2017-01-20 Alan Modra <amodra@gmail.com>
30688 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
30689 optimizing for size.
30691 2017-01-20 Alan Modra <amodra@gmail.com>
30694 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
30695 for strcmp and strncmp from corresponding builtin decl.
30697 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
30699 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
30700 instead of i386/rtems-64.h.
30701 * config/i386/rtems-64.h: Remove.
30703 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
30707 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
30709 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
30711 2017-01-19 Tamar Christina <tamar.christina@arm.com>
30713 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
30714 Change int to HOST_WIDE_INT.
30715 * config/aarch64/aarch64-protos.h
30716 (aarch64_simd_gen_const_vector_dup): Likewise.
30717 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
30719 2017-01-19 David Malcolm <dmalcolm@redhat.com>
30721 * langhooks-def.h (lhd_type_for_size): New decl.
30722 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
30723 * langhooks.c (lhd_type_for_size): New function, taken from
30726 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
30728 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
30729 define_bypass for CR latency.
30730 (power9-cracked-alu): Update bypass latency and remove power9-branch.
30731 (power9-alu2): Add define_bypass for CR latency.
30733 (power9-mul): Update insn latency.
30734 (power9-mul-compare): Update insn latency, bypass latency and remove
30737 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30739 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
30741 * config/aarch64/aarch64.md
30742 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
30743 aarch64_nopcrelative_literal_loads.
30744 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
30746 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
30748 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
30749 TARGET_LOONGSON_3A.
30750 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
30752 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
30755 * config.gcc (supported_defaults): Add lxc1-sxc1.
30756 (with_lxc1_sxc1): Add validation.
30757 (all_defaults): Add lxc1-sxc1.
30758 * config/mips/mips.opt (mlxc1-sxc1): New option.
30759 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
30761 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
30762 __mips_no_lxc1_sxc1.
30763 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
30764 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
30765 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
30767 2017-01-19 Richard Biener <rguenther@suse.de>
30769 PR tree-optimization/72488
30770 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
30771 sure to restore SSA info.
30772 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
30774 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
30776 PR rtl-optimization/79121
30777 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
30778 of the inner type when shifting an extended value.
30780 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
30783 * symtab.c (symtab_node::equal_address_to): Fix comparing of
30784 interposable aliases.
30786 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
30789 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
30790 Use the evmergelohi instruction.
30791 (mov_si<mode>_e500_subreg4_2_le): Likewise.
30792 (mov_sitf_e500_subreg8_2_be): Likewise.
30793 (mov_sitf_e500_subreg12_2_le): Likewise.
30794 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
30795 (mov_si<mode>_e500_subreg4_2_be): Likewise.
30796 (mov_sitf_e500_subreg8_2_le): Likewise.
30797 (mov_sitf_e500_subreg12_2_be): Likewise.
30799 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30801 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
30802 attribute from vecsimple to vecperm.
30803 (altivec_vbpermq2): Likewise.
30805 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30808 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
30810 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30811 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
30812 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
30813 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
30814 case where N arg is SIZE_MAX.
30815 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
30816 (cmpstrsi): Add pattern.
30818 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30820 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30821 __builtin_vec_revb builtins.
30822 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
30823 built-in functions to support generation of the ISA 3.0 XXBR<x>
30824 vector byte reverse instructions.
30825 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
30826 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
30827 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
30828 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
30829 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
30830 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
30831 (P9V_BUILTIN_VEC_REVB): Likewise.
30832 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
30833 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
30834 (p9_xxbrq_v16qi): Likewise.
30835 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
30836 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
30837 (p9_xxbrh_v8hi): Likewise.
30838 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
30839 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
30840 vec_revb built-in functions.
30842 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
30844 PR rtl-optimization/78952
30845 * config/i386/i386.md (any_extract): New code iterator.
30846 (*insvqi_2): Use any_extract for source operand.
30847 (*insvqi_3): Use any_shiftrt for source operand.
30849 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
30851 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
30853 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
30855 2017-01-18 Matthias Klose <doko@ubuntu.com>
30857 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
30859 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30861 * config/rs6000/altivec.h (vec_bperm): Change #define.
30862 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
30863 (altivec_vbpermq2): New define_insn.
30864 (altivec_vbpermd): Likewise.
30865 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
30866 function interface.
30867 (VBPERMD): Likewise.
30868 (VBPERM): New polymorphic function interface.
30869 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
30870 Add entries for P9V_BUILTIN_VEC_VBPERM.
30871 * doc/extend.texi: Add interfaces for vec_bperm.
30873 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30875 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
30876 first letter of error messages.
30877 (s390_resolve_overloaded_builtin): Likewise.
30878 * config/s390/s390.c (s390_expand_builtin): Likewise.
30879 (s390_invalid_arg_for_unprototyped_fn): Likewise.
30880 (s390_valid_target_attribute_inner_p): Likewise.
30881 * config/s390/s390.md ("tabort"): Likewise.
30883 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
30885 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
30886 (ISA_AVOID_DIV_HILO): New macro.
30887 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
30888 (ISA_HAS_DDIV): Likewise.
30890 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30892 * doc/invoke.texi (fabi-version): Correct number of occurrences.
30894 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30896 * doc/invoke.texi (fabi-version): Spelling fix.
30898 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30901 * doc/invoke.texi (fabi-version): Mention mangling fix for
30904 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30907 * doc/invoke.texi (fabi-version): Document discriminator mangling.
30909 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
30912 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
30913 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
30915 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
30917 (stack_protect_test): Ditto.
30918 * config/rs6000/rs6000.opt (mstack-protector-guard=,
30919 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
30921 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
30922 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
30923 -mstack-protector-guard-offset=.
30924 (RS/6000 and PowerPC Options): Ditto.
30926 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30928 * config/i386/i386.h (MASK_CLASS_P): New define.
30929 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
30930 there are no registers from different register sets also when
30931 mask registers are used. Update function comment.
30932 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
30933 to (*k/*r) and (*k/*km) alternatives.
30935 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
30937 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
30938 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
30939 (EH_RETURN_HANDLER_RTX): New define.
30940 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
30941 Force frame pointer in EH return functions.
30942 (aarch64_expand_epilogue): Add barrier for eh_return.
30943 (aarch64_final_eh_return_addr): Remove.
30944 (aarch64_eh_return_handler_rtx): New function.
30945 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
30947 (aarch64_eh_return_handler_rtx): New prototype.
30949 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30951 * config/rs6000/altivec.h (vec_rlmi): New #define.
30952 (vec_vrlnm): Likewise.
30953 (vec_rlnm): Likewise.
30954 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
30955 (UNSPEC_VRLNM): Likewise.
30956 (VIlong): New mode iterator.
30957 (altivec_vrl<VI_char>mi): New define_insn.
30958 (altivec_vrl<VI_char>nm): Likewise.
30959 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
30961 (VRLDNM): Likewise.
30962 (RLNM): New polymorphic function entry.
30963 (VRLWMI): New monomorphic function entry.
30964 (VRLDMI): Likewise.
30965 (RLMI): New polymorphic function entry.
30966 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
30967 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
30968 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
30971 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30974 * dwarf2out.c (field_byte_offset): Restore the
30975 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
30976 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
30977 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
30980 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30983 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
30985 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30988 * dwarf2out.c (add_data_member_location_attribute): For constant
30989 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
30990 instead of DW_AT_data_member_location, DW_AT_bit_offset and
30991 DW_AT_byte_size attributes.
30993 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30995 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
30996 after forcing to constant memory when the code model is medium.
30998 2017-01-17 Julia Koval <julia.koval@intel.com>
31001 * config/i386/avx512fintrin.h
31002 (_mm512_i32gather_ps): Change __addr type to void const*.
31003 (_mm512_mask_i32gather_ps): Ditto.
31004 (_mm512_i32gather_pd): Ditto.
31005 (_mm512_mask_i32gather_pd): Ditto.
31006 (_mm512_i64gather_ps): Ditto.
31007 (_mm512_mask_i64gather_ps): Ditto.
31008 (_mm512_i64gather_pd): Ditto.
31009 (_mm512_mask_i64gather_pd): Ditto.
31010 (_mm512_i32gather_epi32): Ditto.
31011 (_mm512_mask_i32gather_epi32): Ditto.
31012 (_mm512_i32gather_epi64): Ditto.
31013 (_mm512_mask_i32gather_epi64): Ditto.
31014 (_mm512_i64gather_epi32): Ditto.
31015 (_mm512_mask_i64gather_epi32): Ditto.
31016 (_mm512_i64gather_epi64): Ditto.
31017 (_mm512_mask_i64gather_epi64): Ditto.
31018 (_mm512_i32scatter_ps): Change __addr type to void*.
31019 (_mm512_mask_i32scatter_ps): Ditto.
31020 (_mm512_i32scatter_pd): Ditto.
31021 (_mm512_mask_i32scatter_pd): Ditto.
31022 (_mm512_i64scatter_ps): Ditto.
31023 (_mm512_mask_i64scatter_ps): Ditto.
31024 (_mm512_i64scatter_pd): Ditto.
31025 (_mm512_mask_i64scatter_pd): Ditto.
31026 (_mm512_i32scatter_epi32): Ditto.
31027 (_mm512_mask_i32scatter_epi32): Ditto.
31028 (_mm512_i32scatter_epi64): Ditto.
31029 (_mm512_mask_i32scatter_epi64): Ditto.
31030 (_mm512_i64scatter_epi32): Ditto.
31031 (_mm512_mask_i64scatter_epi32): Ditto.
31032 (_mm512_i64scatter_epi64): Ditto.
31033 (_mm512_mask_i64scatter_epi64): Ditto.
31034 * config/i386/avx512pfintrin.h
31035 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
31036 (_mm512_mask_prefetch_i32gather_ps): Ditto.
31037 (_mm512_mask_prefetch_i64gather_pd): Ditto.
31038 (_mm512_mask_prefetch_i64gather_ps): Ditto.
31039 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
31040 (_mm512_prefetch_i32scatter_ps): Ditto.
31041 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
31042 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
31043 (_mm512_prefetch_i64scatter_pd): Ditto.
31044 (_mm512_prefetch_i64scatter_ps): Ditto.
31045 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
31046 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
31047 * config/i386/avx512vlintrin.h
31048 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
31049 (_mm_mmask_i32gather_ps): Ditto.
31050 (_mm256_mmask_i32gather_pd): Ditto.
31051 (_mm_mmask_i32gather_pd): Ditto.
31052 (_mm256_mmask_i64gather_ps): Ditto.
31053 (_mm_mmask_i64gather_ps): Ditto.
31054 (_mm256_mmask_i64gather_pd): Ditto.
31055 (_mm_mmask_i64gather_pd): Ditto.
31056 (_mm256_mmask_i32gather_epi32): Ditto.
31057 (_mm_mmask_i32gather_epi32): Ditto.
31058 (_mm256_mmask_i32gather_epi64): Ditto.
31059 (_mm_mmask_i32gather_epi64): Ditto.
31060 (_mm256_mmask_i64gather_epi32): Ditto.
31061 (_mm_mmask_i64gather_epi32): Ditto.
31062 (_mm256_mmask_i64gather_epi64): Ditto.
31063 (_mm_mmask_i64gather_epi64): Ditto.
31064 (_mm256_i32scatter_ps): Change __addr type to void*.
31065 (_mm256_mask_i32scatter_ps): Ditto.
31066 (_mm_i32scatter_ps): Ditto.
31067 (_mm_mask_i32scatter_ps): Ditto.
31068 (_mm256_i32scatter_pd): Ditto.
31069 (_mm256_mask_i32scatter_pd): Ditto.
31070 (_mm_i32scatter_pd): Ditto.
31071 (_mm_mask_i32scatter_pd): Ditto.
31072 (_mm256_i64scatter_ps): Ditto.
31073 (_mm256_mask_i64scatter_ps): Ditto.
31074 (_mm_i64scatter_ps): Ditto.
31075 (_mm_mask_i64scatter_ps): Ditto.
31076 (_mm256_i64scatter_pd): Ditto.
31077 (_mm256_mask_i64scatter_pd): Ditto.
31078 (_mm_i64scatter_pd): Ditto.
31079 (_mm_mask_i64scatter_pd): Ditto.
31080 (_mm256_i32scatter_epi32): Ditto.
31081 (_mm256_mask_i32scatter_epi32): Ditto.
31082 (_mm_i32scatter_epi32): Ditto.
31083 (_mm_mask_i32scatter_epi32): Ditto.
31084 (_mm256_i32scatter_epi64): Ditto.
31085 (_mm256_mask_i32scatter_epi64): Ditto.
31086 (_mm_i32scatter_epi64): Ditto.
31087 (_mm_mask_i32scatter_epi64): Ditto.
31088 (_mm256_i64scatter_epi32): Ditto.
31089 (_mm256_mask_i64scatter_epi32): Ditto.
31090 (_mm_i64scatter_epi32): Ditto.
31091 (_mm_mask_i64scatter_epi32): Ditto.
31092 (_mm256_i64scatter_epi64): Ditto.
31093 (_mm256_mask_i64scatter_epi64): Ditto.
31094 (_mm_i64scatter_epi64): Ditto.
31095 (_mm_mask_i64scatter_epi64): Ditto.
31096 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
31097 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
31098 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
31099 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
31100 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
31101 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
31102 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
31103 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
31104 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
31105 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
31106 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
31107 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
31108 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
31109 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
31110 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
31111 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
31112 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
31113 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
31114 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
31115 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
31116 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
31117 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
31118 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
31119 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
31120 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
31121 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
31122 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
31123 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
31124 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
31125 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
31126 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
31127 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
31128 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
31129 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
31130 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
31131 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
31132 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
31133 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
31134 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
31135 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
31136 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
31137 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
31138 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
31139 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
31140 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
31141 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
31142 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
31143 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
31144 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
31145 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
31146 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
31147 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
31148 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
31149 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
31150 definitions accordingly.
31152 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
31153 Kuan-Lin Chen <kuanlinchentw@gmail.com>
31156 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
31159 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
31162 * ira-conflicts.c (ira_build_conflicts): Update total conflict
31163 hard regs for inner regno.
31165 2017-01-17 Martin Liska <mliska@suse.cz>
31168 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
31169 assumption and add comment.
31171 2017-01-17 Nathan Sidwell <nathan@acm.org>
31173 * ipa-visibility.c (localize_node): New function, broken out of ...
31174 (function_and_variable_visibility): ... here. Call it.
31176 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31178 PR middle-end/77445
31179 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
31180 correctly set frequency of oudgoing edge.
31181 (duplicate_thread_path): Fix profile updating.
31183 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31186 * configure.ac: Add GCC_BASE_VER.
31187 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
31188 version from BASE-VER file.
31189 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
31190 (gcc.o): Depend on $(BASEVER).
31191 * common.opt (dumpfullversion): New option.
31192 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
31193 * doc/invoke.texi: Document -dumpfullversion.
31194 * doc/install.texi: Document --with-gcc-major-version-only.
31195 * configure: Regenerated.
31197 2017-01-17 Richard Biener <rguenther@suse.de>
31199 PR tree-optimization/71433
31200 * tree-vrp.c (register_new_assert_for): Merge same asserts
31201 on all incoming edges.
31202 (process_assert_insertions_for): Handle insertions at the
31205 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
31207 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
31208 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
31210 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
31213 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
31216 2017-01-17 Alan Modra <amodra@gmail.com>
31219 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
31220 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
31221 symbolic stack limit when pic.
31223 2017-01-16 Martin Sebor <msebor@redhat.com>
31225 PR tree-optimization/78608
31226 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
31228 2017-01-16 Jeff Law <law@redhat.com>
31231 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
31232 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
31233 for several include directories that may be relative to sysroot.
31234 * config/i386/x-mingw32 (gplus_includedir): Define.
31235 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
31236 (native_system_includedir): Likewise.
31237 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
31238 override if TARGET_SYSTEM_ROOT is defined.
31239 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
31241 PR tree-optimization/79090
31242 PR tree-optimization/33562
31243 PR tree-optimization/61912
31244 PR tree-optimization/77485
31245 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
31246 and computed trims into the dump file.
31248 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
31250 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
31252 2017-01-16 Jakub Jelinek <jakub@redhat.com>
31255 * gimplify.c (gimplify_init_constructor): If want_value and
31256 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
31260 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
31261 sequence. Formatting fixes.
31262 (doloop_optimize): Formatting fixes.
31265 * gcc.c (debug_level_greater_than_spec_func): New function.
31266 (static_spec_functions): Add debug-level-gt spec function.
31267 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
31269 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
31270 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
31271 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
31272 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
31273 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
31274 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
31276 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
31278 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
31279 QImode fixups to general and mask registers only.
31281 2017-01-16 Carl Love <cel@us.ibm.com>
31283 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
31284 for built-in functions
31285 vector signed char vec_nabs (vector signed char)
31286 vector signed short vec_nabs (vector signed short)
31287 vector signed int vec_nabs (vector signed int)
31288 vector signed long long vec_nabs (vector signed long long)
31289 vector float vec_nabs (vector float)
31290 vector double vec_nabs (vector double)
31291 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
31293 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
31294 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
31295 * doc/extend.texi: Update the documentation file for the new built-in
31298 2017-01-16 Martin Sebor <msebor@redhat.com>
31300 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
31303 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31305 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
31306 UNSPEC_VSX__XXSPLTD to require special splat handling.
31308 2017-01-16 David Malcolm <dmalcolm@redhat.com>
31311 * system.h: Poison strndup.
31313 2017-01-16 Alan Modra <amodra@gmail.com>
31316 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
31319 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
31321 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
31323 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
31325 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
31326 call recog here. Assert that INSN_CODE (insn) is non-negative.
31328 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
31331 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
31333 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
31334 in the currently scheduled RTL fragment.
31336 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
31338 PR rtl-optimization/78751
31339 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
31342 2017-01-14 Jeff Law <law@redhat.com>
31344 PR tree-optimization/79090
31345 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
31346 variable length stores.
31347 (compute_trims): Delete dead assignment to *trim_tail.
31348 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
31351 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
31353 PR rtl-optimization/78626
31354 PR rtl-optimization/78727
31355 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
31356 of a block, and split such blocks after everything else is finished.
31358 2017-01-14 Alan Modra <amodra@gmail.com>
31361 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
31362 target legitimate_combined_insn.
31363 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
31364 (rs6000_legitimate_combined_insn): New function.
31365 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
31367 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
31368 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
31369 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
31371 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
31373 * doc/frontends.texi (G++ and GCC): Remove references to Java.
31375 2017-01-13 Jeff Law <law@redhat.com>
31377 PR tree-optimization/33562
31378 PR tree-optimization/61912
31379 PR tree-optimization/77485
31380 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
31382 (delete_dead_assignment): Likewise.
31383 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
31384 statement to delete_dead_call and delete_dead_assignment.
31386 2017-01-13 David Malcolm <dmalcolm@redhat.com>
31389 * substring-locations.c (format_warning_va): Strengthen case 1 so
31390 that both endpoints of the substring must be within the format
31391 range for just the substring to be printed.
31393 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
31395 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
31396 * config/i386/i386.c (ix86_target_string): Add missing options
31397 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
31398 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
31399 flags_other and ix86_target_other to flags2_other. Display unknown
31401 (ix86_valid_target_attribute_inner_p): Add missing options and
31402 reorder options by implied ISAs, as in ix86_target_string.
31404 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
31406 * hash-table.h (hash_table::too_empty_p): New function.
31407 (hash_table::expand): Use it.
31408 (hash_table::traverse): Likewise.
31409 (hash_table::empty_slot): Use sizeof (value_type) instead of
31410 sizeof (PTR) to convert bytes to elements. Shrink the table
31411 if the current size is excessive for the current number of
31414 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
31416 * ira-costs.c (record_reg_classes): Break from the inner loop
31417 early once alt_fail is known to be true. Update outer loop
31418 handling accordingly.
31420 2017-01-13 Jeff Law <law@redhat.com>
31422 * tree-ssa-dse.c (decrement_count): New function.
31423 (increment_start_addr, maybe_trim_memstar_call): Likewise.
31424 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
31425 when we know the partially dead statement is a mem* function.
31427 PR tree-optimization/61912
31428 PR tree-optimization/77485
31429 * tree-ssa-dse.c: Include expr.h.
31430 (maybe_trim_constructor_store): New function.
31431 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
31433 PR tree-optimization/33562
31434 PR tree-optimization/61912
31435 PR tree-optimization/77485
31436 * doc/invoke.texi: Document new dse-max-object-size param.
31437 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
31438 * tree-ssa-dse.c: Include params.h.
31439 (dse_store_status): New enum.
31440 (initialize_ao_ref_for_dse): New, partially extracted from
31442 (valid_ao_ref_for_dse, normalize_ref): New.
31443 (setup_live_bytes_from_ref, compute_trims): Likewise.
31444 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
31445 (maybe_trim_partially_dead_store): Likewise.
31446 (maybe_trim_complex_store): Likewise.
31447 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
31448 Track what bytes live from the original store. Return tri-state
31449 for dead, partially dead or live.
31450 (dse_dom_walker): Add constructor, destructor and new private members.
31451 (delete_dead_call, delete_dead_assignment): New extracted from
31453 (dse_optimize_stmt): Make a member of dse_dom_walker.
31454 Use initialize_ao_ref_for_dse.
31456 PR tree-optimization/33562
31457 PR tree-optimization/61912
31458 PR tree-optimization/77485
31459 * sbitmap.h (bitmap_count_bits): Prototype.
31460 (bitmap_clear_range, bitmap_set_range): Likewise.
31461 * sbitmap.c (bitmap_clear_range): New function.
31462 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
31464 2017-01-13 Martin Liska <mliska@suse.cz>
31467 * function.c (set_cfun): Add new argument force.
31468 * function.h (set_cfun): Likewise.
31469 * ipa-inline-transform.c (inline_call): Use the function when
31470 strict alising from is dropped for function we inline to.
31472 2017-01-13 Richard Biener <rguenther@suse.de>
31474 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
31475 for dumping GIMPLE INTEGER_CSTs.
31477 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31479 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
31480 to 201112L since C++17.
31482 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
31485 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
31486 if -fsanitize=kernel-address is present.
31488 2017-01-13 Richard Biener <rguenther@suse.de>
31490 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
31491 as _Literal ( type ) number in case usual suffixes do not
31492 preserve all information.
31494 2017-01-13 Richard Biener <rguenther@suse.de>
31496 PR tree-optimization/77283
31497 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
31498 and ssa-iterators.h.
31499 (is_feasible_trace): Implement a cost model based on joiner
31502 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
31505 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
31506 char or short to __float128/_Float128 directly.
31508 2017-01-12 Martin Sebor <msebor@redhat.com>
31510 to -Wformat-overflow.
31511 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
31512 (min_bytes_remaining): Same.
31513 (get_string_length): Same.
31514 (format_string): Same.
31515 (format_directive): Same.
31517 (pass_sprintf_length::handle_gimple_call): Same.
31519 2017-01-12 Jakub Jelinek <jakub@redhat.com>
31521 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
31522 info.nowrite calls with no lhs that can't throw. Return bool
31523 whether gsi_remove has been called or not.
31524 (pass_sprintf_length::handle_gimple_call): Return bool whether
31525 try_substitute_return_value called gsi_remove. Formatting fix.
31526 (pass_sprintf_length::execute): Don't use gsi_remove if
31527 handle_gimple_call returned true.
31530 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
31531 be removed due to side-effects, don't remove following barrier nor
31532 turn the successor edge into fallthru edge.
31534 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31537 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
31538 element-reversing loads and stores as not swappable.
31540 2017-01-12 Nathan Sidwell <nathan@acm.org>
31541 Nicolai Stange <nicstange@gmail.com>
31543 * combine.c (try_combine): Don't ignore result of overlap checking
31544 loop. Combine overlap & asm check into single loop.
31546 2017-01-12 Richard Biener <rguenther@suse.de>
31548 * tree-pretty-print.c (dump_generic_node): Provide -gimple
31549 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
31551 2017-01-12 Richard Biener <rguenther@suse.de>
31553 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
31554 and TS_TARGET_OPTION directly derive from TS_BASE.
31555 * tree-core.h (tree_optimization_option): Derive from tree_base.
31556 (tree_target_option): Likewise.
31558 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
31560 * config/i386/i386.c (memory_address_length): Increase len
31561 only when rip_relative_addr_p returns false.
31563 2017-01-11 Julia Koval <julia.koval@intel.com>
31565 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
31566 (OPTION_MASK_ISA_SGX_SET): New.
31567 (ix86_handle_option): Handle OPT_msgx.
31568 * config.gcc: Added sgxintrin.h.
31569 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
31570 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
31571 * config/i386/i386.c (ix86_target_string): Add -msgx.
31573 (ix86_option_override_internal): Handle new options.
31574 (ix86_valid_target_attribute_inner_p): Add sgx.
31575 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
31576 * config/i386/i386.opt: Add msgx.
31577 * config/i386/sgxintrin.h: New file.
31578 * config/i386/x86intrin.h: Add sgxintrin.h.
31580 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31583 * fold-const.c (maybe_nonzero_address): Return 1 for function
31585 (tree_single_nonzero_warnv_p): Don't handle function local objects
31589 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
31592 2017-01-11 David Malcolm <dmalcolm@redhat.com>
31595 * opts.c: Include "spellcheck.h"
31596 (struct string_fragment): New struct.
31597 (struct edit_distance_traits<const string_fragment &>): New
31599 (get_closest_sanitizer_option): New function.
31600 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
31602 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31604 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
31606 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
31607 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
31608 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
31609 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
31610 for initial die_offset if dwarf_split_debug_info.
31611 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
31612 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
31614 (output_skeleton_debug_sections): Formatting fix. Use
31615 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
31616 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
31618 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
31620 * config/arm/cortex-a53.md: Add bypasses for
31621 cortex_a53_r2f_cvt.
31622 (cortex_a53_r2f): Only use for transfers.
31623 (cortex_a53_f2r): Likewise.
31624 (cortex_a53_r2f_cvt): Add reservation for conversions.
31625 (cortex_a53_f2r_cvt): Likewise.
31627 2017-01-11 Tamar Christina <tamar.christina@arm.com>
31629 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
31630 to all inlined functions, change static to extern.
31632 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
31635 * config/arm/arm.c (legitimize_pic_address): Handle reference to
31637 (arm_assemble_integer): Likewise.
31639 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
31641 * config.gcc: Use new awk script to check CPU, FPU and architecture
31642 parameters for --with-... options.
31643 * config/arm/parsecpu.awk: New file
31644 * config/arm/arm-cpus.in: New file.
31645 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
31647 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
31649 * config/arm/t-arm: Update dependency rules.
31650 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
31651 of processing .def files.
31652 * config/arm/genopt.sh: Deleted.
31653 * config/arm/gentune.sh: Deleted.
31654 * config/arm/arm-cores.def: Deleted.
31655 * config/arm/arm-arches.def: Deleted.
31656 * config/arm/arm-fpus.def: Deleted.
31657 * config/arm/arm-tune.md: Regenerated.
31658 * config/arm/arm-tables.opt: Regenerated.
31659 * config/arm/arm-cpu.h: New generated file.
31660 * config/arm/arm-cpu-data.h: New generated file.
31661 * config/arm/arm-cpu-cdata.h: New generated file.
31663 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
31666 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
31668 (input_varpool_node): Unpack dynamically_initialized bit.
31670 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
31672 PR rtl-optimization/79032
31673 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
31674 the alignment of the adjusted memory reference against that of MODE,
31675 instead of the alignment of the original memory reference.
31677 2017-01-11 Martin Jambor <mjambor@suse.cz>
31679 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
31681 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
31682 decorated functions.
31684 2017-01-11 Richard Biener <rguenther@suse.de>
31686 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
31687 set range/nonnull info for PHI results. Do not set it on
31688 stmts marked for removal.
31690 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
31692 * expr.c (store_field): In the bitfield case, fetch the return value
31693 from the registers before applying a single big-endian adjustment.
31694 Always do a final load for a BLKmode value not larger than a word.
31696 2017-01-10 David Malcolm <dmalcolm@redhat.com>
31699 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31700 that we correctly handle column numbers greater than
31701 LINE_MAP_MAX_COLUMN_NUMBER.
31703 2017-01-10 Martin Sebor <msebor@redhat.com>
31705 PR middle-end/78245
31706 * gimple-ssa-sprintf.c (get_destination_size): Call
31707 {init,fini}object_sizes.
31708 * tree-object-size.c (addr_object_size): Adjust.
31709 (pass_through_call): Adjust.
31710 (pass_object_sizes::execute): Adjust.
31711 * tree-object-size.h (fini_object_sizes): Declare.
31713 2017-01-10 Martin Sebor <msebor@redhat.com>
31715 PR tree-optimization/78775
31716 * builtins.c (get_size_range): Move...
31717 * calls.c: ...to here.
31718 (alloc_max_size): Accept zero argument.
31719 (operand_signed_p): Remove.
31720 (maybe_warn_alloc_args_overflow): Call get_size_range.
31721 * calls.h (get_size_range): Declare.
31723 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
31725 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
31726 from TI's devices.csv file as of September 2016.
31727 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
31729 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
31731 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
31732 * doc/invoke.texi: Likewise.
31733 * doc/md.texi: Likewise.
31734 * doc/objc.texi: Likewise.
31736 2017-01-10 Joshua Conner <joshconner@google.com>
31738 * config/arm/fuchsia-elf.h: New file.
31739 * config/fuchsia.h: New file.
31740 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
31741 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
31743 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
31745 2016-01-10 Richard Biener <rguenther@suse.de>
31747 PR tree-optimization/79034
31748 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31749 Propagate out degenerate PHIs in the joiner.
31751 2017-01-10 Martin Liska <mliska@suse.cz>
31753 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
31754 (sort_congruence_classes_by_decl_uid): Likewise.
31755 (sort_congruence_class_groups_by_decl_uid): Likewise.
31756 (sem_item_optimizer::merge_classes): Sort class, groups in these
31757 classes and members in the groups by DECL_UID of declarations.
31758 This would make merge operations stable.
31760 2017-01-10 Martin Liska <mliska@suse.cz>
31762 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
31763 usage of m_classes_vec.
31764 (sem_item_optimizer::~sem_item_optimizer): Likewise.
31765 (sem_item_optimizer::get_group_by_hash): Likewise.
31766 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
31767 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
31768 (sem_item_optimizer::verify_classes): Likewise.
31769 (sem_item_optimizer::process_cong_reduction): Likewise.
31770 (sem_item_optimizer::dump_cong_classes): Likewise.
31771 (sem_item_optimizer::merge_classes): Likewise.
31772 * ipa-icf.h (congruence_class_hash): Rename from
31773 congruence_class_group_hash. Remove declaration of m_classes_vec.
31775 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
31777 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
31778 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
31779 * config.gcc: Add avx512vpopcntdqintrin.h.
31780 * config/i386/avx512vpopcntdqintrin.h: New.
31781 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
31782 * config/i386/i386-builtin-types.def: Add new types.
31783 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
31784 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
31785 __builtin_ia32_vpopcountq_v8di_mask): New.
31786 * config/i386/i386-c.c (ix86_target_macros_internal): Define
31787 __AVX512VPOPCNTDQ__.
31788 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
31789 (PTA_AVX512VPOPCNTDQ): Define.
31790 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
31791 TARGET_AVX512VPOPCNTDQ_P): Define.
31792 * config/i386/i386.opt: Add mavx512vpopcntdq.
31793 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
31794 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
31796 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31798 PR middle-end/77484
31799 * predict.def (PRED_CALL): Set to 67.
31801 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
31803 * expr.c (store_field): In the bitfield case, if the value comes from
31804 a function call and is of an aggregate type returned in registers, do
31805 not modify the field mode; extract the value in all cases if the mode
31806 is BLKmode and the size is not larger than a word.
31808 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
31811 * config/i386/cpuid.h: Fix undefined behavior.
31813 2017-01-04 Jeff Law <law@redhat.com>
31815 PR tree-optimization/79007
31816 PR tree-optimization/67955
31817 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
31818 conservative for pt.null when flag_non_call_exceptions is on.
31820 2017-01-09 Jakub Jelinek <jakub@redhat.com>
31822 PR translation/79019
31823 PR translation/79020
31824 * params.def (PARAM_INLINE_MIN_SPEEDUP,
31825 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
31826 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
31828 * config/avr/avr.opt (maccumulate-args): Likewise.
31829 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
31830 * common.opt (freport-bug): Likewise.
31831 * cif-code.def (CIF_FINAL_ERROR): Likewise.
31832 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
31833 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
31834 translatable string.
31835 * config/i386/i386.c (function_value_32): Likewise.
31836 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
31837 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
31839 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
31840 * common/config/msp430/msp430-common.c (msp430_handle_option):
31842 * symtab.c (symtab_node::verify_base): Likewise.
31843 * opts.c (set_debug_level): Likewise.
31844 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
31845 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
31846 missing whitespace to translatable strings.
31847 * config/avr/avr.md (bswapsi2): Fix typo in comment.
31848 * config/sh/superh.h: Likewise.
31849 * config/i386/xopintrin.h: Likewise.
31850 * config/i386/znver1.md: Likewise.
31851 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
31852 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
31853 * double-int.h (struct double_int): Likewise.
31854 * double-int.c (div_and_round_double): Likewise.
31855 * wide-int.cc: Likewise.
31856 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
31857 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
31858 * cfgcleanup.c (crossjumps_occured): Renamed to ...
31859 (crossjumps_occurred): ... this.
31860 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
31863 PR tree-optimization/78899
31864 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
31865 returning bool return struct loop *, NULL for failure and the new
31867 (versionable_outer_loop_p): Don't version outer loop if it has
31868 dont_vectorized bit set.
31869 (tree_if_conversion): When versioning outer loop, ensure
31870 tree_if_conversion is performed also on the inner loop of the
31871 non-vectorizable outer loop copy.
31872 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
31873 LOOP_VECTORIZED in inner loop of the scalar outer loop and
31874 prevent vectorization of it.
31875 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
31876 the outer loop vectorization of the non-scalar version is attempted
31877 before vectorization of the inner loop in scalar version. If
31878 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
31879 vectorization of its inner loop.
31880 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
31881 has 2 inner loops, rename also on edges from bb whose single pred
31882 is outer_loop->header. Fix typo in function comment.
31884 2017-01-09 Martin Sebor <msebor@redhat.com>
31887 * asan.c (asan_emit_stack_protection): Increase local buffer size
31888 to avoid snprintf truncation warning.
31890 2017-01-09 Andrew Pinski <apinski@cavium.com>
31892 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
31893 to reference thunderx2t99 for the tuning structure
31894 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
31896 (thunderx2t99_extra_costs): This.
31897 * config/aarch64/aarch64-tune.md: Regenerate.
31898 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
31899 (vulcan_addrcost_table): This.
31900 (vulcan_regmove_cost): Rename to ...
31901 (thunderx2t99_regmove_cost): This.
31902 (vulcan_vector_cost): Rename to ...
31903 (thunderx2t99_vector_cost): this.
31904 (vulcan_branch_cost): Rename to ...
31905 (thunderx2t99_branch_cost): This.
31906 (vulcan_tunings): Rename to ...
31907 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
31908 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
31910 2017-01-09 Martin Jambor <mjambor@suse.cz>
31914 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
31915 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
31916 (propagate_vr_accross_jump_function): Use the above function for all
31917 value range computations for pass-through jump functions and type
31918 converasion from explicit value range values.
31919 (ipcp_propagate_stage): Do not attempt to deduce types of formal
31920 parameters from TYPE_ARG_TYPES.
31921 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
31922 (ipa_write_node_info): Stream type of the actual argument.
31923 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
31925 2017-01-09 Martin Liska <mliska@suse.cz>
31928 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
31929 (lookup_compiler): Do not show error message with have_E.
31931 2017-01-09 Jakub Jelinek <jakub@redhat.com>
31933 PR tree-optimization/78938
31934 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
31935 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
31936 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
31937 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
31940 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31942 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
31945 2017-01-09 Richard Biener <rguenther@suse.de>
31947 PR tree-optimization/78997
31948 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
31949 name condition properly.
31951 2017-01-09 Richard Biener <rguenther@suse.de>
31954 * dwarf2out.c (is_cxx): New overload with context.
31955 (is_naming_typedef_decl): Use it.
31957 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31959 * invoke.texi (Option Summary): Correct spacing in option lists
31960 and add line breaks to fix over-long lines.
31962 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31964 PR middle-end/17660
31966 * extend.texi (Common Variable Attributes): Add xref to GCC
31967 Internals manual to explain mode attribute keywords.
31969 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31972 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
31973 and Preprocessor Options.
31974 (Options for Linking): Document -pthread here....
31975 (RS/6000 and PowerPC Options): ...not here.
31976 (Solaris 2 Options): ...or here.
31977 * doc/cppopts.texi: Document -pthread.
31979 2017-01-08 Martin Sebor <msebor@redhat.com>
31981 PR middle-end/77708
31982 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
31983 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
31984 New member functions.
31985 (format_directive): Used them.
31987 (pass_sprintf_length::handle_gimple_call): Same.
31988 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
31989 to avoid truncation for any argument.
31990 (extract_affine_mul): Same.
31991 * tree.c (get_file_function_name): Same.
31993 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31995 PR middle-end/77484
31996 * predict.def (PRED_INDIR_CALL): Set to 86.
31998 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32000 PR preprocessor/54124
32001 * doc/cppopts.texi: Reformat -d subtable to list the full name
32002 of the options. Add cross-reference to the docs for the general
32003 compiler -d options.
32004 * doc/invoke.texi (Developer Options): Add cross-reference to the
32005 preprocessor-specific -d option documentation.
32007 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32009 PR preprocessor/13498
32010 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
32011 redudant material, and reflect new command-line options.
32012 (System Headers): Likewise.
32014 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32016 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
32017 -isystem, and -idirafter. Copy-edit.
32018 * doc/cppopts.texi: Copy-edit. Remove contradiction about
32019 default for -ftrack-macro-expansion. Delete obsolete and
32020 badly-formatted implementation details about -fdebug-cpp output.
32021 * doc/cppwarnopts.texi: Copy-edit.
32023 2017-01-07 David Malcolm <dmalcolm@redhat.com>
32026 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32027 that the transition from a max line width >= 1<<10 to narrower
32028 lines works correctly.
32030 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
32032 * doc/options.texi (PerFunction): New.
32033 * opt-functions.awk (switch_flags): Map both Optimization and
32034 PerFunction to CL_OPTIMIZATION.
32035 * opth-gen.awk: Test for PerFunction flag along with
32037 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
32038 it only when the latter is present. Skip those that don't in
32039 the hash function generator.
32040 * common.opt (fvar-tracking): Mark as PerFunction instead of
32042 (fvar-tracking-assignments): Likewise.
32043 (fvar-tracking-assignments-toggle): Likewise.
32044 (fvar-tracking-uninit): Likewise.
32046 2017-01-07 Jakub Jelinek <jakub@redhat.com>
32048 PR translation/79018
32049 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
32052 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
32055 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
32056 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
32057 TARGET_LONG_JUMP_TABLE_OFFSETS.
32058 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
32059 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
32060 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
32061 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
32062 * config/m68k/m68k.md (tablejump expander): Likewise.
32063 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
32064 TARGET_LONG_JUMP_TABLE_OFFSETS.
32065 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
32066 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
32068 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
32069 David Holsgrove <david.holsgrove@xilinx.com>
32071 * common/config/microblaze/microblaze-common.c
32072 (TARGET_EXCEPT_UNWIND_INFO): Remove.
32073 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
32075 * config/microblaze/microblaze.c (microblaze_must_save_register)
32076 (microblaze_expand_epilogue, microblaze_return_addr): Handle
32078 (microblaze_eh_return): New function.
32079 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
32080 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
32081 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
32082 * config/microblaze/microblaze.md (eh_return): New pattern.
32084 2017-01-06 Jakub Jelinek <jakub@redhat.com>
32086 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
32087 GCC_DIAGNOSTIC_STRINGIFY): Define.
32089 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
32091 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32093 * config/arm/arm.md (<mcrr>): New.
32095 * config/arm/arm.c (arm_arch5te): New.
32096 (arm_option_override): Set arm_arch5te.
32097 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
32099 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
32100 (arm_mcrr_qualifiers): ... this. New.
32101 (MRRC_QUALIFIERS): Define to...
32102 (arm_mrrc_qualifiers): ... this. New.
32103 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
32105 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
32106 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
32107 (MRRCI, mrrc, MRRC): New.
32108 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
32109 VUNSPEC_MRRC2): New.
32111 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32113 * config/arm/arm.md (<mcr>): New.
32115 * config/arm/arm.c (arm_coproc_builtin_available): Add
32116 support for mcr, mrc, mcr2 and mrc2.
32117 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
32118 (arm_mcr_qualifiers): ... this. New.
32119 (MRC_QUALIFIERS): Define to ...
32120 (arm_mrc_qualifiers): ... this. New.
32121 (MCR_QUALIFIERS): Define to ...
32122 (arm_mcr_qualifiers): ... this. New.
32123 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
32125 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
32126 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
32127 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
32128 VUNSPEC_MRC2): New.
32130 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32132 * config/arm/arm.md (*ldc): New.
32136 * config/arm/arm.c (arm_coproc_builtin_available): Add
32137 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
32138 (arm_coproc_ldc_stc_legitimate_address): New.
32139 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
32140 'qualifier_const_pointer'.
32141 (LDC_QUALIFIERS): Define to...
32142 (arm_ldc_qualifiers): ... this. New.
32143 (STC_QUALIFIERS): Define to...
32144 (arm_stc_qualifiers): ... this. New.
32145 * config/arm/arm-protos.h
32146 (arm_coproc_ldc_stc_legitimate_address): New.
32147 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
32148 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
32149 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
32150 stc2, stcl, stc2l): New.
32151 * config/arm/constraints.md (Uz): New.
32152 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
32153 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
32154 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
32155 VUNSPEC_STC2L): New.
32157 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32159 * config/arm/arm.md (<cdp>): New.
32160 * config/arm/arm.c (neon_const_bounds): Rename this ...
32161 (arm_const_bounds): ... this.
32162 (arm_coproc_builtin_available): New.
32163 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
32164 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
32165 (CDP_QUALIFIERS): Define to...
32166 (arm_cdp_qualifiers): ... this. New.
32168 (arm_expand_builtin_args): Add case for 6 arguments.
32169 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
32170 (arm_const_bounds): ... this.
32171 (arm_coproc_builtin_available): New.
32172 * config/arm/arm_acle.h (__arm_cdp): New.
32174 * config/arm/arm_acle_builtins.def (cdp): New.
32176 * config/arm/iterators.md (CDPI,CDP,cdp): New.
32177 * config/arm/neon.md: Rename all 'neon_const_bounds' to
32178 'arm_const_bounds'.
32179 * config/arm/types.md (coproc): New.
32180 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
32181 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
32182 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
32183 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
32185 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32187 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
32188 (UBINOP_QUALIFIERS): New.
32190 (acle_builtin_data): New. Change comment.
32191 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
32192 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
32193 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
32194 arm_acle_builtins.def.
32195 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
32196 (arm_init_acle_builtins): New.
32197 (CRC32_BUILTIN): Remove.
32198 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
32199 crc32cb, crc32ch and crc32cw.
32200 (arm_init_crc32_builtins): Remove.
32201 (arm_init_builtins): Use arm_init_acle_builtins rather
32202 than arm_init_crc32_builtins.
32203 (arm_expand_acle_builtin): New.
32204 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
32205 * config/arm/arm_acle_builtins.def: New.
32207 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32209 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
32210 (arm_builtin_datum): ... this.
32211 (arm_init_neon_builtin): Rename to ...
32212 (arm_init_builtin): ... this. Add a new parameters PREFIX
32213 and USE_SIG_IN_NAME.
32214 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
32215 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
32216 'arm_builtin_datum'.
32217 (arm_init_vfp_builtins): Likewise.
32218 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
32219 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
32220 (arm_expand_neon_args): Rename to ...
32221 (arm_expand_builtin_args): ... this. Rename builtin_arg
32222 enum values and differentiate between ARG_BUILTIN_MEMORY
32223 and ARG_BUILTIN_NEON_MEMORY.
32224 (arm_expand_neon_builtin_1): Rename to ...
32225 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
32226 values, arm_expand_builtin_args and add bool parameter NEON.
32227 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
32228 (arm_expand_vfp_builtin): Likewise.
32229 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
32231 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32233 PR middle-end/77484
32234 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
32235 * predict.c (tree_estimate_probability_bb): Reverse direction of
32236 polymorphic call predictor.
32238 2017-01-06 David Malcolm <dmalcolm@redhat.com>
32240 * passes.c (execute_one_pass): Split out pass-skipping logic into...
32241 (determine_pass_name_match): ...this new function and...
32242 (should_skip_pass_p): ...this new function.
32244 2017-01-06 Nathan Sidwell <nathan@acm.org>
32246 * ipa-visibility.c (function_and_variable_visibility): Reformat
32247 comments and long lines. Remove extrneous if.
32248 * symtab.c (symtab_node::make_decl_local): Fix code format.
32249 (symtab_node::set_section_for_node): Fix comment typo.
32251 2017-01-06 Martin Liska <mliska@suse.cz>
32254 * lra-constraints.c: Rename invariant to lra_invariant.
32255 * predict.c (set_even_probabilities): Initialize e to NULL.
32257 2017-01-05 Martin Sebor <msebor@redhat.com>
32259 PR tree-optimization/78910
32260 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
32261 (format_integer): Correct off-by-one error in the handling
32262 of precision with negative numbers in signed conversions..
32264 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
32266 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
32268 2017-01-05 Jakub Jelinek <jakub@redhat.com>
32270 PR tree-optimization/71016
32271 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
32272 factor_out_conditional_conversion. Formatting fix.
32273 (factor_out_conditional_conversion): Add cond_stmt argument.
32274 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
32275 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
32278 2017-01-05 David Malcolm <dmalcolm@redhat.com>
32280 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
32281 read-rtl-function.o, and selftest-rtl.o.
32282 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
32283 (selftest::aarch64_test_loading_full_dump): New function.
32284 (selftest::aarch64_run_selftests): New function.
32285 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
32286 selftest::aarch64_run_selftests.
32287 * config/i386/i386.c
32288 (selftest::ix86_test_loading_dump_fragment_1): New function.
32289 (selftest::ix86_test_loading_call_insn): New function.
32290 (selftest::ix86_test_loading_full_dump): New function.
32291 (selftest::ix86_test_loading_unspec): New function.
32292 (selftest::ix86_run_selftests): Call the new functions.
32293 * emit-rtl.c (maybe_set_max_label_num): New function.
32294 * emit-rtl.h (maybe_set_max_label_num): New decl.
32295 * function.c (instantiate_decls): Guard call to
32296 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
32297 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
32299 * gensupport.c (gen_reader::gen_reader): Pass "false"
32300 for new "compact" param of rtx_reader.
32301 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
32302 rather than an empty string for NULL strings.
32303 * read-md.c: Potentially include config.h rather than bconfig.h.
32304 Wrap include of errors.h with #ifdef GENERATOR_FILE.
32305 (have_error): New global, copied from errors.c.
32306 (md_reader::read_name): Rename to...
32307 (md_reader::read_name_1): ...this, adding "out_loc" param,
32308 and converting "missing name or number" to returning false, rather
32310 (md_reader::read_name): Reimplement in terms of read_name_1.
32311 (md_reader::read_name_or_nil): New function.
32312 (md_reader::read_string): Handle "(nil)" by returning NULL.
32313 (md_reader::md_reader): Add new param "compact".
32314 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
32315 (md_reader::read_file): New method.
32316 * read-md.h (md_reader::md_reader): Add new param "compact".
32317 (md_reader::read_file): New method.
32318 (md_reader::is_compact): New accessor.
32319 (md_reader::read_name): Convert return type from void to file_location.
32320 (md_reader::read_name_or_nil): New decl.
32321 (md_reader::read_name_1): New decl.
32322 (md_reader::m_compact): New field.
32323 (noop_reader::noop_reader): Pass "false" for new "compact" param
32325 (rtx_reader::rtx_reader): Add new "compact" param.
32326 (rtx_reader::read_rtx_operand): Make virtual and convert return
32327 type from void to rtx.
32328 (rtx_reader::read_until): New decl.
32329 (rtx_reader::handle_any_trailing_information): New virtual function.
32330 (rtx_reader::postprocess): New virtual function.
32331 (rtx_reader::finalize_string): New virtual function.
32332 (rtx_reader::m_in_call_function_usage): New field.
32333 (rtx_reader::m_reuse_rtx_by_id): New field.
32334 * read-rtl-function.c: New file.
32335 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
32336 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
32337 (selftest::verify_three_block_rtl_cfg): New decl.
32338 * read-rtl-function.h: New file.
32339 * read-rtl.c: Potentially include config.h rather than bconfig.h.
32340 For host, include function.h, memmodel.h, and emit-rtl.h.
32341 (one_time_initialization): New function.
32342 (struct compact_insn_name): New struct.
32343 (compact_insn_names): New array.
32344 (find_code): Handle insn codes in compact dumps.
32345 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
32346 (bind_subst_iter_and_attr): Likewise.
32347 (add_condition_to_string): Likewise.
32348 (add_condition_to_rtx): Likewise.
32349 (apply_attribute_uses): Likewise.
32350 (add_current_iterators): Likewise.
32351 (apply_iterators): Likewise.
32352 (initialize_iterators): Guard usage of apply_subst_iterator with
32353 #ifdef GENERATOR_FILE.
32354 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
32355 (md_reader::read_mapping): Likewise.
32356 (add_define_attr_for_define_subst): Likewise.
32357 (add_define_subst_attr): Likewise.
32358 (read_subst_mapping): Likewise.
32359 (check_code_iterator): Likewise.
32360 (rtx_reader::read_rtx): Likewise. Move one-time initialization
32362 (one_time_initialization): New function.
32363 (rtx_reader::read_until): New method.
32364 (read_flags): New function.
32365 (parse_reg_note_name): New function.
32366 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
32367 Handle reuse_rtx ids.
32368 Wrap iterator lookup within #ifdef GENERATOR_FILE.
32369 Add parsing support for RTL dumps, mirroring the special-cases in
32370 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
32371 values, and calling handle_any_trailing_information.
32372 (rtx_reader::read_rtx_operand): Convert return type from void
32373 to rtx, returning return_rtx. Handle case 'e'. Call
32374 finalize_string on XSTR and XTMPL fields.
32375 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
32376 "(nil)" values were omitted. Call the postprocess vfunc on the
32378 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
32379 class ctor. Initialize m_in_call_function_usage. Call
32380 one_time_initialization.
32381 * rtl-tests.c (selftest::test_uncond_jump): Call
32382 set_new_first_and_last_insn.
32383 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
32384 * selftest-rtl.c: New file.
32385 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
32386 (selftest::get_insn_by_uid): New decl.
32387 * selftest-run-tests.c (selftest::run_tests): Call
32388 read_rtl_function_c_tests.
32389 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
32390 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
32393 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
32395 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
32396 operands in a special way. Assert that pos+len <= mode precision.
32398 2017-01-05 Jakub Jelinek <jakub@redhat.com>
32400 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
32401 3 argument Alias with unlimited for the negative form.
32402 (fno-vect-cost-model): Removed.
32404 2017-01-05 Martin Liska <mliska@suse.cz>
32406 * hsa-gen.c (gen_hsa_divmod): New function.
32407 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
32409 2017-01-05 Martin Liska <mliska@suse.cz>
32412 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
32415 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32417 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
32418 small constant length operands.
32420 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32422 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
32423 between loop iterations.
32425 2017-01-05 Martin Liska <mliska@suse.cz>
32428 * gimplify.c (gimplify_decl_expr): Compare to
32429 asan_poisoned_variables instread of checking flags.
32430 (gimplify_target_expr): Likewise.
32431 (gimplify_expr): Likewise.
32432 (gimplify_function_tree): Conditionally initialize
32433 asan_poisoned_variables.
32435 2017-01-04 Jeff Law <law@redhat.com>
32437 PR tree-optimizatin/78812
32438 * rtl.h (contains_mem_rtx_p): Prototype.
32439 * ifcvt.c (containts_mem_rtx_p): Move from here to...
32440 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
32441 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
32442 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
32443 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
32445 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32447 * input.c (assert_char_at_range): Default-initialize actual_range.
32449 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32451 * df-scan.c (df_ref_create_structure): Make regno unsigned,
32452 to match the caller.
32454 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32456 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
32457 insns after final jump in test to emit dummy move.
32459 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32461 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
32462 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
32464 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32466 * multiple_target.c (create_dispatcher_calls): Init e_next.
32467 * tree-ssa-loop-split.c (split_loop): Init border.
32468 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
32471 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
32476 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
32477 (altivec_register_operand): Do not return true if the operand
32478 contains a SUBREG mixing SImode and SFmode.
32479 (vsx_register_operand): Likewise.
32480 (vsx_reg_sfsubreg_ok): New predicate.
32481 (vfloat_operand): Do not return true if the operand contains a
32482 SUBREG mixing SImode and SFmode.
32483 (vint_operand): Likewise.
32484 (vlogical_operand): Likewise.
32485 (gpc_reg_operand): Likewise.
32486 (int_reg_operand): Likewise.
32487 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
32488 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
32489 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
32491 (rs6000_emit_move_si_sf_subreg): New helper function.
32492 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
32493 fixup SUBREGs involving SImode and SFmode.
32494 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
32495 numbers for the new peephole2 optimization.
32496 (peephole2 for SFmode unions): New peephole2 to optimize cases in
32497 the GLIBC math library that do AND/IOR/XOR operations on single
32498 precision floating point.
32499 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
32500 target macros to say whether we need to avoid SUBREGs mixing
32502 (TARGET_ALLOW_SF_SUBREG): Likewise.
32503 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
32504 (UNSPEC_SI_FROM_SF): Likewise.
32505 (iorxor): Change spacing.
32506 (and_ior_xor): New iterator for AND, IOR, and XOR.
32507 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
32508 (movdi_from_sf_zero_ext): Likewise.
32509 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
32510 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
32511 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
32512 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
32513 (fms<mode>4): Likewise.
32514 (fnma<mode>4): Likewise.
32515 (fnms<mode>4): Likewise.
32516 (nfma<mode>4): Likewise.
32517 (nfms<mode>4): Likewise.
32519 2017-01-04 Marek Polacek <polacek@redhat.com>
32522 * doc/invoke.texi: Document -Wpointer-compare.
32524 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32526 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
32529 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
32530 descriptions for -gdwarf-5 and emit them as uleb128 instead of
32533 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
32536 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
32537 documentation of the powerpc_popcntb_ok attribute.
32538 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32539 code to issue warning messages if a requested CPU configuration is
32540 not supported by the binary (assembler and loader) toolchain.
32541 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
32542 made to define a built-in function that has been disabled.
32543 (paired_init_builtins): Add assertion to prevent ICE if attempt is
32544 made to define a built-in function that has been disabled.
32545 (altivec_init_builtins): Add comment explaining why definition
32546 of the DST built-in functions is not preceded by an assertion
32547 check. Add assertions to prevent ICE if attempts are made to
32548 define an altivec predicate or an abs* built-in function that has
32550 (htm_init_builtins): Add comment explaining why definition of the
32551 htm built-in functions is not preceded by an assertion check.
32553 2017-01-04 Jeff Law <law@redhat.com>
32555 PR tree-optimizatin/67955
32556 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
32557 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
32558 the points-to solution does not include pt_null. Use DECL_PT_UID
32561 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
32563 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
32564 Use gen_int_mode instead of gen_lopwart for const_int operands.
32566 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32568 PR tree-optimization/71563
32569 * match.pd: Simplify X << Y into X if Y is known to be 0 or
32570 out of range value - has low bits known to be zero.
32572 2017-01-04 Alan Modra <amodra@gmail.com>
32574 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
32575 * configure: Regenerate.
32576 * config.in: Regenerate.
32578 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32581 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
32582 a substring of the message, but strcmp with the whole message. Ifdef
32583 ENABLE_NLS, translate the message first using dgettext.
32585 2017-01-03 Jeff Law <law@redhat.com>
32587 PR tree-optimizatin/78856
32588 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
32589 (mark_threaded_blocks): Remove code to truncate thread paths that
32590 cross multiple loop headers. Instead invalidate the cached loop
32591 iteration information and handle case of a thread path walking
32592 into an irreducible region.
32594 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
32597 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
32598 assertions. Add support for doing the signbit if the IEEE 128-bit
32599 floating point value is in a GPR.
32600 * config/rs6000/rs6000.md (Fsignbit): Delete.
32601 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
32602 Update the length attribute if the value is in a GPR.
32603 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
32604 the sign or zero extension instruction, since the value is always 0/1.
32605 (signbit<mode>2_dm2): Delete using <Fsignbit>.
32608 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
32609 extracting SImode to a GPR register so that we can generate a
32610 store, limit the vector to be in a traditional Altivec register
32611 for the vextuwrx instruction.
32613 2017-01-03 Ian Lance Taylor <iant@google.com>
32615 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
32617 2017-01-03 Martin Sebor <msebor@redhat.com>
32619 PR tree-optimization/78696
32620 * gimple-ssa-sprintf.c (format_floating): Correct handling of
32621 precision. Use MPFR for %f for greater fidelity. Correct handling
32623 (pass_sprintf_length::compute_format_length): Set width and precision
32624 specified by asrerisk to void_node for vararg functions.
32625 (try_substitute_return_value): Adjust dump output.
32627 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
32629 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
32631 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
32633 * doc/invoke.texi (SPARC options): Document -mlra as the default.
32634 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
32635 -mlra/-mno-lra was passed to the compiler.
32637 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
32639 PR rtl-optimization/65618
32640 * emit-rtl.c (try_split): Move initialization of "before" and
32641 "after" to just before the call to emit_insn_after_setloc.
32643 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
32645 * doc/md.texi (Standard Names): Remove reference to Java frontend.
32647 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
32649 * dwarf2out.c (gen_enumeration_type_die): When
32650 -gno-strict-dwarf, add a DW_AT_encoding attribute.
32652 2017-01-03 Jakub Jelinek <jakub@redhat.com>
32654 PR tree-optimization/78965
32655 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
32656 Change first argument from const call_info & to call_info &. For %n
32657 set info.nowrite to false.
32659 PR middle-end/78901
32660 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
32661 possibly throwing calls.
32663 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
32664 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
32665 and fns handling, rather than in a separate case SSA_NAME.
32667 2017-01-02 Jeff Law <law@redhat.com>
32669 * config/darwin-driver.c (darwin_driver_init): Const-correctness
32670 fixes for first_period and second_period variables.
32672 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
32675 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
32676 (*insvqi_1): New insn pattern.
32677 (*insvqi_1_mem_rex64): Ditto.
32678 (*insvqi_2): Ditto.
32679 (*insvqi_3): Rename from *insvqi.
32681 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
32683 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
32685 * doc/cfg.texi (Edges): Remove reference to Java.
32686 (Maintaining the CFG): Ditto.
32688 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32690 PR middle-end/77674
32691 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
32692 transparent aliases.
32694 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32696 PR middle-end/77484
32697 * predict.def (PRED_CALL): Update hitrate.
32698 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
32699 * predict.c (tree_estimate_probability_bb): Split CALL predictor
32700 into direct/indirect/polymorphic variants.
32702 2017-01-01 Jakub Jelinek <jakub@redhat.com>
32704 Update copyright years.
32706 * gcc.c (process_command): Update copyright notice dates.
32707 * gcov-dump.c (print_version): Ditto.
32708 * gcov.c (print_version): Ditto.
32709 * gcov-tool.c (print_version): Ditto.
32710 * gengtype.c (create_file): Ditto.
32711 * doc/cpp.texi: Bump @copying's copyright year.
32712 * doc/cppinternals.texi: Ditto.
32713 * doc/gcc.texi: Ditto.
32714 * doc/gccint.texi: Ditto.
32715 * doc/gcov.texi: Ditto.
32716 * doc/install.texi: Ditto.
32717 * doc/invoke.texi: Ditto.
32719 Copyright (C) 2017 Free Software Foundation, Inc.
32721 Copying and distribution of this file, with or without modification,
32722 are permitted in any medium without royalty provided the copyright
32723 notice and this notice are preserved.